DE2747702A1 - Automatic telephone number information service - reduces possible errors by using only initial letters of main words in name, address and occupation - Google Patents

Abstract

The automatic telephone information service installation has a main index contg. name, address and occupation information for all subscribers. The information is stored in main code or as a concept term, ie. only the first few letters of each main word are stored. The main indices are store e.g. on tape and the indices at the end of each track also stored in an access memory, e.g. a core matrix, as key words. Enquiry information from a subscriber seeking a telephone number is received at a control desk and entered into the system in abbreviated form as an enquiry index to which an enquiry number, control desk number and a format control symbol are added. The enquiry index is compared with the key words.

Description

Automatic telephone notification ~~

The invention relates to an automatic telephone directory assistance system at least one input device that is operated by an operator to a Query information including a format control pulse for a query index enter, with at least one output device, that the information in response which delivers queries, with at least one data store with direct access to it the memory data inputs and with at least one central computer and control unit, especially for comparison operations.

One known automatic directory assistance system is in the United States patent 3 242 47o. In this known system, the entire content of a Telephone directory entered in digitally coded form in a computer memory.

The contents of the telephone directory are useful for retrieving this information or the phone book summarized in rough indices, each of which is again in 200 fine indexes is divided. This coarse index is used for the first search operation used for the storage area to be searched on i / 2oo of the entire Narrow down the contents of the telephone directory. Each of the 200 entry groups is des further divided into 200 finer parts, indicated by fine indices as another Search levels are shown. Finally, each of the 200 finer parts is in So Feinstgruppen subdivided by fine indices. This continued subdivision of the indices the search is continued until it is determined under which address of the memory the desired entry can be found. This rough and fine search becomes complete conducted on an alphabetical basis. In the event that the information system forwarded information request is incomplete, which often occurs, provides the known system does not provide the correct telephone subscriber information, i.e. the exact phone number.

If the requester has found the correct and complete spelling of the Knows the name of the subscriber and if the alphabetical coarse and fine search leads to the memory address for the desired entry, there is the next one step in the well-known information system in it, the entire block of telephone information entries, to translate those found under this memory address into a visible form, and the entire block of information on a viewing screen for review to be represented by the information operator. This procedure requires complicated and expensive electronic equipments, and the operator must also read the information block shown.

The disadvantage here is that only 1/8 of an information block, which is contained in a memory address, at the same time on the display or screen is shown so that the font can remain large enough to be easily read to be able to. The operator must then manually open the image window Move over the insignificant parts of the block information until the essential Part is localized before moving on to picking out the specific one you want Information can begin.

A difficulty with known automatic telephone number directory assistance services results from the fact that the inquirer generally has only imprecise knowledge from the subscriber whose phone number he is looking for.

This difficulty is compounded by the need for a quick one Access to the information based on an imprecise query is made difficult.

In practice it has been found that the coincidence ratio between rough information in an inquiry and in a telephone directory on the basis approximate the correspondence of each character in the information for the name 3o%, for the address approx. 13% and for the serum approx. 4%.

Furthermore, it was found that the simultaneous coincidence for the name, the address and for the profession make up a maximum of about 13%.

It is also true that with an increasing number of keyword elements the accuracy of the answer drops if the input information is inaccurate.

In the known system according to US Pat. No. 3,242,470 only one name is used for the search index. However, in this system, the number becomes the responses are larger, inversely proportional to the degree to which the index is simplified so that the operator is faced with a large block of data will. This requires increased operator intervention.

In some known systems, a so-called 'random method' used for access using the spoken word. This 'random method' divides the word pattern into random digits for the memory address by a given one mathematical operation on. Because the letters in a human name are statistical have seen unequal distribution, the result of the random method is not satisfactory to get the true randomization of the letters. this leads to an increase in the overlapping of the entries at common positions. Therefore, in the known systems, the alphabetical order is in the regular Order of entries not suitable for direct search access.

The object of the invention is to provide an automatic directory assistance system to create a conversion of digitally encoded data into visible data not applicable, correct information is received even if the inquiries are incomplete Information about the name, address and employment of the telephone subscriber and in which the three pieces of information, which are independent of one another, These are the name, the address and the profession, so correlated with each other, that the number of possible answers that the operator has to judge, is decreased.

As part of this task is also to solve a direct problem Access to the data entry in alphabetical order using a Allow language index without any positional gaps or overlaps of entries occur and, moreover, to create a system that is faster works, requires less operator effort and both operator errors as well as inaccuracies or incompleteness of the request information.

According to the invention, this object is achieved by the characterizing features of claim 1 solved.

The further embodiment of the invention results from the characterizing Features of the following claims.

With the invention, the advantage is achieved that correct information on a question is received, even if that is an inaccuracy or incompleteness by retrieving information on the basis of a direct access concept instead of a sequential alphabetical concept. This is done at the beginning of the subscriber index from the independent input data, that is Name, place of residence, address and occupation of each subscriber, prepared. Some these indexed input elements are first converted into main codes or concepts and their surplus characters can be cut off.

Some of these inputs are on the border of two adjacent tracks or unit areas stored in a data memory are called a keyword used. The keywords are compared with the query index obtained from the Inquiry information is made to the address of the stored in the data store To locate information.

In the case of data entry, the name of a telephone subscriber is used as handles a combination of words N1, N2, N3, etc., of which N1 is the first or represents the most important word, for example in the case of a company the plant name reproduces, or in the case of a natural person, the surname of that person, while N2 is the second or the next most important word in the name of a company or for a natural person, for example, is the first name.

If the company name 'Alexander Chemical Industry Inc.' considered, N1 is equal to 'Alexander', N2 is equal to 'Chemical', while at the name of a natural person ~ George Washington ', N1 is equal to' Washlngton 'and N2 is equal to ~ George '.

The data A about the address of the telephone subscriber represent a Combination of A1, A2 etc., in which A1 is the district name, A2 the sub-district, A3 represents the address number etc. in cities such as Tokyo or London. In cities For example, in the United States, A1 may be the postal zone or a district of a city, A2 a street in this city, etc. The The occupation of the telephone subscriber is made up of a main term or concept and one specific term or a corresponding concept O2.

In practice it has been found that only the first four or five letters of N1 and N2 are required to find, although every entry in the index always has the same fixed word length. The usage a shortened name word of this kind in no way results in anything essential Loss of information as it turned out to be the first four or five letters any name are sufficient to establish a reasonable coincidence according to the invention, when the name is used in conjunction with two other independent terms or concepts how the occupation and the home address are correlated. On the other hand, it should be noted that the greater the possibility of human error, the greater it is Is the number of letters of a name that the inquirer should enter, that is all the more The probability that he will make a mistake is also greater, and the less so is therefore the overall effectiveness of the system.

As a result of the general classification described above or shortening the input data used has been found to be effective each element of the query to the stored element is: With respect to the name coincidence is 3o%, the coincidence for N being 95% and for N2 being 85%.

Compared to the address, the coincidence is 13%, with the coincidence for A1 is 95% of the coincidence of N1, while the coincidence of A2 is 68% amounts to.

Compared to a specific occupational coincidence of 4%, the coincidence from 0 92% after conversion to a general word code.

Regarding the negligence of 5% error in determination of A1 the proportion of incorrect information is 2%, while ignorance of one Inquirers on the basis of statistical studies by the telephone number information office in Tokyo accounts for 2.9%. The insignificance of 8% at O gives an error of 2.7%, while the ignorance by an inquirer around 5.4%, based on a general classification.

According to the invention, a search index I is used as a combination of m N1, A1 and 0, which constitute main codes or a concept term, are prepared. Some of the Indexes Imx stored in the data store are used as keywords K is picked out and stored in an access memory such as a Core matrix. Compare between the query index 1 q and the key words K. take place in a comparison device. Additional circuits are provided to the Data storage using the keywords K, the are stored in the access memory to be scanned with a division point search system. The division point search system comprises successive divisions by two of the Keyword data as a function of the result of the comparison between the query indices 1q with the keyword indexes K.

The invention also increases the number of responses that the operator receives can be shown for further decision by sieving with a reduction index the second most important data in the request information, the are N2 and A2. As a result of the reduction process, the resulting Nearly the number of responses presented to the operator in practice in all cases equal to one.

There is generally only a single correct answer from the operator is presented instead of an entire block of data, as in the prior art, the operating time is shortened and the system is made much more effective. Furthermore, the advantage is achieved that the information is printed and not on a screen presented that the need for the use of complicated and there is no need for expensive converting devices from digital to video signals.

There is also an error correction system for the automatic Correction of some misspelled or incorrect request data is planned. The error correction system includes an error correction memory table which contains various Contains names, addresses, etc. that are very often confused with other such data will. For example, it can happen that the postal zone of a store, which is very close to the borderline between two zones, often with the wrong one postal zone is confused.

If a query does not return a result, the memory table is scanned for information that is similar to the query information. Similar dates that stored in the table will then be matched with the appropriate Inquiry data exchanged and the data memory queried again.

This procedure continues until either a result of the Data storage is obtained or any similar information in the error correction storage table were used.

The invention is illustrated in the following with reference to FIG Embodiments explained in more detail. They show: FIG. 1 - a block diagram an automatic directory assistance system according to the invention; Figures 1A and 1B - Block diagrams of an additional device for input or output in with reference to the system of Figure 1; Figure 2A - a circuit diagram of a signal block and the relationship of the signal to the strobe pulses and to the channels; Figure 2B - a diagram of the digits D5 and D; Figure 2C - a diagram of the words Ws, We and W1, W2 Wn; Figure 2D is a diagram of a group of query data; Figures 3A and 3B - a block diagram of a central computing and control unit of the directory assistance system according to Figure 1, wherein Figure 3A a comparison device, a data memory and Contains additional equipment, while Figure 3B equipment for forming or manufacturing the index from the query information, a checking device, an index reforming device and includes ancillary equipment; Figure 4A is a block diagram of an input divider according to Figure 3B; Figure 4B - a timing diagram for opening and closing the input divider according to Figure 4A; FIG. 5 is a block diagram of an embodiment of a converter device in the central computing and control unit according to FIG. 3B; figure 6 is a logic block diagram detailing the converter components shown in FIG FIG. 5 for carrying out the conversion in the central control and processing unit according to Figure 3B; FIG. 7 - a block diagram of an identifier circuit for an initial letter the central computing and control unit according to FIG. 3B; Figure 8A - a logic block diagram of the connections between the index formers 306, 308, 3o9 and 310i FIG. 8B - a logical block diagram of the arrangement of the index formation directions, such as the device 31o in Figure 3B, and the design of the converter register in the central computing and control unit 273 according to FIG. 3B; Figure 9 - a block diagram of a detail of the N1 test circuit, namely a device 311 according to FIG. 3B, FIGS. 10A and 10B - a block diagram of a reduction comparator 314 in FIG the central computing and control unit according to FIG. 3B; Figure 11 - a diagram of a Response counter in the central computing and control unit according to Figure 3B; Figure 12 - a logic block diagram of the error correction circuits; and Figure 13 - a logical one Block diagram of a comparison device for use in the search method by means of a single coordinate.

