JP4076533B2 - Information conversion apparatus and program - Google Patents

Description

  The present invention relates to an information conversion apparatus and program, and in particular, every time a character string composed of a plurality of words combined to represent a predetermined object is input, the input character string is input to the program. The present invention relates to an information conversion device for converting into identification information given to a predetermined object presumed to be represented by a character string, and an information conversion program for causing a computer to function as the information conversion device.

  When a financial transaction such as remittance to an account of a different financial institution is requested, a message in which the name of the customer related to the financial transaction (for example, the name of the holder of the remittance account) is set as a character string is financial Messages are sent and received between institutions. On the other hand, a financial institution assigns a customer code (this customer code is called, for example, a CIF (Customer Information File)) to each customer, and the computer system in the financial institution is the customer code. Since it is configured to identify customers based on codes and execute designated financial transactions, when receiving a message from another financial institution, a character string representing the customer name set in the received message Needs to be converted to the corresponding customer code.

  In relation to the above, Patent Document 1 discloses that a customer code table in which a combination of a plurality of words specifying a customer code and a corresponding customer code is registered in a storage device in advance and converted from input data. There is disclosed a technique for extracting a plurality of words, comparing the extracted plurality of words with a plurality of words registered in a customer code table, and converting them into a customer code if they match.

Patent Document 2 also includes a basic word dictionary composed of words, a nuclear name dictionary composed of a set of two words, and a full name dictionary composed of a set of a symbol and another word and target data. The memory address is symbolized on the main memory as a symbol, the address corresponding to the word contained in the converted lexical phrase is obtained by referring to the symbolized basic word dictionary, and the pair in which two addresses match is symbolized A technique for converting a converted lexical into target data by finding from a nuclear name dictionary and then finding a pair that matches another address from a symbolized full name dictionary is disclosed.
JP 2002-56005 A JP 2004-31873 A

  However, for the customer name set in the above message, the name specified by the client who requested the execution of the financial transaction is set as it is as a character string, so an extra word is added to the regular name, for example. In some cases, the proper name is not set correctly, for example, some of a plurality of words constituting the canonical name are missing. For this reason, conventionally, even when the character string representing the customer name set in the received message is a character string with an extra word added to the legitimate name, conversion to the correct customer code In order to enable this, a character string representing an authorized customer name and a customer code are registered in the name dictionary in association with each other, and a character string (input character string) representing a customer name set in the received message is obtained. Each character string registered in the name dictionary (character string representing a regular customer name) is collated to search for a character string in which each constituent word is present in the input character string. A customer code associated with a character string is determined to be a customer code corresponding to an input character string.

  However, even if a search is performed under the above conditions, for example, for the input character string `` HAYASHI ELECTRIC CO., LTD SHINZYUKU '', the character string `` HAYASHI ELECTRIC CO., LTD '' and the character string `` SHINZYUKU ELECTRIC CO. , LTD "may be extracted, and a plurality of character strings representing customer names may be extracted by the search. In such a case, it is difficult to automatically determine which of the plurality of customer names from which the input character string is extracted, and conventionally there is no solution other than leaving it to the operator's decision. The actual situation is that a great burden is placed on the operator.

  The present invention has been made in consideration of the above facts, and a predetermined object presumed to represent a character string composed of a plurality of words combined to represent the predetermined object. It is an object to obtain an information conversion apparatus and an information conversion program that can improve the probability that conversion to identification information assigned to the item can be automatically performed without bothering an operator.

  A character string that is set for the purpose of representing a specific target (for example, a specific customer) but is different from the regular name of the specific target is an extra word for the regular name as described above. The extra word added to this type of string, for example, “HAYASHI ELECTRIC CO., LTD” has an extra word “SHINZYUKU” added to it. In many cases, it is a fixed word for each individual object. In the above example, if an extra word “HAYASHI” is not added to the character string representing the customer name “SHINZYUKU ELECTRIC CO., LTD”, the input character string is the customer name “HAYASHI ELECTRIC CO. ., LTD "can be regarded as a character string. The inventor of the present application pays attention to the above facts, and identifies a character string formed by combining a plurality of words to represent a predetermined object, given to the predetermined object estimated to be represented by the character string. When converting to information, if an extra word (unnecessary word) added to a character string that has been converted to identification information in the past is stored and used to convert a new character string, the character string The present invention has been made on the assumption that it is possible to improve the probability of appropriate automatic conversion to identification information.

  Based on the above, the information conversion apparatus according to the first aspect of the present invention provides an input character string each time a character string configured by combining a plurality of words to represent a predetermined object is input. An information conversion apparatus for converting into identification information given to the predetermined object presumed to be represented by the inputted character string, and for each object to which identification information is given, A first storage means for storing a name dictionary registered by associating a combination of words constituting a character string representing each identification object and a character string representing each object, and a specific object inputted in the past And a combination of words obtained by adding a single or a plurality of specific words to a combination of words registered in the name dictionary in association with identification information given to a specific object Said single or multiple in the string A second storage means for storing an unnecessary word dictionary registered as an unnecessary word of the specific target in association with identification information given to the specific target, and an input character string The combination of words to be searched is compared with the combination of words registered in the name dictionary, thereby searching for a combination of words in which each constituting word is present in the input character string. Based on a search means and a search result obtained by the search means, a target estimated to be represented by the input character string is determined, and the identification information given to the determined target is determined based on the input character string And a combination of words in which a single or plural first specific words are excluded from a plurality of words constituting the inputted character string in a combination of words extracted by a search by the search means And a word excluding the first specific word when a combination of words excluding a single or plural second specific words from a plurality of words constituting the input character string is included. Whether the first identification information registered in the name dictionary in association with the combination is registered in the unnecessary word dictionary in association with at least one of the single or plural first specific words. The second identification information registered in the name dictionary in association with a combination of words excluding the second specific word is searched for the single or plural second specific words. It is searched whether or not it is associated with at least one and registered in the unnecessary word dictionary, and only one of the first and second identification information is at least one of the single or plural first specific words, Or the single or plural When the unnecessary character dictionary is associated with at least one second specific word and registered in the unnecessary word dictionary, the input character string is registered in the unnecessary word dictionary out of the first and second identification information. Conversion means for determining that the object corresponding to the identification information is represented.

  The information conversion device according to the first aspect of the present invention provides an input character string that is input to a character string that is formed by combining a plurality of words to represent a predetermined object. It converts into the identification information given to the predetermined object estimated that the column represents. As the above character string, for example, a character string representing the name of a customer set in a message sent and received between financial institutions is suitable, but is not limited to this, and a plurality of characters are represented to represent a predetermined target. The character string may be a combination of words. In the invention described in claim 1, each individual object to which identification information is assigned is associated with a combination of identification information given to the individual object and a word constituting a character string representing the individual object. The registered name dictionary is stored in the first storage means. In addition, the search means compares each combination of words constituting the input character string with each combination of words registered in the name dictionary, so that each of the constituent words is included in the input character string. Search for a combination of existing words, and the conversion means determines an object that is assumed to be represented by the input character string based on the search result by the search means, and determines the input character string. It is converted into the identification information given to the target.

  In the first aspect of the present invention, a combination of words registered in the name dictionary associated with the identification information given in the past and representing the specific target and associated with the specific target is registered. On the other hand, the single or plural specific words in the character string composed of a combination of words to which the single or plural specific words are added are associated with the identification information given to the specific target, and the specific target An unnecessary word dictionary registered as an unnecessary word is stored in the second storage means. In addition, the conversion unit includes a combination of words obtained by excluding a single or a plurality of first specific words from a plurality of words constituting the input character string in a combination of words extracted by the search by the search unit. A combination of words excluding the first specific word when a combination of words excluding the single or plural second specific words from the plurality of words constituting the input character string is included. While searching whether or not the first identification information associated and registered in the name dictionary is associated with at least one of the single or plural first specific words and registered in the unnecessary word dictionary, The second identification information registered in the name dictionary in association with the combination of words excluding the second specific word is associated with at least one of the single or plural second specific words, and the unnecessary word dictionary Registered in Search whether or not. The conversion means is unnecessary because only one of the first and second identification information is associated with at least one of the single or plural first specific words, or at least one of the single or plural second specific words. If it is registered in the word dictionary, it is determined that the input character string represents a target corresponding to the identification information registered in the unnecessary word dictionary among the first and second identification information.

  As described above, in the invention described in claim 1, a word registered in the unnecessary word dictionary as an unnecessary word for a specific target is used for determining whether or not the input character string represents a specific target. For example, for the input character string “HAYASHI ELECTRIC CO., LTD SHINZYUKU”, a combination of words from the name dictionary (character string) “HAYASHI ELECTRIC CO., LTD” and “SHINZYUKU ELECTRIC CO., LTD” by searching by the search means ”Is extracted, the specific word“ SHINZYUKU ”is registered in the unnecessary word dictionary as the target unnecessary word corresponding to the word combination“ HAYASHI ELECTRIC CO., LTD ”, and the specific word“ HAYASHI ” Is not registered in the unnecessary word dictionary as an unnecessary word corresponding to the word combination “SHINZYUKU ELECTRIC CO., LTD”, the input string corresponds to the word combination “HAYASHI ELECTRIC CO., LTD”. Automatically determines that the object is represented It is, will be automatically converted to the corresponding identification information. Therefore, according to the first aspect of the present invention, a character string formed by combining a plurality of words to represent a predetermined object is given to the predetermined object estimated to be represented by the character string. It is possible to improve the probability that conversion to identification information can be automatically performed without bothering the operator.