In the system shown in Figure 1 runs over a request line loo a request. An operator at a control panel receives lol the inquiry for telephone number information. A control console 1o2 includes a keyboard for the operator to enter data. One Printing device 105 provides the information output for the operator Als a standby processor for emergencies or for an overload the plant. A data memory 1o4 stores all data entries and has a direct Accessibility. Further data memories 1o4 'and 1o4' 'store the data entries in a different order from that of the data memory 104. If, for example the data memory 1o4 stores the data input in the order NAO, that is the name, place of residence, address and occupation are saved in the data memory 1o4 'the entries in the order NOA, while the data memory 1o4' 'the entries saves in the order OAN. The structure of an additional input device 1o6 is shown in detail in Figure 1A. A fast output device 107 can, for example be a line printer. The structure of the output device 1o8 is shown in detail in FIG 1s shown. A reading device 1o9 is provided for entering new data, when a data group is first compiled, when it is complete is revised or when minor changes need to be made. A buffer store 11o is intended for a zone that connects different charge zones to the Establish mediation between them. Another buffer memory 111 operates as a multiplex that carries the information from or to all input or output devices, which are available to any information provider, collects or distributes. A storage device 112 stores the query information for the relevant responses cannot be found for later statistical analysis.

Switching devices s / w enable switching from the regular ones Devices to the auxiliary devices.

As shown in Figure 1A, the auxiliary input device 106 of each Information helper used to answer inquiries from telephone subscribers received, including the name of the desired subscriber, and as far as possible together with information about their place of residence. If the request information is not sufficient, the operator asks again. The additional input device 1o6 records the information in place of the operator. For example can the message generated by the additional input device 1o6 read 'Please enter the name, place of residence and occupation of the desired subscriber '. This message is recorded in the message unit 1061. This news unit 1061 is due to a connection signal with the inquiry line loo via the conventional telephone switching device connected and transmits this message to the inquiring party via a differential transformer 1062.

In conventional electronic transmission systems in which the Network impedance NW is equal to the line impedance, are the transmission losses large, almost infinite, between the message unit 1061 and a recorder 1063. In these circumstances, any output from the messaging unit 1o61 transmitted to the line and not recorded by the recorder 1063.

If the inquirer with the name, the place of residence and the occupation of the desired Telephone subscriber answers, his answer is to the recording device 1063 and the communication unit 1061 via the differential transmitter 1062.

The answer is recorded in the recorder 1063 which transmits the answer to the control panel. When the operator closes the exit of the recorder 1063 is addressed, the recorded content for the Usage by the operator.

FIG. 1B shows an output device 108 which is used for additional output processes is provided and a controller 1081, an automatic sounder 1o82 and a sound storage unit 1083. The central computer and control unit 1o3 generates a response that includes an operator identification symbol. This Symbol actuates controller 1081, causing a connection switch the corresponding control panel lol is located. The sounder 1082 is activated the language stored in the sound storage unit 1o83.

The sound storage unit 1083 stores spoken words such as for example the phonetic character 'wo', which corresponds to the digit 1, in the binary code o1, 'tau', which corresponds to the number 2, in the binary code lo, etc. The sound generator 1o82 scans and addresses the corresponding stored words.

The sound output 1082 receives signals and generates in response the same number pronunciation that corresponds to the coded number and in addition to it Word strings such as 'the number is', this phrase being pre-recorded is, '321-6291', this part of the binary codes being output again.

According to the invention, the request information is transmitted by converting means converted in the central computing and control unit 103 so that the request information is composed of main codes or concepts for each query element. Of the The query index formed from the query information is combined with the keywords compared that in the same way as the query index from the data entries be prepared in the data memory 1o4. The query index has the format NAO (name, Address, occupation) for each subscriber.

Before continuing the description of the directory assistance system, it is useful, briefly, the preferred embodiment for the signal format that used in conjunction with the system.

As shown in Figure 2A, the signal used in the system consists of digitally coded data. With every occurrence of key impulses, which are triggered by a system key are generated, an 8-bit character is recorded as a bit parallel. Every word W comprises a number of digits D1 .... Dn. Each data block in turn comprises again a number of words Wi .... Wn. Furthermore, each block contains a specific one Word WS in the first position of the block. W5 is the first word that marks the beginning each block. Another one certain word We are at that last position of a group of blocks and is used as the final word designated.

As FIG. 2B shows, each word comprises a certain number of digits Db, D1, D2 .... Dns where the break digit Db at the beginning of each word is the Indicates word breaks. A certain word We includes a break digit Db and three special digits Ds. Another particular word We includes three interruption digits Db and a special digit Ds, as shown in Figure 2C. Accordingly, it is possible count the number of data by counting the number of words W8, or it is also possible to count the number of words by counting the number of digits Count db.

As shown in Figure 2D, the request information is in a block organized, in particular from an initial word Ws, an order index Iq, another start word Ws, the corresponding operating point number and one Question number exists. The block of inquiry information is called the 'inquiry sentence' and an information group during the information search. Therefore this request block can also be called the start block. Thus there is an information group from a query record and an end block when the query information enters the system is fed in for processing. The information group includes a starting block at the beginning and a data block B1 .... Bf, followed by an end block in the im System found information.

The data block combines Ws, the data input Im with respect to the Main index and the subscriber's data. The end block includes W5 and Although the description of this invention applies only to an automatic directory assistance system It is understood that the invention is also applicable to more general ones Applications, such as looking for real information due to of ambiguous request data.

Regarding the procedure for entering new data in the Data storage, the details will be discussed later. However is it is necessary at this point to better understand the information search that is described below, followed by a brief explanation of how to enter data allow.

The data, which have the format of the data block shown in Figure 2D, are stored in the data memory 1o4 by a data processor or programmer saved. The data processor or programmer prepares the in the main index Information to be stored for each item of information for each subscriber to be registered before. The data of each telephone subscriber was on a fine tape or punched on cards.

The cards are in the reading device 109, which is shown in FIGS and FIG. 3B is inserted. This data is then converted into the main index format converted by unit 32o in Figure 3B. The main index and the other data of each subscriber flow into the data memory 104 shown in Figure 3A is. The counter 218 counts the number of data received from the unit 32o fed into the data memory 1o4. At any point in time at which the data count 85% of the capacity of a column or track of the data memory is reached the counter 218 a signal that the control unit (Figure 3A) of the data memory 1o4 is fed. The control unit reacts to this from the counter 218 generated signal and changes from the recording track just written of the data memory 1o4 to the next recording track.

The remaining area of the recording track, that is 15 8, will be released to provide additional recording space for further future subscribers to own. At every point in time at which a recording track change takes place, becomes the corresponding main index at one end of the recording track is arranged, a direct access memory 2o1 (Figure 3A> transferred and stored as a keyword. All data entries are made in Data storage 1o4 recorded. The data in each recording track or in each column of the Data memory 1o4 are then transferred to a unit 32o (FIG. 3B) for calculation read in a level or row number. The level value of each specification is indicated by determines the inclusion of sub-elements of the main index with respect to each data entry. The multiplication of the sub-elements will be discussed later. The level number then becomes the data storage controller (Figure 3A) along with each data entry transferred, which in turn is reversed for storing the level number area of the corresponding recording track of the data memory 1o4 is used.

It can be seen that all data entries are stored in the data memory 1o4 will. In practice, the operator receives an inquiry via telephone directory inquiries.

For example in the case of 'Alexander Chemical Industry Inc.', 14 Lexington Avenue, Postzone 16, New York, the index is made up as follows: Alexander N1 Chemical N2 16 A1 Lexington A2 Chemical Industry Occupation Ol k O2 1 In With regard to the occupation data k and 1 listed above, there is a table for Coding of job titles given in Appendix 2 at the end of the description. As can be seen from Appendix 2, the main code k includes the chemical industry, while the more specific code is given by 1. The request information will be entered into the system in the following way: Alexander # Chemical # 16 + Lexington # Chemical Industry. In the system according to the invention, the request information shortened and converted to a given length and format, which reads as follows: ALEXA # CHEMI # o16 + LEXIN # k 1 before a search process for a telephone number corresponding to this identified telephone subscriber is started.

The index for each input quantity is called the 'main index Im'. The index that is determined by the query information is called 'query index 1q' Some of the main indexes of all telephone subscribers required for a comparison process are used with 1g are selected for storage in the access memory and referred to as 'Keyword KW'.

Furthermore, it has proven to be useful to specify the desk number, add the request number and a format symbol 1 or II to 1q. The lectern and the inquiries are made automatically through the keypad on the control panel 1o2 (Figure 1) immediately after the operator presses the request button, transfer. The control panel 1o2 includes a counter for the number of set Queries and a wired memory, not shown, that holds the desk identification number contains. The control panel with the keyboard 1o2 is of conventional design, so that a detailed description is not necessary.

The format symbol 1 or II enables the operator to create a to select from two search modes for the data memory 1o4. The format icon 1 indicates that the data memory 1o4 is to be queried. It is reminded that the data memory 1o4 contains the data in the order NAO. on the other hand the format symbol II indicates that a search is being carried out in the data memory 1o4 ', which, as already mentioned, contains the data stored in the order NOA.

Another possibility is for the operator to use bracket symbols (and) used when it is unclear which part of a multi-part name is considered to be the most significant part N1 should be selected. For example from Asks the inquirer the name 'Miami Hilton Hotel', it is unclear whether now the inquirer the 'Hilton Hotel' in Miami or dds 'Miami Hotel' owned by Hilton thinks. In this case, the operator refrains from making a decision by typing the input data as' N (N ') AO'. In practice, the operator becomes Enter the request as follows: N1 (N1,> Nr N2 ~ A1 * A2 ~ 0 ~ 1 (or II).

In addition to this data, the desk and request number, the Following the request information, transmitted. The desk and request number are automatically generated by the control panel.

It should be noted that it is often not necessary to use the bracket symbol as the most significant part of a name is generally known. The uncertainties in deciding which part of a name is most significant by using the 'rule for N1 ', which is listed in Appendix 1 at the end of the description. However can always be used by the operator in the event of certain uncertainties respectively.

If there are uncertainties, use the bracket symbol.

The function of the symbol recognition circle, that of the central computer and control unit according to Figures 3A and 3B is assigned, will be described below.

Symbol recognition circuit (Figure 4A) As Figure 4A shows, the Inquiry information, which has the format N (N ') AO, of the symbol recognition circuit from buffer memory 111 (Figure 3B) along line 348. One detector 331 responds to the open bracket symbol t and transmits a recognition signal to a gate 334 as well as a switch 335 and a counter 330.

The normal current path is through switch 335 from the gate 334 to a switch 341. The element N1 contains the query index which is indicated by the Switch 341 flows into register 336. Immediately afterwards, responding to the Output of the detector 331, the switch 335 changes the current path in direction a register 338.