  In the first aspect of the invention, whether or not the first identification information is registered in the unnecessary word dictionary in association with at least one of the first specific word or the plurality of first specific words is searched. As a result of searching whether or not the identification information is associated with at least one of the single or plural second specific words and registered in the unnecessary word dictionary, the first and second identification information is single or plural There may be a case in which at least one of the first specific words or at least one of the single or plural second specific words is registered in the unnecessary word dictionary. In consideration of such a case, in the invention according to claim 1, the conversion means may convert the first specific word or the second specific word out of the first and second identification information as described in claim 2, for example. Among the included words, words that are associated with the first and second identification information and registered in the unnecessary word dictionary (that is, unnecessary as unnecessary words in the first identification information and unnecessary identification words in the second identification information) A word other than a word other than a word registered in the word dictionary (that is, it is included in the input character string and is registered in the unnecessary word dictionary as an unnecessary word of either the first identification information or the second identification information) It is preferable that the identification information associated with the word) is determined and it is determined that the input character string represents an object corresponding to the determined identification information.

  In the invention described in claim 2, the first and second identification information is unnecessary to be associated with at least one of the single or plural first specific words or at least one of the single or plural second specific words. When it is registered in the word dictionary, it is included in the input character string as the first specific word or the second specific word, and is an unnecessary word dictionary as an unnecessary word of either the first identification information or the second identification information. Since the target represented by the input character string is determined based on the words registered in, it is possible to automatically determine the more probable target that the input character string is supposed to represent. It is possible to further improve the probability of automatically converting a column into identification information without bothering the operator.

  Further, in the invention according to claim 1, the conversion means, as described in claim 3, for example, includes the unnecessary word dictionary in which the first identification information is associated with at least one of the single or plural first specific words. As a result of searching whether or not the second identification information is registered in the unnecessary word dictionary in association with at least one of the single or plural second specific words, The first and second identification information is registered in the unnecessary word dictionary in association with at least one of the single or plural first specific words, or at least one of the single or plural second specific words. In this case, among the first and second identification information, the identification is included in the input character string as the first specific word or the second specific word and is registered in the unnecessary word dictionary in association with a larger number of words. Determine and enter information May be configured to string is judged to represent the object corresponding to the determined identification information.

  In the invention of claim 3, the first and second identification information is not required because it is associated with at least one of the single or plural first specific words or at least one of the single or plural second specific words. Based on the number of words registered as unnecessary words in the unnecessary word dictionary among the words included in the input character string as the first specific word or the second specific word when registered in the word dictionary, Since the target represented by the input character string is determined, it is possible to automatically determine a more probable target that is estimated to be represented by the input character string. It is possible to further improve the probability of automatically converting a column into identification information without bothering the operator.

  In the invention described in claim 1, for example, as described in claim 4, in the unnecessary word dictionary stored in the second storage unit, the character string represented in the character string input in the past is specified. Frequency information indicating the appearance frequency of the unnecessary word of the target is registered for each unnecessary word registered in the unnecessary word dictionary, and the conversion means inputs the input characters based on the search result by the search means Each time an object that is estimated to be represented by a column is determined, the input character string is registered in the unnecessary word dictionary based on whether or not the input character string includes a specific target unnecessary word represented by the character string. Update the frequency information of unnecessary words of a specific target, and, among the combinations of words extracted by the search by the search means, single or multiple first words from a plurality of words constituting the input character string 1 specific unit And a combination of words excluding single or plural second specific words from a plurality of words constituting the input character string, and the first identification information is It is searched whether or not it is associated with at least one of the single or plural first specific words and registered in the unnecessary word dictionary, and the second identification information is at least one of the single or plural second specific words. As a result of searching whether or not it is registered in the unnecessary word dictionary in association with one, the first and second identification information is at least one of a single or a plurality of first specific words, or a single or a plurality of first When it is associated with at least one of the two specific words and registered in the unnecessary word dictionary, it is included in the input character string as the first specific word or the second specific word, and the first and second identification information Are associated with each Based on the individual frequency information of unnecessary words registered in the word dictionary, it may be determined on the object where the input character string represents.

  In the invention of claim 4, the first and second identification information is unnecessary to be associated with at least one of the single or plural first specific words, or at least one of the single or plural second specific words. If it is registered in the word dictionary, it is included in the input character string as the first specific word or the second specific word, and is registered in the unnecessary word dictionary in association with the first and second identification information. Since the target represented by the input character string is determined based on the appearance frequency of each unnecessary word, it is possible to automatically determine a more probable target that the input character string is supposed to represent. As in the inventions of claims 2 and 3, it is possible to further improve the probability that the input character string can be automatically converted into the identification information without bothering the operator.

  In the first aspect of the invention, whether or not the first identification information is registered in the unnecessary word dictionary in association with at least one of the first specific word or the plurality of first specific words is searched. As a result of searching whether the identification information is associated with at least one of the single or plural second specific words and registered in the unnecessary word dictionary, both the first and second identification information are the first specific If the word or the second specific word is not registered in the unnecessary word dictionary, the conversion unit displays the input character string on the display device as described in claim 5, for example. It can be configured such that the determination of the object represented by the input character string is left to the operator, and the object represented by the input character string is determined based on the determination result input by the operator via the input device.

  Further, in the invention described in claim 1, for example, as described in claim 6, a single combination of a plurality of words constituting an input character string is included in a combination of words extracted by a search by a search means. Or a combination of words excluding a plurality of specific words is included, and the character string input by the conversion means represents a specific object corresponding to the combination of words excluding the specific words When it is determined, it is preferable to provide an unnecessary word registering unit that associates the specific word with the identification information given to the specific target and registers the specific word as an unnecessary word of the specific target in the unnecessary word dictionary. Thereby, unnecessary words are automatically registered in the unnecessary word dictionary by the unnecessary word registering means, and the operator does not need to perform complicated operations such as unnecessary word registration.

  Further, in the invention according to any one of claims 1 to 6, for example, as described in claim 7, character strings corresponding to individual words constituting the input character string and given to individual words A third storage means for storing a basic word dictionary registered in advance with a corresponding logical index, and the basic word using character strings corresponding to individual words constituting the input character string as a key. A word conversion means for converting the input character string into a logical index in units of individual words constituting the character string by performing a dictionary search, and the name dictionary includes a combination of words The logical index combination is registered as an unnecessary word dictionary, and the logical index is registered as an unnecessary word in the unnecessary word dictionary, and the search means and the conversion means are input character strings as input character strings. It is preferably configured to perform the processing by using a combination of logical indexes obtained by being converted word by word by word conversion means.

  In general, text data representing a character string has a larger amount of data than the amount of information (for example, text data represented using a character code that represents one character as data of 2 bytes is the number of characters × 2 bytes of data). If the character string is processed as text data, the load applied to the information conversion apparatus according to the present invention and the processing delay are increased, and the storage capacity necessary for storing the name dictionary and the unnecessary word dictionary also increases. On the other hand, in the invention according to claim 7, the character strings corresponding to the individual words constituting the input character string and the logical index (word code information) assigned to the individual words are associated with each other in advance. Each registered basic word dictionary is stored in the third storage means, and the word conversion means searches the basic word dictionary by using the character strings corresponding to the individual words constituting the input character strings as keys. By doing so, the input character string is converted into a logical index in units of individual words constituting the character string, so that the data amount of the input character string (character string to be converted) can be reduced. it can. Also, by registering a combination of logical indexes as a combination of words in the name dictionary and registering a logical index as an unnecessary word in the unnecessary word dictionary, a storage capacity necessary for storing these dictionaries Can also be reduced. Then, the search means and the conversion means perform each process using a combination of logical indexes obtained by converting the input character string in word units by the word conversion means as the input character string. The load applied to the information conversion apparatus according to the present invention can be reduced and the processing speed can be increased.

  The information conversion program according to the invention described in claim 8 is an information conversion program, wherein each time a character string composed of a plurality of words combined to represent a predetermined object is input, the input character string is An information conversion program for causing a computer to function as an information conversion device for converting into identification information assigned to the predetermined object presumed to be represented by the input character string, wherein the computer is provided with identification information. A first storage means for storing each registered name dictionary in association with identification information given to each object and a combination of words constituting a character string representing each object And a single or a plurality of special features for a combination of words registered in the name dictionary associated with identification information given in the past and representing a specific target and assigned to the specific target. The single or plural specific words in the character string composed of a combination of words with added words are associated with the identification information given to the specific target and registered as unnecessary words of the specific target Second storage means for storing an unnecessary word dictionary, and the computer compares each combination of words constituting the input character string with each combination of words registered in the name dictionary. And a search means for searching for a combination of words in which each of the constituent words is present in the input character string, and the input character string is represented based on a search result by the search means. A combination of words extracted by a search by the search means, and converting the input character string into identification information given to the determined object If a combination of words excluding a single or a plurality of specific words from a plurality of words constituting the input character string is included, the single or a plurality of specific words are excluded. Whether or not the specific identification information registered in the name dictionary in association with the combination of the registered words is registered in the unnecessary word dictionary in association with at least one of the single or plural specific words When the specific identification information is associated with at least one of the single or plural specific words and registered in the unnecessary word dictionary, the input character string is It is characterized by functioning as a conversion means for judging that it represents an object to which identification information is given.