The open parenthesis signal is obtained by blocking gate 334 selected, which closes when the open parenthesis signal appears. Then N1 ' stored in a register 336 and a detector 332 switches the switch 335 back to the normal current path as described above. Simultaneously closes gate 334 on occurrence of the closed parenthesis signal).

The counter 33o starts the strobe pulses at the time of recognition to count the bracket signals. Having generated a predetermined number of key pulses the output of counter 33o opens gate 334. A delay circuit 333 delays the information signal during the switching transition of switch 335 by prevent erroneous signal overlap as shown in Figure 4B.

A detector 342 is responsive to the symbol # which identifies the elements of the Data groups that make up the index are separated from each other. A detector 343 controls in response to the number marking, a switch 341. After changing the State of the switch 341 as a result of the number marking, the data AO and the subsequent information is stored in registers 337 and 339. Both registers 337 and 339 store the end of the request information, that is the desk number and the request tuner.

Memory cells for the desk number and the request number are in the Registers 337 and 339 respectively, and these cells include one-bit redundant Cells one. Since the request index in the format N (N1 ') N2A1 # A2O102I (or II) nd (desk number) ng (request number) is formed, the first symbol # is followed by the Detector 342 detected the current path from register 336 to registers 337 and 339 to be switched. The switch 341 does not return to its original state State back after it was toggled by the first symbol, ~, despite the Fact that the detector 342 identifies the second symbol # until the switch 341 receives an idle signal from the control counter 347.

The detector 34c responds to the symbol 1 or II to the gate 343, which usually closes when it responds to the clock pulses on the Line 349 responds. After the detector 34o detects the format symbol 1 or II has, the clock pulses trigger the contents of the AO register 337.

These output pulses are counted by the counter 347. When the pulse count reaches a predetermined number, the counter 347 transmits a readout signal, whereby causing this to pass the clock pulses on line 349. The counter 347 also transmits a terminator CL1 to counter 346, switches 341 and 347 to registers 336 and 337.

The counter 346 starts counting the clock pulses at the point in time to which he receives the final character CL1.

When register 339 receives a read out signal from counter 347, the redundant cell of the register 339 changes to the logical state 1. the corresponding cell of register 337 retains the logic state zero. Corresponding then, when the query volume in the binary code is lol, the read-out signal is loll in register 339 and 1010 in register 337.

The final signal CL1 closes the gate 333 and runs the switch SW2, which carries the position monitor 341, returns to its original position. After this the counter 346 has reached a predetermined count value, it transmits a hold signal CL2 to gate 344 to block the clock pulses.

This hold signal CL2 makes both registers 338 and 339 free and the Index 1q is separated into the two parts NAO and N'AO.

The circuit 3o5 (FIG. 3B) operates as an address converter and receives Address data in unencrypted form and contains a conversion table, a street address, a postal zone or a district name in to convert a numeric address code A1 for further processing. These Further processing takes place with the occupation information by adding an occupation table in another converter 307 is included, as well.

Address converter 3o5 and occupation data converter 3o7 As the figures 5 and 6, rough information A1A2, which in some cases can be a noun, or in other cases is a mnemonic symbol, one Decoding circuit 4o1 fed to the coarse information in an input register 4o2 feeds. A memory size 405, which in the case of the address converter 3o5 Area titles A1 or, in the case of the job information converter, 307 job titles ° 1 ° 2 which match the numeric codes corresponding to either the areas or job titles represent continuously feeds their content into registers 4o4 and 4o6.

Each test data entry A1 or ° 1 ° 2 from the memory table 4o5 is sent to the attempt register 4o4. At the same time the corresponding Code representing a field or occupation, a converter register 4o6 fed in. A comparison device 403 compares the contents of the successive ones Test designations in register 4o4 with the query designation in register 406. If the comparator 403 finds a coincidence, it provides one Trigger pulse to register 406, which causes the contents of the register 4o6 are transmitted to an output register 4o7. +) will In certain situations The operator may be unsure of how best to use a particular job title in accordance with the classification as given the table in the Annex II at the end of the description specifies classifying. In other cases it can happen that the person ordering the desired telephone subscriber's occupation does not know.

In such a case, the operator gives a blank code - in the occupational group position of 1q, so that this index reads NA-. On the other hand, if the requesting party does not know the address of the desired subscriber, the operator inputs the blank code in the address group position so that the index 1q then reads NO-.

As can best be seen from FIG. 5, the converter converts 4o4 of the address converter 305 or the occupation information converter 307 the blank code to a broad or 'at the top of the list' numeric code, a numerical connection code and an identification code ++, for example as follows: oil ++ 80. This identification code acts as an instruction to the system, the area of query indexes I must be sampled from N1A1o1 to N1A180. The symbols 01 ++ 80 q are used by the detector 213 found.

Regarding the address information, it should be said that after the Designation A1 or its corresponding empty code was transmitted to the decoding circuit 4o1 (Figure 5) responds to the division character + and A2 directly to the A2 register 3o6 '(Figure 3B). This assumes that A2 does not require any conversion and can be processed in the rough form.

It should be noted that the computer system of the invention is suitable is used by a number of operators in real time at the same time to become. This is one of the important advantages of the invention because it a quick telephone inquiry can be provided with minimal personnel expenditure. To make this possible, the request is sent from each control panel 1o1 (FIG. 1) added an identification number that enables the system to make an operator information request to separate from the other. The registry 4o1 stores this identification number, while it only forwards the address or the professional information to register 4o2. Accordingly, the register 401 transmits after the appropriate code has been received and the output from register 4o6 to register 4o7 has been made, the identification number to register 407. In this, the identification number with the address information recombined in coded form as a result of the conversion process. the combined identification and address information is then stored in register 4o7 for transfer the further processing.

The occupation information converter 3o7 is described in detail with reference to FIG Figure 6 described. A decoding circuit 6o1 in Figure 6 corresponds to the circuit 4o1 in FIG. 5. A buffer store 6o1-1 stores at the input of the circuit 6o1 the request information. The stored request information is from the memory 6o1-1 read successively into a register 6o2 via a switch 6o1-3. One transferred to the input register 602, which corresponds to the register 402 in FIG Occupation information is stored in register 6o2-2.

If the occupation contains a blank code, such as NA-, identified the blank code device 6o8 the blank code signal. The blank code identification signal is then transferred to a blocking gate 6o6-2 for storage. The blocking gate 6o6-2 blocks the occupational information for passage through the circuit according to figure 6, while the memory of this circuit stores the blank code signal. The inquiry job data are transferred to an output register 607, which corresponds to register 407 in FIG and stored in this.

The request information has the form ~ occupation information 1 (or II) # Desk number # Inquiry number 'or' - (blank code) # Desk number # Inquiry number '. That first symbol is identified by a detector 6o1-2, which detects the state of the Switch 6o1-3 changes. After the position of switch 6o1-3 has been changed, the data will be ~ 1 (or II) # desk number # Inquiry number 'a Output register 6o7 supplied and stored in this.

The request information signal is also sent to a pulse gate 6o5-1 in circuit section 605, which corresponds to section 405 in FIG and the information signal is stored in the memory associated with the gate 6o5-1 is. While the gate 6o5-1 stores the information signal, it leaves clock or Time impulses to read out the memory or general occupation codes. The clock or time pulses are a counter 6o5-2 and the storage units in the Circuit section 6o5 fed to read this. The counter 6o5-2 counts the impulses and controls the timing. The meter transmits at the same time 6o5-2 a signal to the comparison device 6o3 and a further counting signal the storage controller 605-3. The memory controller 6o5-3 controls the reading of the memory units in the circuit section 6o5 and in particular the memory unit 6o5-5 feeds theirs Contents in register 6o4. At the same time, the in the memory unit 6o5-6 stored general code, that of occupation or special code, that is stored in the memory unit 6o5-5 is read into the register 606-1. The entire register part 6o6 corresponds to register 4o6 in FIG. 5. The contents the registers 6o2 and 6o4 are compared with one another in the comparison device 6o3 and a coincidence signal to the register 6o6-1 for the purpose of signal readout, the Blocking gate 6o6-2 for storage and the OR gate 6o7-5 for signaling the completion of the comparison process or to indicate the end of the conversion forwarded. The contents of the registers 6o6-1 are transferred to the OR gate 607-4 in the register 3o8 fed in.

The blocking gate 606-2 receives this from the memory regulator 605-3 Signal indicating that the output of the memory unit 605 containing the memory circuits 605-5 and 605-6 which has reached the end data. When coincidence between the Contents of registers 6o2 and 6o3 occurs, the output signal of the memory controller 6o5-3 blocked by a gate 6o6-2. If no coincidence up to the last level occurs, the signal of the memory controller 6o5-3 passes through the gate 6o6-2 and activates the symbol generator 6o6-3 for a. The generator 6o6-3 transmits a to the output register 607. The output of the gate 6o6-2 is delayed until the signal a has been transmitted and then gate 606-4 opens to pulse a To pass output register 6o7.

The contents of the individual register circuits 607-1, 6o7-2 and 6o7-3 are transferred to the designated control panel in their order, and then to the storage device 112, for example a magnetic tape shown in FIG is.

The transmission of the signal a indicates that a specific job information could not be found by the storage unit 6o5 and that it is necessary is to use a different specification for the request.

A detector 6o8-1 transmits the dummy code identification signal. A blank code generator 6o8-2 produces an output signal such as ol ++ 8o. The latter signal is via an OR gate 6o7-4 in the output register 6o7 of the device 308a in FIG 3B forwarded. The former signal goes directly into device 308b, which is shown in FIG Figure 8A is shown.

It should be noted that with regard to the search method used in the present invention is applied, in one embodiment a continuous one Table checking is done, while in another embodiment a partial comparison is made. These methods of retrieval are detailed below to be discussed.

The output of the empty signal detector 6o8-1 or the comparison device 6o3 indicates the completion of the operation and is stored in a storage unit, which is contained in an AND gate 6o1-4. Similarly, the concluding Signal from an address converter, for example the converter 3o5 (Figure 3B) is applied to gate 6o1-4. The first of the two signals is waiting the latter signal, and when the signals are fed into gate 6o1-4 at the same time the gate transfers the output to latch 601-1. on the other hand makes the termination signal through the OR gate 6o7-5 all memory units and Registers in the occupation information converter circuit shown in Figure 6 are free.

As described above, a pseudo-numeric or an alphabetical representation of the name information N1 is supplied to the register 3o4, an alphabetical representation of the name information N2 is entered in register 3o4 ' fed a numerical representation of an appropriate code that of the address information A1 corresponds, - is fed to register 3o6, an alphabetical representation the address information A2 to the register 3o6 'and a numerical representation of a corresponding code, which corresponds to the professional information information ° 1 ° 2, arrives in the job information register 308.

Intermediate processing of the request information 1g The circuit 3o4 according to FIG. 3B, the excess characters are cut off when these are carried out by the operator entered into the directory assistance system. Circuits for disconnection of excess characters are known and therefore such a specific one Circuit not described in detail. Each of registers 304, 304 ', 3o6, and 306 'contains a register for storing the input characters and a blocking element, which is placed in front of the storage register and is normally open, however is closed after the input of a signal from a counter.