  An information conversion program according to an eighth aspect of the invention is a program for causing a computer including the first storage unit and the second storage unit to function as the search unit and the conversion unit. By executing the information conversion program according to the eighth aspect of the invention, the computer functions as the information conversion apparatus according to the first aspect. Converting a character string formed by combining a plurality of words to represent into identification information given to a predetermined object presumed to be represented by the character string, without bothering the operator The probability of being automatically performed can be improved.

  As described above, according to the present invention, for each object to which identification information is assigned, a combination of identification information given to each object and a combination of words constituting a character string representing each object is associated with a name dictionary. Each of the combinations of words registered in the name dictionary associated with the identification information given in the past and representing the specific target and associated with the specific target. One or more specific words in a character string composed of a combination of words to which one or more specific words are added are associated with identification information given to a specific target, and unnecessary words are used as unnecessary words of a specific target By registering in the dictionary and comparing each combination of words constituting the input character string with each combination of words registered in the name dictionary, each of the constituent words is included in the input character string. Each present A combination of words obtained by excluding a single or a plurality of first specific words from a plurality of words constituting an input character string in a combination of words extracted by the search. And a combination of words excluding the first specific word when a combination of words excluding a single or plural second specific words from a plurality of words constituting the input character string is included. And searching whether or not the first identification information registered in the name dictionary and associated with is registered in the unnecessary word dictionary in association with at least one of the single or plural first specific words. The second identification information registered in the name dictionary in association with the combination of words excluding the second specific word is associated with at least one of the single or plural second specific words and is an unnecessary word Registered in the dictionary Whether or not only one of the first and second identification information is associated with at least one of the single or plural first specific words, or at least one of the single or plural second specific words If it is registered in the unnecessary word dictionary, it is determined that the input character string represents a target corresponding to the identification information registered in the unnecessary word dictionary among the first and second identification information. Therefore, converting a character string formed by combining a plurality of words to represent a predetermined object into identification information given to the predetermined object estimated to be represented by the character string. In addition, it has an excellent effect that it is possible to improve the probability of being automatically performed without bothering the operator.

  Hereinafter, an example of an embodiment of the present invention will be described in detail with reference to the drawings. FIG. 1 shows a computer system 10 according to the present embodiment. The computer system 10 includes a computer 12 installed in an information center of a specific financial institution. The computer 12 is composed of a mainframe computer and includes a CPU 12A, a ROM 12B, a RAM 12C, and an input / output port 12D, which are connected to each other via a bus 12E such as an address bus, a data bus, and a control bus. A communication control device 14, a display 16, a mouse 18, a keyboard 20, and an HDD (hard disk drive) 22 are connected to the input / output port 12D as various input / output devices.

  The HDD 22 stores a basic word dictionary, a name dictionary, and an unnecessary word dictionary (details will be described later), and the HDD 22 corresponds to the first storage means and the second storage means according to the present invention. This also corresponds to the third storage means described. Further, in the computer 12, a customer code conversion program for performing a customer code conversion process described later is installed in the HDD 22 in advance. This customer code conversion program corresponds to the information conversion program according to the eighth aspect, and the computer 12 functions as an information conversion apparatus according to the present invention by the CPU 12A of the computer 12 executing the above programs. The computer functioning as the information conversion apparatus according to the present invention is not limited to the mainframe computer, and any computer such as a personal computer or a microprocessor can be applied.

  The communication control device 14 of the computer 12 is connected to a computer network 26 built in a specific financial institution. The computer network 26 is configured by connecting a large number of computers to each other via a communication line. In addition, among a large number of computers constituting the computer network 26, there are included computers for performing financial transactions instructed from the computer 12 (for example, depositing into an instructed account). Further, the communication control device 14 of the computer 12 is connected to a computer 30 of another overseas financial institution via a network 28 (international network for message exchange between banks). Hereinafter, the specific financial institution in which the computer 12 is installed is referred to as a first financial institution, and the financial institution in which the computer 30 is installed is referred to as a second financial institution.

  Next, the operation of this embodiment will be described. When requesting a foreign remittance to a financial institution, the remittance requester specifies information such as the name of the financial institution, branch name, account holder name, account number, and amount of remittance of the remittance destination account. In the second financial institution, when remittance is requested from the remittance requester to the specific account established in the first financial institution, a message is generated to request the first financial institution to deposit money in the specific account. However, each piece of information designated by the remittance requester is set in this electronic message, and information such as the name of the holder of the remittance destination account is set in the electronic message as a character string (text data). When the creation of the message is completed, the second financial institution transmits the created message to the computer 12 of the first financial institution via the network 28 by the computer 30. When the computer 12 of the first financial institution receives a message from another financial institution via the network 28, the name of the name of the remittance destination account (name of the customer of the first financial institution) set in the received message. Processing for extracting text data of a character string to be represented and converting the text data of the extracted character string into a customer code (identification information) assigned to each customer by the first financial institution to identify each customer (Customer code conversion processing described later) is performed. The basic word dictionary, the name dictionary, and the unnecessary word dictionary stored in the HDD 22 of the computer 12 are provided to realize the customer code conversion process.

  That is, the character string representing the customer name is formed by combining a plurality of words. However, in the first financial institution, a multi-digit numerical value is added to each word constituting the character string representing each customer name. In the customer code conversion process, a character string to be converted representing a customer name extracted from the received message is assigned to each word constituting the character string. Processing is performed after conversion to a combination (logical index set) of corresponding logical indexes. For this reason, as shown in FIG. 2A as an example, the basic word dictionary is associated with text data representing individual words and numerical data corresponding to the logical index assigned to each word. Each information is registered (in FIG. 2A, information such as part of speech, attribute (name) and attribute (code) is also added to each word). In addition, as shown in FIG. 2B as an example, the name dictionary described above corresponds to the names of individual customers, a logical index set corresponding to each customer name, and a customer code assigned to each customer. Each is attached and registered.

  By the way, among the names of the individual customers, a combination of a specific word added to a combination of words constituting the character string representing the name of the first customer is a character string representing the name of the second customer. There is a name having a relationship (this relationship is called an inclusion relationship) that matches the combination of the words that constitute. As an example, as shown in FIG. 2B, the character string “HAYASHI CO, .LTD” representing the name of the customer B is composed of two words “HAYASHI” and “CO, .LTD” (FIG. 2). If the logical index shown in FIG. 2 (B) is used, the logical index set of the above character string is “105,107”), but the combination of words obtained by adding the specific word “ELECTRIC” to this combination of words is the customer C Matches the combination of words constituting “HAYASHI ELECTRIC CO, .LTD” (the logical index set of this character string is “105,106,107”). When the character string to be converted is a character string corresponding to a customer name having an inclusion relationship, it uniquely identifies which of the plurality of customer names having an inclusion relationship the character string to be converted represents There is a problem that it cannot be judged. For example, if the character string to be converted is “HAYASHI CO, .LTD”, the character string to be converted is likely to represent the name of the customer B described above, but the character string to be converted is the customer C's name. The specific word “ELECTRIC” in the character string “HAYASHI ELECTRIC CO, .LTD” (hereinafter referred to as non-inclusive word) is missing for some reason, and the character string to be converted is the name of the customer C We cannot deny the possibility that it represents.

  Therefore, in the customer code conversion process, it is necessary to perform different processing depending on whether the character string to be converted represents a name having an inclusion relationship or a name having an inclusion relationship. For this reason, out of the customer information registered in the name dictionary, the customer information whose name has an inclusion relationship has an inclusion relationship as shown in FIG. 2B as an example. An inclusion code (inclusion identification information) indicating that the Specifically, the same inclusion code is added to the information of a plurality of customers whose names have an inclusion relationship (the first customer and the second customer described above), and the same inclusion code is added. Among the customers, the logical index of the non-inclusive word is also added to the inclusion code of the customer (second customer) in which the non-inclusive word is added to the name (in FIG. 2B, the name is “HAYASHI ELECTRIC In this example, the logical index “106” of the non-inclusive word “ELECTRIC” is added to the inclusion code of the customer C, “CO, .LTD”).

  Although the details of the unnecessary word dictionary will be described later, the conversion target character string representing the name of the specific customer has a redundant word other than the plural words represented by the logical index set of the specific customer registered in the name dictionary. If it is an added character string, the extra word is associated with the customer code of the specific customer and registered as an unnecessary word of the specific customer in the unnecessary word dictionary.

  Subsequently, after extracting text data of a character string representing a customer name from a message received from another financial institution, the customer code conversion process performed by the computer 12 by the CPU 12A of the computer 12 executing the customer code conversion program. This will be described with reference to FIG. This customer code conversion process is executed every time a telegram including text data of a character string representing a customer name is received from another financial institution.

  In step 50, the conversion target character string extracted from the received electronic message is divided into a plurality of words with a space existing in the conversion target character string as a delimiter. In step 52, text data of an unprocessed word is extracted from a plurality of words obtained by dividing the character string in step 50. In the next step 54, a basic word dictionary is created using the text data of the extracted word as a key. Search for. In this search, for example, the entire basic word dictionary is read from the HDD 22 and expanded in the memory, the text data of the extracted word is stored in the memory (RAM 12C, etc.), and then the text data of the word to be searched is expanded in the memory. The data may be compared with the text data of the individual words registered in the basic word dictionary in order, or the text data of a single word is read from the basic word dictionary of the HDD 22 and stored in the memory. It is also possible to repeat the comparison of the text data stored in the text data with the text data of the search target word stored in the memory until text data matching the text data of the search target word appears. is there.