The next procedural step for the request information I g consists in passing the query index I1 = N1A1O1O2 to register 31o, except the index information N2 and A2 for the reduction that is directly der Comparison device 314 are supplied. The index 1q can be in the register 31o can be entered in any order required for a particular information request suitable is. For example, if the inquirer provides the name and address of the desired subscriber knows, but only approximate information about his occupation can make, the operator will terminate the inquiry information complete like 1. The end character 1 instructs the directory assistance system according to the invention to arrange the query information so that a query index 1g in the register 31o is entered in the order NAO. If on the other hand the inquirer details of the name and occupation of the desired subscriber than can do via its address, the operator will use a different terminator like Ii at the end of the query information, indicating that the Inquiry index 1q should be entered into register 31o in the order NOA. A controller 3o9 differentiates the end characters 1 and II. The controller 3o9 breaks down the Output of the control panel 1o2 (Figure 1) in sections and regulates the order in which the registers 304, 3o6 and 3o8 feed the information N, A and O to the register 31o.

Further processing of the information is then carried out under the Assumption described that the request information 1g the system according to the invention is supplied in the order NAO. But it goes without saying that the operational sequence is similar for a query information that has the order NOA or 0102A1, with the exception that another section is the data memory 104, for example a magnetic drum storage (Figure 1) is used.

The query index I = NAO is used together with the operator panel identification code from register 310 to register 273 (Figure 3A). The register 273 stores the information N1A10102 but removes the control panel identification code and transmits this code to a register 214.

Devices 273, 309 and 31o The circuits I-IDG and II-IDF im Circuit section 3o9 in FIG. 8A are the recognition circuits for the end characters 1 and II. One of the two circuits transmits a start signal to the register 306 or 3o8 through the gate G1 or G2. For example, in the case of one End character 1 circuits CTR2, OR3 and G6 in circuit section 3o6 for the transmission the contents of the address register A (address code) to the register 31o by the O register or

Triggering the occupational data register 3o8 is planned. A blocking gate INH.G1 in circuit section 308a maintains the exact order of the operation of the Register 3o6 and 3o8.

A switch SW of the register 310, which is activated by the output of the O register in the circuit section 3o8, and the A register in the circuit section 3o6 are connected to register 310 ', which is similar to register 31o in FIG. 8B.

If the end character I is considered, the result is that the information AO is fed into a register REG3 of register 31o (FIG. 8B), while register REG1 and REG2 store the information N1 generated by the register 3o4.

A circuit CTR.DEV1 in register 31o (FIG. 8B) is used as a counting circuit for the transfer of the contents of the registers REG1 and REG2 into the device 273 (Figure 3A or 8B) and provided in a comparison device 272 (Figure 8A).

The register REG1 is output on the basis of the signal from the Gate OR8 in register device 31o (Figure 8A), the signal indicating that the transmission of the information AO is completed. A predetermined number of Bits contained in N1 are counted by CTR1, which is the count signal to the Gates G2 in the circuit CTR.DEV1 transmits.

After the contents of REG1 are transferred to the comparator 272 the contents of REG2 are then fed in. The latter will be also the a-transformation 1er of the device 273 (Figure 8B) via the Switch SW3 supplied. In the a converter, the AO code is converted into a numeric value converted for multiplication in a calculator 216 (Figure 3A).

If an empty code is found in index 1g in converter 3o5 or 3o7, a gate OR6 of the register device 31o in FIG. 8A receives an identification signal. This signal activates the coding device for the empty code (see register device 31o in FIG. 8A) in order to generate a symbol and this in register REG4 of the register device 31o to be recorded. The counting circuit CTR.DEV2 in the register device 31o in Figure 8B is actuated by a signal from gate OR8 and transmits the contents the registers REG1, REG3 and REG4 to the device 214 in the form NA-.

As has been described in connection with FIG. 6, the contents of register 3o8 (FIG. 8B> a format corresponding to olt8o, if an empty code is contained in the query index 1q. This information o1t80 is sent to the facility 273 transferred. The symbol t is identified by circuit 213 in Figure 3A or 8B and this symbol controls switch SW1 in device 273 (Figure 8B>. For example, the portion ol is an upper address register UPP.

REG and then the t-characteristic switches the signal path to a lower address register LOW.REG in device 273 (FIG. 8B). The counting circuit CTR.DEV3 transmits the readout signals to the mentioned lower ones and upper registers when the counting circuit has a predetermined number of signals after receiving a signal from a gate G9 or G10 in Figure 8A Has. Address code A was registered before this counting step.

Aol and A8o go to different address converters on different paths a headed. The switch SW2 in Figure 8B corresponds to that indicated by the position number 215 marked switch SW in Figure 3A.

Partial search method In order to be able to bring about a quick determination, on which drum memory track the desired information is located the key words K, which represent all memory tracks, are checked in such a way that referred to as the 'split point' method. For example, in one Group of key tracks the keyword for the track names T / 2 the 1/2-point keyword this group.

Similarly, the T / 4 keyword represents the 1/4 dots, the T / 8 and 3T / 8 Keywords representing 1/8 and 3/8 subdivisions of the group, etc.

The division point search system involves the successive division by the number 2 of the key tracks, i.e. it is formed T / 2, T / 4 and T / 8 etc., as long as T is an integer power of 2, i.e. T = 2m, with m = one integer. Accordingly, the total number of Troinel memory tracks becomes two Parts divided, namely into a group that represents the greatest possible power of 2 and in residual groups. Contains, for example, a characteristic drum storage system 12ovo traces, so the greatest power contained in the number 1200 = 2m = 21o = 1o24. The first 1024 memory tracks therefore form the first group. Of the The first group step for a search consists of the query index Iq, which contains the represents the desired subscriber, with the last keyword K the To compare the first sub-point search group, that is the keyword K of the 1o24-th Storage track. If the query code 1q is less than or equal to the 1o24-th trace code K, then it is certain that the desired subscriber is in the first Partial search group, that is contained in the first 1024 recording tracks is.

On the other hand, it is certain if the query index 1q is a larger one has numeric value as the keyword K for the recording track owner 1024, that the desired recording track in the rest group, that is in the last 176 recording tracks out of the total number of 12ovo recording tracks is to be found. To find out which of these 176 recording tracks is the right one a partial point approximation is made with the 176 recording tracks. In particular, the remaining 176 recording tracks can be grouped into a sub-point group and a secondary residual group. Then search in the same Wise continued by applying partial point analysis to this group. Finally, a last-ranked remainder group is reached, which is too small to have another Partial point analysis to justify and the keywords K of this last ranking Remaining groups are simply sequential for a comparison with the query index I. scanned. In practice, g is the first-rate remainder group of 176 recording tracks not judged to be large enough for a block of memory to approximate a partial point to justify.

Two-dimensional partial point search All recording track keywords K are stored in the computing system's access core memory for fast To enable comparison with the query index 1q.

In order to locate each desired keyword K as quickly as possible in the core memory, the step of determining the memory address of each keyword K included in the partial point analysis can be avoided by directly using the memory address in the division process as an x, y coordinate memory address. According to this method, each subdot keyword K takes the form K. K represents each keyword in general, xy while the letter x represents both the x storage coordinate and the facility number divided by two. The letter y represents both the y-memory coordinate and the operating number divided by two within a given division level. If, for example, the group is divided into quarters within a second division level for which x- = 2, the first subpoint separates Y. = 1 the first quarter of the group from the second quarter, while the second subpoint Y = 2 separates the third quarter from the fourth quarter. This process is summarized by the following diagram, which gives a general method for developing the subdot keywords of the form K xy: As the diagram shows, xKy = 1R for the first division level. There are two division points 2K1 and 2X2 in the second level, and this continues in the following division levels. 1K1 is the 1/2 division point, 2K1 the 1/4 division point, etc. It can help to improve understanding when assessing this diagram if it is thought of as a drum store that has been spread out on a plane. Record tracks of the memory extend from the top of the diagram to the bottom, with the record track represented by 1K1 being in the middle of the trointel. This recording track is meant by the 1/2 division point. 2K is the 1/4 division point corresponding to a recording track between the tip of the drum and the center thereof, while 2K2 is equal to the 3/4 division point. In general, the division point is n / t = (2y »2X, where n is a specific record number with a keyword ### and t is the total number of record tracks in the sub-point group.

It is pointed out that if the address number of the Core memory 2o1 by the number following the core address much easier the following formula can be obtained: z = 2X 1 + y - 1 Further, 2z = 2 (X + 1) -1 + (2y-1) - 1 and 2z + 1 = 2 (X + 1) -1 + (2y) - 1 The next address of the core memory is selected by calculating '2z' or '2z + 1', in accordance with, whether the query index 1g is smaller or larger than the keyword that starts with 'z' corresponds. This one-coordinate method will be discussed later with reference to FIG 13 should be explained.

The following explanation refers to the two-coordinate method.

If the value of in or the number of keywords is too large, to store key words in the given volume of the memory 2o1 all keywords divided into many groups that start on the letter based on N1.

Initial Letter Identifier 321 The initial letter identifier circuit shown in Figure 3A 321 checks the first letter of N1 in register 3o4 and defines the appropriate one Group. The output of the identification circuit 321 gives the instruction to an external memory FOND, for example a fast magnetic drum storage, a defined group to read out the corresponding Has keyword. figure 7 shows details of the identification circuit 321 which stores in a memory 361 the alphabetical Letters A to Z, completed by the assignment numbers na to nz in the external Includes memory FMD. The assignment numbers na to nz correspond to the alphabetical ones Letters A to Z. If the signal, this is the request information N1 in the register 3o4 is stored, or the switch 3o2 the signal path from register 3o4 to the Register 304 'has changed, the first letter identifier circuit 321 is reached, the reading of the table 361 into a register 362 or into a register 363 is initiated. The register 362 stores the letters A to Z as a label and the register 353 stores the FMD area numbers. If the comparison circuit 364 makes a match between the contents of register 362 and register 3o4 for the initial letters of N1, a coincidence signal operates register 363. The output of the Register 363 is sent to an FND controller (not shown). The group of Keywords that are transmitted from the FMD memory are called the core memory 2o1 (Figure 3A> supplied, while the rough query 1q in an encrypted Index 1q in N1 register 3o4 by the Iq circuit 310 described above was, is transformed and together with the group a certain value m for the group is transferred from the memory FND to the circuit 2o9 (FIG. 3A).

The query index 1q = NAO is stored in the comparison device 272 (FIG 3A), which carries out successive stages of the comparison in order to determine the Find the corresponding keyword XKy that is the desired memory record track n denotes. At the beginning the x-register 222 and the y-register 22o are reset, so that their contents correspond to the logical state 1. The logical value 1 in x register 222 causes an x address driver stage 2o3 to energize the line, which corresponds to the x = 1 coordinate in the core memory 2o1. The logical value 1 in y register 22o is fed to the computer circuits 2o6 and 2o6 'and additionally causes he that a y-address driver stage 2o2 energizes the line in the core memory 2o1, which corresponds to the coordinate y = 1. The coincidence of the x and y drive signals causes the contents of the core part 2o1 in the address x = 1, y = 1 in the register 2o8 are managed. The information that register 2o8 receives includes this Word xKy = 1K1 together with the number n of the memory record track that includes it is detected. This information is then transferred to register 271 which the xKy component separates from the n component. The K component becomes the comparison xy means 272 for the comparison with the Iq information received from the register 273 forwarded.