  In step 56, it is determined whether or not the search target word is extracted from the basic word dictionary by the search in step 54, that is, whether or not the search target word is registered in the basic word dictionary. If the determination is affirmative, the process proceeds to step 64, where the logical index (numerical data) associated with the extracted word to be searched and registered in the basic word dictionary is extracted, and the words constituting the conversion target character string The process proceeds to step 66 after being stored in the memory as a logical index. On the other hand, if the determination in step 56 is negative, the process proceeds to step 58 where the logical index assigned to the search target word is determined so as not to overlap with the logical index already registered in the basic word dictionary. In step 60, the text data of the search target word is registered in the basic word dictionary in association with the logical index (numerical data) determined in step 58. Further, in step 62, the logical index determined in step 58 is stored in the memory as a logical index of words constituting the conversion target character string, and then the process proceeds to step 66.

  In step 66, it is determined whether all words have been extracted from the conversion target character string. If the determination is negative, the process returns to step 52, and steps 52 to 66 are repeated until the determination of step 66 is affirmed. As a result, each word constituting the conversion target character string is converted into a logical index, and a combination of logical indexes of the conversion target character string (logical index set) is stored in the memory. Become. Steps 52 to 66 correspond to the word conversion means described in claim 7.

  If the determination in step 66 is affirmative, the process proceeds to step 68, where information on a single customer is extracted from the name dictionary and stored in the memory, and the logical index set included in the extracted information is converted into the logical index set of the character string to be converted. Compare with The comparison of the logical index sets may be performed after the entire name dictionary is read from the HDD 22 and expanded in the memory, or the single customer information is read from the HDD 22 and stored in the memory repeatedly. It is also possible. In the next step 70, the logical index set of the conversion target character string includes a logical index set extracted from the name dictionary (each logical index constituting the logical index set extracted from the name dictionary is converted into the logical string of the conversion target character string. Whether all exist in the index set).

  If the determination is denied, the process proceeds to step 74 without performing any processing. If the determination is affirmed, the process proceeds to step 72, and the customer code is obtained from the single customer information extracted from the name dictionary in step 68. And the number of logical indexes (number of unmatched words) not included in the logical index set extracted from the name dictionary among the logical indexes constituting the logical index set of the conversion target character string is counted and extracted. After the customer code and the counted number of mismatch words are stored in the memory as the association search result information with the logical index set extracted from the name dictionary, the process proceeds to step 74. In step 72, if the single customer information extracted from the name dictionary includes an inclusion code, the inclusion code is also extracted and stored in the memory as search result information.

  In step 74, it is determined whether all the logical indexes registered in the name dictionary have been extracted (compared with the logical index set of the character string to be converted). If the determination is negative, the process returns to step 68, and steps 68 to 74 are repeated until the determination of step 74 is affirmed. In the search processing of step 68 to step 74, a single logical index set is often extracted as a logical index set that matches the determination condition of step 70. For example, the conversion target character string has an inclusion relationship. A character string corresponding to a customer name including a non-included word (for example, “HAYASHI ELECTRIC CO, .LTD” in the above example), or a character string in which a word different from the non-included word is further added to the character string ( For example, "HAYASHI ELECTRIC CO, .LTD SHINZYUKU"), a plurality of logical index sets to which the same inclusion code is added are extracted.

  Even when the conversion target character string does not have an inclusion relationship, a plurality of logical index sets may be extracted by the search processing of step 68 to step 74. For example, the combination of words in the character string “AKASAKA BANK CO., LTD SHINZYUKU” is the name “AKASAKA BANK CO., LTD” of customer A and the name “SHINZYUKU BANK CO., LTD” of customer E shown in FIG. ”And the conversion target character string is“ AKASAKA BANK CO., LTD SHINZYUKU ”, the logical index sets of customer A and customer E are respectively extracted. . Steps 68 to 74 described above correspond to the search means according to the present invention.

  If the determination in step 74 is affirmed, the process proceeds to step 76, and a customer code determination process is performed. This customer code determination process will be described with reference to FIG. In step 80, it is determined whether or not a logical index set that matches the determination condition in step 70 has been extracted by the search processing in steps 68 to 74 described above. If this determination is negative, it can be determined that there is a serious error in the conversion target character string extracted from the received message, so the process moves to step 82 and the operator is called by displaying a predetermined message on the display 16 or the like. The process is terminated, and the decision of the customer name and customer code represented by the conversion target character string is left to the operator.