If the result of the comparison in the comparison device 272 indicates that 1q is less than the keyword 1K1, this is the 1/2 keyword, causes an output signal of the comparison device 272 to a computer 276, the performed the computation y = 2Y-1 immediately after finding the y-value of the register 22o has received the contents of the register 22o as a new trial value for y is entered.

On the other hand, if the comparison in the comparison device 272 indicates that 1q exceeds the xKy value, an output from the comparator 272 to the computer 2o6 'which has carried out the calculation y = 2y, immediately after receiving the y-value and furthermore the content of the Register 22o entered as a new trial value y. After one of the calculators 2o6 or 2o6 'has changed the content of the y register 22o, it carries the new one Value for y again the two computers simultaneously with the transfer of the new one Value for y from register 22o to register 2o2.

Otherwise, an output signal from the comparison device 272 is added at every point in time at which a comparison is carried out, regardless of whether 1q or the keyword xKy is greater, the number 1 corresponds to the value x in register 2o2 and the number 1 is added to the content of a counting circuit 2o9 added, which begins with the initial value zero. In the same way, at the new keyword xKy together with the new record number for each comparison n is fed from the core memory 2o1 to the registers 2o8 and 271 and the new value from x Ç to the comparison device 272 for further comparison with Tq.

A third possibility arises when the comparison device 272 Equality between 1q and one of the successive test values of the Keyword xKy determines that an output signal from the comparison device 272 the gate 2o9 'triggers. One of the outputs of the gate 2o9 'then reads the existing recording track number n from the register 271 to use them for the Instruction to the data memory 1o4 to use to output this recording track. Further outputs of the gate 209 'represent the y register 22o and the x register 222 to its initial value 1, run the counting circuit 2o9 to its initial value logic value O and clear the comparison device 272. The counting circuit 2o9 records how many comparison steps, i.e. how many stages x of the partial point analysis were performed by the comparison device 272. As mentioned before, is, since x = the number of divisions by 2, the last possible level of partial point analysis achieved when x = in, since the partial point search group is only the first 2m record tracks of the data memory. In a practical example this means that in 1o and 21o - 1024 recording tracks are. Therefore only in = lo steps can be used in the partial point comparison can be performed by the comparison device 272.

The gate 209 'generates a readout signal at the time when Iq = xKy only if x is not equal to m. Even if x is m and Iq equals xKy, since the signal from the gate 2o9 can pass through the output of the comparator 272 is delayed, the readout signal can be generated by the gate 2o9 '. After that Signal representing 1g = #Ky appears at the output of the comparison device 272, also occurs the signal x = m from the # gate 209 and the signal x = m is from the gate 2o9 'to clear the circuits 2o9, 22o and 222 or to return to the basic position.

In summary, it should be noted that the gate 2o9 'is a readout signal generated in response to the signal 1 = K, g xy before the state x = m is reached or that the counting circuit 2o9 does not reach this state. If there is no signal for 1g = #Ky, the gate 2o9 'generates a readout signal when the counting circuit 2o9 generates the signal x = m. In both situations, when the counting circuit 209 reaches the state x = m, the gate 2o9 'generates a ~ delete and on basic position bringing signal 'for the circuits 2o9, 22o and 222.

If at stage m = 1o a K = 1 is not generated, then xy becomes q the partial point search canceled. If x = m and 1g is even greater # Ky # then press the outputs from the counting circuit 2o9 and the comparator 272 an AND gate 21o, which causes the circuit 211 to send a logic 1 signal to the adder 212 transmits. The adder 212 performs an addition N + 1, which is the next Identifies the recording track number in the data memory 1o4.

Practical significance of the partial point search If 1g is greater than KW (5/8), which shows the keyword corresponding to the 5/8 division point, the number 1 is added to the number 5 and KW (3/4) is reached, which is the keyword for the represents earlier step. When the comparison step is completed in the state ~ 1q greater than the keyword ', the numerical value n in the x register 222 must be increased by the digit 1.

In the val. That more than two columns of 1q are identical, the means that the keywords are the same, it comes with the definition the column number causes confusion or ambiguity. For example In one case, the system can use the keyword as a second recording track or define column, or in another case, the system defines the keyword as the third column of the group of the same keywords. Indeed it must Computer read out the data from the first column to the last column of each group. Therefore, after the first step of data entry, the keywords are removed from the Core memory 2o1 successively in a working memory (not shown) read out and the associated recording track n modified only for this group.

As an example, assume that p, p + 1, p + 2, p + 3 is the column number of one group with the same keyword. A modification of the record number n means that p + 1 is reduced to p, p + 2 together with p + 3 is reduced to p etc., being in accordance with such a modification whenever the comparison with respect to a group is stopped, the readout instruction always starts with the first Column p begins. The change from the p to the p + 1 column is done by an overflow mark, which will be explained with reference to the level or level definition.

Each column has a key word compared to the previous one Column different keyword.

Entries of the same index I as this keyword result in q the connection to the next column. Therefore, partial entries must be made in the next Column outside this group due to overflow from the previous column can also be read out during the overflow procedure.

Level definition in the data storage recording tracks As indicated above, the reading out of all levels in a recording track n in correspondence with a sub-dot group by the Calculation of the upper and lower limits of the level range of this recording track or group of recording tracks within which a desired subscriber should be found should be avoided. This calculation is made by the calculator 216 (Figure 3A) receiving AO information from register 273, after the information is stored in a corresponding order quantity a with the help of a Conversion table in register 273 has been converted. The table is working due to input information according to the formula = = (A-1) x8o + o (if o 8o x occur can) and = = b x 5 + a, with b = the number of blocks in a recording track, 5 = the quotient, and a = the remainder of the division, smaller than b; that is, that then AO is converted to a. A portion of the calculator 216 multiplies a times q, um to determine the upper limit of the level number within the recording spectrum. A another portion of the calculator 216 performs the operation (a-1) x q + 1 to obtain the to determine the lower limit of the level number in the recording track. The information The computer 216 then provides information on the range of the recording track levels output to the control unit (FIG. 3A) of the data memory 1o4.

In accordance with a random selection of telephone subscribers, contained in the same AO block, a special constant q is determined. It should be noted that this constant q is not absolutely necessary, if the address calculation is only carried out by a partial point comparison. However, if the level definition is used, q is calculated as follows: The mean the number of telephone subscriber entries with the same AO plus one Standard deviation igmas. In practice it has been found that q is a whole Number is on the order of 1o. This value was according to the results determined from data samples in a public telephone directory. This statistical Calculation ensures that normally the number of participant entries, the successive recording track levels are assigned, the capacity not exceed a recording track. For example, if the usual number of entries belonging to a block, q = lo, the Entry area below and above q = 10.

How the level definition system works assumed that A ol = a1; and A 8o = a2, then the decoding circuit recognizes 213 (Figure 3A) the symbol and responds to it by energizing switch 215, Via which the alternative a1 from the register 273 to the one input channel of the Computer 216 and a2 to the other input channel or the input line of the computer 216 is forwarded. Along one of the input channels of the computer 216 is the smaller ordinal number a1 by 1 decreased and multiplied by g and then the result increased by 1 by the lower limit for the reading range of the recording tracks. In the other input channel of the computer 216 is the higher order value a2 multiplied by q to the upper limit of the range to be calculated (compare computer 216 in Figure 3A). As already mentioned, is the value of q is about 1o in practice.

After the determination or calculation of the recording track from the Adding circuit 212 and the results of the level range calculations from the computer 216 have been received, the control unit of the data memory 1o4 checks the designated Areas or parts of the memory. The desk code combining unit 214 combines the retrieved data with the desk name and the request number and transmits the information on the comparison device 311 (FIG. 3B).

Overflow detection The second detection circuit 217 (FIG. 3A) detects an overflow symbol% at the end of the response flow to a request and causes the Control unit, starting with the indicated location of the overflow at the end of the previous one Entry group to read the entries together with the symbol%. The symbol % is intended for the detection circuit 217 to go to the next readout address recognize which is transferred to the control unit of the data memory 1o4.

For the description below, a group of q levels will be referred to as called a 'block'. If the number of entries in a group is occasional q exceeds the first g entries in the first q levels of the block A, and a final level is used, the superlaur symbol% and to store a level number h. This creates an overflow command such as ~ Advance to the next level h, which belongs to block B 'given. Then in inscribed at the levels of block B markers §, the mean, 'this entry is an overflow'.

Each additional entry is then made below the level h des Block B entered. If necessary, especially if the number of telephone subscribers changed, the overflow can start at the first empty level of a block, that is not fully occupied. In this case, overflow markers § at level h and the following levels, the group of entries in levels 1 to h - 1 are stored from the overflow of another recording track, which is stored in the levels h + 1 and the following levels., to separate. The overflow mark is at each level of these overflow entries § saved together with the entries mentioned. The § symbol indicates that the entry belongs to the overflow from another block, and it can be detected by detector 312, the task of which is to detect incorrect information caused by an incorrect query index, to determine.

If a third recognition circle 312 (Figure 3B) the symbol § at the beginning of a readout block in a recording track, indicating that the following data belong to an overflow, then this means that with respect to 1q something must be wrong; on the other hand, namely, the calculations become first led to a non-overflow entry of a header block of an A1O1O2 group. The detection circuit or detector 312 examines the response of the circuit 214 and recognizes the overflow flag when it is in the first entry of the responses appears. The detector 312 then provides an output to a no response circuit 313. The output of detector 312 prepares this circuit 313, but without it trigger.

The block of entries determined by the computer 216 is shown in rearranged a block in the order of 'a' on the recording track, even if they are different as far as their N1 is concerned. Because of this, you can the Answers from a block of entries with different N1 with respect to the component N1 of a corresponding 1q.

The comparison circuit 311 receives the N1 code of the request directly from register 3o4 and compares it with the N1 codes contained in the system response information are included to determine if they apply. All the information that comes from come to the data memory 1o4 and who do not have an N1 code that corresponds to the N1 code in matches register 3o4 are excluded from further consideration by the system. When all responses from comparator 311 are turned off, prepare this presents the 'no response' circuit 313, but without triggering it.

N1 check circuit 311 Referring to Figure 9, details will be given the N1 test circuit 311, which is similar to the response counter 314 in Figure 3B, will be described. At the entrance, a detector 351 constitutes a selector with respect to N1, one End marker 1 or II and a start word Ws. A counter 352 is counted by a Initial word Ws signal triggered to count the clock pulses and transmits signals, to time the start and end of N1 while via a gate G2 the input information is running. The input information is also in a Memory 353 stored.

The N1 part in the input information is stored in an N1 response register 356-2 registered. The N1 component is processed with the N1 request in register 356-1 is stored, compared. A coincidence signal on line a from one Comparator 356-3 opens gate 354 to transmit a readout signal. A Coincidence signal on line b causes an output control circuit 357 to turn the amount of a data block jumps further. However, the signal is on the line b blocked as long as an output of an end detector 351-5 or of the detector 351-2 for the symbol 1 or II.

When an end detector 359 has a closing block in the output information identified, the detector transmits a clear or hold signal through which each Circuit or each circuit is returned to its original starting state, by delaying the signal as long as the counter 356-4 is operating. The counter 356-4 counts the number of coincidence pulses reached, if the count is zero an output of the end detector 359 enters the counter 356-4 and transmits a 'no answer' signal.

When the circuit 313 is prepared by either circuit 311 or 312 it transmits a signal to the N1 test circuit 311 which causes this transmits the query index 1q to the register 316 (FIG. 3B). The end sign 1 or II accompanies this information and is used to indicate to register 316 whether the search should be carried out in the order NAO or NOA. As answer on the end character symbol, the register 316 transfers the information in the displayed Sequence for the circuits connected to register 316.

First, N is fed directly to register 316 without any change. #Then A and O are sent to Error A converter 317 and Error O converter 318 (Fig 3B) supplied. The order of this transfer depends on whether the search is in the NAO or NOA sequence must be carried out, which is indicated by the end mark 1 or II is determined. The converter 317 changes A to the form A 'and the converter 318 Re-index O into the form O '. A 'and O' are trial corrections that make the trial represent to correct possible errors in the information provided by the inquirer made in his request.

The error correction will then be described in conjunction with the description of FIG. 12 will be explained.

If part of the information generated by the data memory 1o4 matches the original request with respect to N1, the information becomes one Reduction comparison device 314 forwarded. This receives the Contents of the N2 register 3o4 'and A2 register 3o6' and checks whether the N2 and A2 information contained in the response information with the N2 and / or A2 information of the original request recorded in these registers is to match. Answers that are not coincident on this will become excreted conventionally.

Since the N2 or A2 component, these are each the second most important terms of the request data, a high percentage of it is incorrect, can often be the case occur that a wrong N2 or A2 in a query with a correct N2 or A2 is compared in a registered entry of the data memory 1o4. Self if the comparison in the comparison circuit 314 does not result in coincidence, can it is a correct answer N2 or A2. This means that im If there is no coincidence, not all responses from data memory 1o4 are dropped must be, but that they are to be viewed as a system output that has been found. Only when a coincident entry occurs will the remaining ones fail coincident data, eliminated.

Reduction comparison device 314 with the aid of FIGS. 10A and 10B the operation of the reduction comparator 314 will be explained in detail. Area 314-1 in Figure 10A represents control symbols Ws, 1 and We of the input data group firmly and in response, actuates area 314-2. This area educates a first controller which includes three counters and two gates.

An area 314-3 represents a registration circuit, the two registers, an OR gate, switches SW1 and SW3 and a readout circuit with a switch SW2 contains. Area 314-3 delays the input data, while N2 and A2 in the response from the data memory 1o4 is compared with N2 and A2 in the query index 1q While the contents of R1 in area 314-3 through the switch SW3 are read out, N2 and A2 in the next answer, which is the content of R2 is compared to # 2 and A2 in the query index by comparator 314-4.

The comparator 314-4 includes a register for storing N2 and another register for storing A2 in the input data, and another an OR gate and an AND gate. The OR gate is for generating a Provide output signal in response to a coincidence for the control of SW3. SW3 in turn switches the output signal path to the memory circuit 3 in the area 314-7 (Figure 10B) around. The output of the AND gate switches to none in response Coincidence, the output path to the storage unit 4 in the area 314-8 (Figure loB) around. The non-coincident data are stored in register 4.

The area 314-5 is intended for the storage of N2 and A2, based on the desk number and the request number contained in 1q were found and compared in the comparator 314-4. The area 314-6 represents a second control unit, the five gates, two blocking gates, four OR gates are included for controlling the output data from the registration area 314-3 in such a way that the output data in the storage unit 3 or in the storage unit 4 depending on a coincidence or a non-existent coincidence with respect to of the reduction index N2 or A2 can be saved. This ensures that in both Storage units of the header (Figure 2D) and the trailer are not dropped will. When the output of the OR gate OR1 in the comparator 314-4 of Figure loA is transferred immediately, it is in the memory part of the gate G11 in the second Control device 314-6 and at the same time in the memory part of the blocking gate INH-G2 saved to block the current path.

After a predetermined count after the detection of an end mark We by detection circuit We1 in the detection area 314-1, the output of the Counter CTR1 in the first control device 314-2 to the gate G11 in the second Control device 314-6 supplied. The output of the gate G11 operates the gate G12 in memory section 314-7. The gate G12 transmits the readout pulses of the Storage unit 3. The storage section 314-7 stores the coincident data with respect to the reducing index N2 or A2 in the comparator 314-4, while non-coincident Data relating to the reducing index is stored in the storage section 314-8.

When the blocking gate INH-G2 receives a signal from the counter CTR1 in the range 314-2 passes, just during the time when the detection circuit We1 im Area 314-1 operates, the gate G13 in the memory section 314-8 starts the readout pulses to be transferred to the storage unit 4.

Additional comments on the reduction data N2, A2 According to empirical data Statistics is the percentage that both A2 and N2 are wrong and not are correct, almost zero. The percentage at which A2 is incorrect or absent, accounts for 32%, while the percentage of faulty N2 or total missing N2 15%. Since A2 and N2 are independent of each other, the probability is that A2 and N2 are invalid at the same time, equal to o, 32 x o, 15 = o, o5, that is to say the Probability that either A2 or N2 is valid is equal to 1, oo - o, o5 = o.95, in other words, A2 or N2 are 95% available at all times. It is therefore very rare that all responses from the N1 comparator 311 be transmitted to the counter circuit 315. As in the following empirical Table shows the number of requests that are distributed for the same Value of N1 can be found, as shown in the left column. Of the The mean of this distribution is clearly above 300 to 4ovo.

The solution to N1 is obtained using the other remaining input elements A or 0 improved. The NA column or NO column show the distribution with a Average value from 2o to 30.

If N, A and 0 are evaluated together, the mean value is of the distribution, as shown in the right column of the table, at 3.1. The number of queries for which more than two responses due to the same 1q obtained is 97 - 56 = 41. This result is obtained by the reduction comparison circuit 314 reduced to 2, according to the calculation 41 x o, o5 = 2.

Number of entries found by the same element N1 N1A1 N1O1O2 NAO 1 3 2o 19 56 2 6 6 14 14 3 1 7 6 5 4 7 2 9 5 1 7 3 6 5 4 2 7 2 4 4 2 8 1 2 1 9 - 1o 2 2 7 2 11 - 2o 5 14 13 3 21 - 4o 1o 14 12 2 41 - 8o 12 6 2 81 - 16o 12 7 4 161 - 32o 12 2 2 321 - 640 15 1 1 641 - 2560 3 2561 over 5 together 97 97 97 97 From the table it can be seen that in 97 attempts where NAO was entered, only a single number was found 56 times, while six numbers were found only twice. The system never gives again than 40 numbers when NAO is present in the request information.

Accordingly, it can be stated that the majority of the answers of the system can be given in the form of a single phone number that includes the desired Identify telephone subscribers. This allows the automatic directory assistance system according to Figure 1B can give the answer without the assistance of the operator. Will however, if more than two telephone numbers are output from the system, the operator must consult with the questioner which of these telephone numbers is the desired Can be number, that means in practice that an operator is always available have to be. As discussed above, the N2 and A2 request information each as a reducing index to reduce the number of considerations Answers are used and help reduce the operator's workload to belittle. The system answer to a question that goes through all tests is fed to counter 315 (Figure 3B).

Reply counter 315 It can be seen from FIG. 11 that the reply counter 315 comprises a detector part 351 which includes the 'header' and the 'tail' block, including Reference is made to FIGS. 2A to 2D. A gate 351-4 is through the determination of an initial word W5 opened in an identification circuit 351-1 and a query index format symbol 1 or II is detected by a flag 351-3. A gate 352 is opened by a header (Figure 2D> to the second Let signal W5 through, this is the starting word associated with the first data block is, in addition to the initial word associated with the second data block stands.

A counter 352 counts the number of data or entries that can be entered into the data memory 1o4. One signal corresponding to each data block activates the controller 355 which stores the input data. The data will be stored in a memory 353 in an order such that the header is stored in a first memory level, while the first data block is in is stored in the second memory level, etc.

When the counter 352 counts the data for identification, generated it sends an output to a switch 360 which has an output to the desk addressing device 1o5 (Figures 1 and 3B) and another output to the high speed printer 1o7 owns. A detector 351 detects the end block or the end value with a End detector 351-5. The output of the end detector 351-5 opens a gate 354 through an OR gate 358. Similarly, the output of a counter 352 opens Gate 354 through an OR gate 358.

Clock pulses for data readout lead through gate 354 to the Memory 353. The circuit paths are switched over by the control device 357, which is excited by each output of counter 356. The counter 356 transmits a Output signal corresponding to each data block cycle. If an end block is replaced by the Detector 351-5 detected or if the contents of counter 352 exceed zero, starts reading out the memory 353 and continues until a second End detector 359 the end block or word in the output from memory 353 identified. At this point the end detector 359 transmits a reset and stop signal to each element of the system.

If the number of answers found is a predetermined size C is greater than or equal to C, all responses would be to the high speed output device 1o7 transmitted for quick printing. However, the number of answers is the same or smaller than the specified size C, then it is assumed that there are so few answers that they are due to the relatively slow desk pressure unit 105, for example a printer, can be output. Hence the answers is transferred to the printer via the buffer memory 111 for printing.

Error Correction Circuits 317 and 318 The function of the error correction circuits 317 and 318 will be described in detail with reference to FIG. A faulty one Input information A or O is first given to an input register Sol, the the original numeric code A or O from the console identification code for the Information desk separates. The numerical code A or O is then sent to a comparator 5o2 forwarded, while the identification code for the information desk to an output register 5o6 is transmitted. An error correction memory table 504 is continuously scanned and their entries are passed on to registers 5o3 and 5o5 one after the other. the Information stored in table 5o4 comprises a list or

a catalog of correct A or O code entries along with the corresponding list of incorrect entries. These data are empirical compiled on the basis of experience that certain names or addresses and the like are very often confused with each other. Each pair of corresponding to each other correct and incorrect entries are read from table 5o4, with the wrong entry in the test register 5o3 and the correct entry in the Conversion register 5o5 is stored.

The comparator 502 checks the coincidence between the tentative incorrect entry in the test register and the actual erroneous information in the input register 501. If a coincidence is detected by the comparator 5o2 is, it generates a signal for the conversion register 505, which thereupon a transfers correct A or O information to the output register 5o6. The right A or O information is associated with the console identification code for the Information desk, as shown, combined and finally transmitted to the I'q register 319 and in this composed as a corrected query index I'q = NA'O or NAO ', with the apostrophe indicates that the term has been corrected.

In this way, the error correction circuits 317 and 318, respectively, serve in addition, the most likely errors in the requests made by the operator entered to be corrected, regardless of whether the error was made by the questioner or caused by the operator.

For example, it has been found through experience that the questioner The postal zone for a certain location is often incorrect if this location is in is close to an adjacent postal zone. The correction circuit 316 is used for this case is provided with a stored error correction table in order to make it frequent Correct mistakes. On the other hand, if an employment statement is frequent is confused with similar employment information, correction circuit 318 equipped with a fixed table for correcting errors in the profession be.

Reached the error correction table 5o4 at the end of a read cycle the test register 503, which is a counter, takes a maximum count value and provides one Impulse that clears the input register Sol and opens a gate that is connected to this connected is. Since the recovery facilities consisting of the second Ig register 316, error correction circuits 317 and 318 (Figure 3B) are not gated such as circuit 302, the error correction circuits must have 317 and 318 have their own inputs for the next incoming question in the A or O Xode open or close yourself. When the error correction circuits 317 and 318 are at their Infonotlon correction processes have ended and the A and O information in A 'and O'-information converted kind of the second Iq ## ister 316 an impulse to the Error correction circuits 317 and 318, which in turn provide the A 'or O'information to register 319 in the order NAO or NOA depending on the presence an end marker I or II transferred.

After the corrected A 'or O'information in the I' -q register 319 and there again with the original, unchanged N information is merged, the register 319 transmits the now corrected query index I 'to q the NAO register 273, where a new search cycle in the data memory 104 with the new query index I 'using the q method described above is initiated. If, due to the improved index I ', a response to the N1 -q test circuit 311 comes back a second signal from the N test circuit 311 forwarded to the 'no response' gate 312 if there is still no response from the data memory 1o4, which agrees with the request information I '. That 'no Answer 'gate 313 is prepared in advance by an input signal as above has been described. The second signal triggers gate 313 which is a command signal forwards via the buffer memory 111 to the printer 105, which at the desk 101 the Print out the 'no answer' message.

One-Coordinate Z Search Method Referring to Figure 13, the Function of the circuit with which the one-coordinate Z search method is carried out, described in detail. As mentioned above, in the event that that the address number of the core memory 2o1 is arranged serially, the following The core address number can be obtained by the formula Z = 22X 1+ y - 1 in FIG. A comparator 372, a register 371 and a core memory 375 correspond to the units 272, 271 and 2o1 in Figure 3A. The query index 1g enters the comparator 372, coming from the output of register 273 <Figure 3A) to the output of the core memory 375 to trigger.

A Z register 374 has the logical initial value 1 and a computer 376 doubles this initial value of Z (1x2 = 2) and another calculator 376 ' doubles the initial value of Z and adds a 1, so that 1 x 2+ is obtained 1 = 3. This result is entered in every invoice.

The content Z = 1 of the register 374 is forwarded to the control device 373, which determines the core address of the first memory output. The content of the memory 375, that is K <Z) and n corresponds to the coordinate Z = 1, which is stored in the register 371. The relationship between K (Z) and Z is as follows, with the number in brackets indicating the address corresponding to K (Z): In the table above, the keyword number is 20, 21, 22, 23 .... or 2m 1.

If the partial point comparison comprises m operations, then applies to the total number of keywords This means that the largest value of Z cannot exceed the total number of keywords SO. The content of the register 376 or 376 'lies between the values 2m and (2m + 1 - 1), so that the value exceeds SO = 2m - 1.

At this time, the Z check circuit 379 confirms that the value of Z exceeds the total number SO after the last comparison has ended and produces an output to an OR gate 378. The remaining parts of the device are the same as the corresponding circuitry relating to Figure 3A.

General remarks Experiments with the one described above Appendix have shown that this is three times more effective than known directory assistance services in which the person providing the information has the telephone subscriber numbers in one printed telephone directory searches. It is not with the search system described required that the information or operator give the full name of the searched Telephone subscriber enters, or only essential parts of this name; it is sufficient if the fragmentary information provided by the search index N1N2A1A20102 is given, which will take much less time to complete takes.

In a conventional directory assistance system in which the information manually searched in a telephone directory by the operator takes time It takes an average of 6o to 100 seconds to get a phone number from the information service to be obtained when the amount of telephone subscribers between 1.5 and 2 million lies. In the search system according to the invention, an answer is received almost immediately, d.

H. approximately within 0.6 seconds after the operator has finished entering the search code. In most cases the operator receives only a single answer to be forwarded to the inquirer, occasionally more as an answer, and almost never a large number of answers.

It can be seen that the invention is a fast automatic Telephone directory retrieval as a result of a novel and unique enables an empirical approach to information storage and retrieval, wherein the information is independent of one another according to a correlation of different ones Factors such as the main and secondary name, the main and secondary address and the main job, saved and retrieved later. Furthermore he follows access to the memory during the search operation is more random than in one specific order, so that the search time required to find a specific Memory address is reduced to a minimum.

The present invention concept was in connection with a Telephone directory assistance system described, but it can also be used in a control system Find application. For example, in the case of systems in use, the demand after a part, like a used car, not always with the registered part, this is e.g. a car in stock, match. In such a case it is conceivable that the inquirer has the desire to receive a part that corresponds to his / her ideas comes as close as possible. This is with the help of the present invention the main classification of each element in the index possible. This invention can can therefore also be used in the used car trade, for example Whenever a customer is supposed to be satisfied with a car, this is possible if possible corresponds to his request.

In the following example it is assumed that the desired part is characterized by the characteristics a, b and c, while the characteristics of the registered part, ie the one that is in stock, are a ', b' and c1. If the generic concept of a and a 'is A, that is, a and a' are identified under a common description and similarly the generic concept of b and b 'is given by B, etc., the relationship between the searched part and the registered part are represented as follows: Identifier Identifier Identifier Concept A Concept B Concept C If it is assumed that the features of the requested part are represented by the term a + b + c = A + B + C, in which the sign + indicates an 'overlap' or a cut, the features are of the registered part a '+ b' + c '= A + B + C in the genre classification correct. If A + B + C are taken as an index, a small number of similar parts can be retrieved. When a second scan is performed by 'a' or 'b' or 'c', the part closest to the desired part is output. For example, the output part can have the features a ', b' and c '. This was not possible with the previous search systems and accordingly the number of entries found was also very large in the prior art.

In this way, the invention can be used to advantage for finding parts in a short time, even when using inaccurate or incorrect input information.

a + b + c Appendix 1 The rule for N1 The following rules have been found at the choice of name codes found useful: 1. If the name of the telephone subscriber contains a place name and its own name, the own name is used for the Main code N1 preferred over the place name, as this only contains the address information repeated that is contained in the address code or even from the fact may result in a directory assistance center for a particular Area is selected, such as 'Miami Hilton Hotel'.

2. If the name contains a function word and its own name, the own name is preferred over the function word for the main name code N1, since the function word only repeats the information that comes with the employment code is detected. For example, if a request for the telephone number of the Hotel Hilton arrives, then the main name code N1 refers to the own name 'Hilton' and not the term 'hotel' that describes business operations.

3. If the name includes a place name and a common name that lacks distinctiveness, such as ~ standard 'or 'General', then the place name becomes the basis for the main name code N1.

4. If the name contains a place name and an activity word, for example ~ New York Trust CO. ', then the place name is used as the basis used for the main name code N1.

5. If the name consists of a place name that serves to a branch of a widespread company, be it public or private, to designate, e.g.

New York District Headquarters of the International Revenue Service, then it is likely that the New York Tower will simply go to the International Revenue Service will ask. Consequently, for creating the name cord it is more likely the distinctive part uses the 'International Revenue Service' as the geographical Part of 'New York District Headquarters'.

6. If the name has a number of expressions of foreign origin contains, e.g. 'Sukiyaki / tei', the name code N1 is based on the first Expression determined that means 'Sukiyaki' and the second part 'tea' is not taken into account, as the meaning of foreign expressions is not captured anyway and the easiest way is to use the beginning of the name.

7. If the name is a surname plus a place name plus a Name of the activity in any order (e.g. Miami Hilton Hotel), then the surname has the greatest additional information content and its meaning goes beyond the address and the job title. Therefore, the surname is used as the basis for the main name code N1.

8. If the name consists of a surname and a place name and a The name of the activity exists, but this is determined by the place name Area is so small that it corresponds to the second area code A2 closer than that Main area code A1, then there is a cross-reference in the manner that the name is filed twice. The name is first using in a memory area the surname as the main name code N1 and the place name as the second name code N2 is stored, and is then further stored in a further memory area inscribed, in which the place name is used for the main name code N1 and the surname as the second name code N2. In this way there is a cross-correlation or to a cross-correlation of the main code and the special name code.

9. Brackets () are used to indicate alternative codes. if an information or operator receives a request and does not make a decision which word of the name should be used for the main name code N1 they send a request in the form N1 (N1,), which means for the computer circuitry N1 or N1 '. Thus, N1 and N1 'are alternative possibilities that matter of the main name code, each of which will generate the desired response can.

lo. When a company has a number of subsidiaries each of which is known by its own name, then each of the subsidiaries both under their own name and under the name of the main company saved.

Appendix II Employment Code In the table below, a detailed exemplary embodiment for the coding of employment in a search system Listed: Code main code main sub-extract from the occupations class class a Agriculture and forestry, horticulture, fishing, animal breeding, etc.

a-1 Agriculture, forestry, animal husbandry, food a-2 Horticulture a-3 Fishing, fishing tackle e entertainment e-1 entertainment establishments e-2 Entertainment halls e-3 entertainment equipment m mining and metals m-1 mining m-2 metal processing m-3 metal goods manufacturing m- 4 metal processing m-5 metal factories 5 services s-1 restaurants, hotels s-2 hairdresser 5-3 cleanings, Garbage collection 5-4 advertisements, drafts 5-5 research s-6 consultations 5-7 administrations s-8 real estate t transport, news services, information t-1 transport t-2 newspapers t-3 intelligence services h hospitals, medical treatment, pharmaceuticals h-1 Hospitals, clinics h-2 Medical treatment h-3 Drugs, medical Article k Chemical companies and products k-1 chemical treatment, chemical Products k-2 chemicals, toiletries, fertilizers k-3 synthetic resins and fibers k-4 Oils and fats k-5 Gases, lacquers, paints n Machines, equipment and parts n-1 General machines n-2 Food industry machines n-3 Vehicles n-4 Electrical machines, equipments and tools n-5 precision machines, scientific Devices, optical machines and tools n-6 textile machines, mining equipment n-7 motors n-8 machine accessories n-9 machine parts c construction and civil engineering c-1 General construction c-2 Construction specialists, e.g. carpenters, etc.

c-3 factory buildings, etc.

f food f-1 cereals f-2 meat, eggs, dairy products f-3 Agricultural products, fish industry products. f-4 Food generally not in other classes) f-5 spices and ingredients f-6 cakes and F-7 Bread F-8 luxury food industry, ice cream makers and traders p paper, publishing p-1 paper and pulp p-2 paper products p-3 publishing x textiles, fibers and products made from them x-1 fiber producers x-2 fiber products x-3 Clothing 1Light and Fuel 1-1 Electricity and Gasworks 1-2 Others Fuels g rubber and leather g-1 leather products g-2 rubber products r religion, Culture, education r-1 schools, cultural institutions r-2 religious organizations, Funeral home r-3 clubs i Finance, banks, pawn shops, Insurance i-1 ~ Finance, Banks, Pawn Shops i-2 Insurance, Securities d department stores, commerce, works of art, items d-1 department stores, trading companies d-2 Jewels, works of art, jewelry d-3 Handicrafts and toys, tools d-4 Haberdashery, articles, accessories (buttons, buckles, zippers, hooks and eyes) d-5 Second hand goods dealers w Wood and bamboo products, furniture, ornaments w-1 Wood and wood products w-2 bamboo and cane w-3 furniture, ornaments, fittings q stones and ceramics q-1 ceramics q-2 stone products and building materials b graphic arts b-1 prints b-2 publishing, books b-3 photography The fine division for the right column, for example subclass d-2 (jewels, works of art, jewelry) then reads: handicrafts (including 'rarities') pictures and paintings (including 'Art Galleries') Precious Metals (including 'Jewels') Ivory Beaded Jewelry alarm clock and watches and glasses-driven jewelry.

It goes without saying that for each subclass a certain There is a fine division.

As can be seen from the table above, the letter code for the main division, e.g. 'b' for the graphic trade as the main feature of oil is used for employment and the numerical code for the subclass is used second Feature O2 of employment is used. In the case of a photographer, for example, the following applies ° 1 ° 2 = b3.

Blank page

Claims (6)

Claims 1. Automatic Telefonauskunftsanige with at least an input device, which is operated by an operator, to a query information including a format control pulse for a query index with at least one output device that provides the information in response to the queries supplies, with at least one data memory with direct access to the memory data inputs and with at least one central computing and control unit, in particular for comparison operations, d u r c h e k e n n n z e i c h n e t that the central computing and control unit (103) includes an index recognition circle (FIG. 4A>, which is connected to the additional input device (1o6) is connected and detectors (331, 332, 340, 342) for detecting the delimiter between the individual data information in the request and each of these Feeds data information into corresponding registers (336, 337, 338, 339) which are connected to the detectors (331, 332, 340, 342) via switches (335, 341), and store the data information and the number of characters of each data information limit to a predetermined number that converters (305, 307, see Figures 5, 6 or 317, 318, see Figure 12) with the detectors (331, 332, 340, 342) in connection stand to convert each of the data information into a corresponding main code, that a register <32o) with the converters (305, 307 or 317, 318) for storage each main code is connected, that with this register (32o) further registers (304, 306, 308, 322) and an integration register (31o) are coupled to the data information to summarize to a set of query information obtained from the query index, a desk and a request number and a format control symbol exist that a Control unit (3o9> with the register and the other registers (304, 306, 308, 322), which are parallel to a circuit (3o2), is connected and the format control symbol in the request information for controlling the transmission sequence the Recognizes main codes from the other registers to the integration register (31o), that a separator with the integration register for separating the query index of the set of query data information that is coupled to the separator and a combining device is connected to the output channel of the data memory that receives the set of request data information and with the response data of the data memory {wo4), which are output in the form of key words xKy are and compared in a comparison device (272) with the query index 1q Will. Continuation with claim 2 2. Automatic directory inquiries according to claim 1, characterized in that the comparison device (272) has a Access memory with memory cells for storing the keywords xKyt in are prepared in the same format as the query index, and a control unit comprises that a further access memory (2o1) for storing the area unit numbers is provided that correspond to the keywords, the access memory Contains memory cells that correspond to the corresponding memory cells in the access memory the comparison device (272) is assigned that a comparison device with the two access memories for determining the query index from a number of keywords stored in the access memory and the memory areas of the data memory (104) correspond to that circuits (206, 206 ') with the comparison device (272) and a y-register (22ob of the direct Access memory (2o1) are connected to the memory address of the next keyword xKy in accordance with a last comparison result determine, that indicates whether the keyword is larger or smaller than the query index 1g, that a keyword storage means (271), which between the direct access memory (2o1) and the comparison device (272) is arranged and connected to these, receives and stores the keyword xKy for the next comparison step, that with this keyword memory device an area number memory device is connected, which is also connected to the direct access memory (2o1) and receives and stores the area unit numbers corresponding to the keywords and their content via an adder (212) during the determination the position of the query index under the keywords, and that testers (311, 312) between an adder (212) and the fast output device (107l are turned on to see if one or none of the from the data store (1o4) received responses with the query index at least in certain data information and then the coincident data to the fast output device (1o7) or a signal 'no answer' via a register (313) to the fast output device (1o7) to be transmitted if no matching data is available. 3. Automatic directory assistance system according to claim 1, characterized in that that the comparison means (272) includes circuitry to determine whether the Inquiry index 1q is greater, less than or equal to each keyword and for transmission the corresponding results to addressing devices that use an x-coordinate counter circuit (209) between an x-address driver stage (2o3) of the direct access memory (2o1) and the comparison device (272) is connected, which is used for each comparison inputs a 1 into the counting circuit (2o9) and the value of x in an x register (222) a 1 adds that a y-register (22o) with a y-address driver stage (2o2> for direct access memory (2o1) connected to at each comparison step to register a calculated value that a y-coordinate calculating circuit (206 ') is connected to the y-register (220) to find its newly stored y-value multiply by 2 and re-enter it in the y-register (220) as soon as the Inquiry index 1q is greater than the keyword xKy that another y-coordinate computer circuit (2o6) is provided to the y-register (22o) connected and its content is multiplied by 2 and reduced by 1, back into the y-register (22o) if the query index 1q is smaller than the keyword xKy, that circuits are included in the comparison device (272), which in each Compare step add 1 to the keyword x Ç and check whether the summed up Value has reached a predetermined value and to send an output signal to a gate (2o9 ') to be transmitted when the totalized value reaches the specified value has the direct access memory (201) based on the keywords x Ky which stores x and y coordinates corresponding to the area units of the data memory (1o4), the register for the area units at each comparison step stores the area number, which is determined according to the formula t (2y-1) / 2x, if t is the number of all area units in the data memory (104) that the OR gate (2o9 ') is connected to the test circuits of the comparison device <272) and a stop signal to the comparison device (272) for the comparison process, a reset signal for the x and y registers (222 and 22o, respectively) and the counting circuit (2o9) and a command signal to the area number register (271) supplies the stored Area number to transmit to the data memory (1o4) that an AND gate (21o) with the test circuits the comparison device (272) and a signal to an adding device (212) transmits, if the query index 1q is greater than the keyword K, that between xy the adding circuit (212) and the AND gate (210) a circuit (211) is interposed, which is a signal of the value 'one' in the adder (211) to form the value n which is the next recording track number indicates in the data memory (1o4), and that between the adding circuit (211) and the Data memory (1o4) a multiplier circuit (216) for determining the upper and the lower level number is switched in the recording track. Continuation with claim 4 4. Automatic directory assistance system according to claim 1, characterized in that a z-register (374) with a control device (373) of a core memory (375) is connected to a calculated value at each Compare step to store that a z coordinate calculator (376) with the z register (374) and with a comparator (372) in connection and the content of the z-register (374) doubled each time the content is changed, stored and in the z-register (374) new if the initial index 1q is less than the keyword Kz, that another z-coordinate computer (376 #) with the z-register (374) and with the comparator (372) is coupled to multiply the contents of the z-register (374) by two and to increment 1 each time the content changes, store this calculated value and to be re-entered into the z-register (374) if the query index 1q is greater than the keyword Kz is that an OR gate (378) with the comparator t372) and a test circuit (379) is connected that an AND gate (38c) to the test circuit (379) is connected that the test circuit (379) signal to the AND gate <380) when the z coordinate has a specified value achieved, and that the core memory (375) is designed as a direct access memory and has storage units in which the keywords Kz in a sequence of the z coordinate value is stored in accordance with the relationship z = 2x 1 + y - 1 are. Continuation with claim 5 Automatic directory assistance system according to claim 1, characterized in that the device for finer subdivision in the data memory t1c4) an isolating circuit for isolating certain data information from the request record and the conversion of this data information into a serial number, which is transmitted contains that a first detection circuit (213) with the integration register (31o) is coupled and connected in parallel with the isolating circuit to the symbol to recognize the boundary between the upper and lower value of the data information indicates, and to generate a control signal to a switch (215), that a multiplier circuit (216) in a first multiplier unit the order number with a specific one Integer multiplied by the mean of the number of entries in a Area unit corresponds, and that a second multiplier unit corresponds to the ordinal number reduced by 1 and the obtained value multiplied by the whole number and that Product increased by 1 that the switch (215) with the isolating circuit in the form of a Register t273) and with the multiplier circuit (216) as well as with the first Detection circuit (213) is connected, the output signal of which controls the switch (215), both of which Outputs with the two multiplier units in the multiplier circuit (216) are connected, in the case that the first detection circuit (213) the boundary symbol in the specific data information does not recognize this of the first multiplier unit in the multiplier circuit (216) and in the event that the detection circuit (213) determines the limit symbol in the data information, the lower limit value the data information of the second multiplier unit in the multiplier circuit (216) supplies that a second detection circuit (217) with the data memory (1o4) is coupled to an overflow symbol in the response data of the data memory (1o4) and to set a range unit address for this overflow symbol in the Enter data memory (1o4), with the help of which the remaining entries were found can be, and that a third detection circuit (218) with the register (32o) is coupled to determine whether the first data of the response contain a symbol, which indicates an overflow from another area unit of the data memory (1o4) and generates a 'no response' signal upon the occurrence of such a symbol. 6. Automatic directory assistance system according to claim 5, characterized in that that control registers are connected to the test device and its set of query data save as well as separate the data information from each other in order to from each data information to recognize the format control signal and a signal for the newly formed order according to the format control signal that is sent to the control register converter that receive the data information and that of each data information determine the closest concept and match the converted codes with the newly formed sequence create that one integration register with the the control register and converter register are coupled to the dividing device is connected to a query index by summarizing the converted codes for each query data information, this query index for search purposes the dividing device is fed that a 'no answer' circuit a Has output channel which is connected to the output device and another Output channel 'connected to the index forming is connected and input channels connected to the third detection circuit and the test device includes to receive and store a 'no response' signal and at least one of the third detection circuits and the test device, whereby the feed a signal 'no entry' to the output device is prepared that furthermore a signal for triggering the index converter and for the second search can be fed back is, and that the 'no answer' circuit by the second signal 'no answer' the second search is triggered and a 'no entry' signal is sent to the output device enters following the examination by the index converter.