  On the other hand, if the determination in step 80 is affirmed, the process proceeds to step 83, and it is determined whether a plurality of logical index sets that match the determination condition in step 70 have been extracted by the search processing in steps 68 to 74. If the determination is negative, the process proceeds to step 84. If the determination is affirmative, the process proceeds to step 126, and is stored in the memory corresponding to each of the plurality of logical index sets extracted by the search process. Based on whether the same inclusion code is included in each search result information, it is determined whether there is an inclusion relationship between customer names corresponding to a plurality of logical index sets extracted by the search process To do. If this determination is negative, multiple-set extraction processing is performed in step 128. If the determination is positive, the process proceeds to step 84. Note that the process for extracting a plurality of sets in step 128 will be described later.

  In step 84, among the logical index sets stored in the memory by the search in the previous step 68 to step 74, the logical index set having the smallest number of unmatched words (hereinafter, this logical index set is referred to as “first logical index set for convenience”). "). When only a single logical index set is extracted (stored in the memory) by the search in steps 68 to 74, this logical index set is treated as the first logical index set in the following processing. In step 86, based on whether or not an inclusion code is included in the search result information corresponding to the first logical index set, whether or not the customer name corresponding to the first logical index set has an inclusion relationship. To determine. If this determination is negative, the process proceeds to step 90, and the customer code corresponding to the first logical index set is output as the customer code corresponding to the conversion target character string.

  For example, in the example shown in FIG. 2, the conversion target character string is “AKASAKA BANK CO., LTD” (logical index set is “100, 101, 107”) or “AKASAKA BANK CO., LTD MARUNOUCHI” (logical index set is “100, 101, 107, 110”). If it is, the processing of the previous step 68 to step 74 corresponds to the character string “AKASAKA BANK CO., LTD” representing the name of the customer A as a logical index set that matches the determination condition of step 70. The logical index set “100, 101, 107” is extracted from the name dictionary. Here, when the extracted logical index set is only the above-described logical index set (when the determination in step 83 is negative), the inclusion corresponding to the extracted logical index set (first logical index set). The code is not registered in the name dictionary (see Fig. 2 (B)), and the name of customer A has no inclusion relationship. ., LTD MARUNOUCHI ", it can be determined that the character string to be converted represents the name of customer A. In such a case, the determination in step 86 is affirmed as described above, and in step 90, the customer code “1230000” corresponding to the first logical index set is used as the customer code corresponding to the character string to be converted. Will be output.

  In the next step 94, it is determined whether the logical index set of the character string to be converted completely matches the logical index set of the customer name corresponding to the output customer code (the number of mismatched words is 0). . If the determination is affirmative, the customer code determination process (customer code conversion process) is terminated. For example, in the above example, if the conversion target character string is “AKASAKA BANK CO., LTD”, the logical index set of the conversion target character string completely matches the logical index set of the customer name corresponding to the output customer code. Therefore, the determination at step 94 is affirmed and the process is terminated. If the conversion target character string is “AKASAKA BANK CO., LTD MARUNOUCHI”, the conversion target character string for the character string “AKASAKA BANK CO., LTD” representing the customer name corresponding to the output customer code The extra word “MARUNOUCHI” is added, and the logical index set of the conversion target character string does not completely match the logical index set of the customer name corresponding to the output customer code. Then, the process proceeds to step 96.

  Here, the extra word added to the conversion target character string (in the above example, the word “MARUNOUCHI”) is a word that is essentially unnecessary for determining the customer code corresponding to the conversion target character string. When a character string representing the same customer name as the character string is input in the future, it is highly likely that the same word is added to the character string, so the customer name represented by the character string to be converted is There is a possibility that the word can be used when determining whether or not the customer name corresponds to the output customer code. For this reason, in step 96, first, an unnecessary word dictionary is searched using the output customer code as a key to determine whether the output customer code is registered in the unnecessary word dictionary, that is, an unnecessary word corresponding to the output customer code. After confirming whether or not is registered in the unnecessary word dictionary, the character representing the customer name corresponding to the output customer code included in the conversion target character string in the logical index of the registered unnecessary word Check whether there is a logical index of unmatched words (word "MARUNOUCHI" in the example above) that is not included in the column. If the unmatched word (logical index) is not registered as an unnecessary word, the logical index of the mismatch word is registered as an unnecessary word in the unnecessary word dictionary in association with the output customer code, and the process is terminated (FIG. 2 (C)).

  If there are a plurality of mismatch words, the logical index of each mismatch word is registered in the unnecessary word dictionary as an unnecessary word in step 96. In step 96, if the logical index of the mismatched word is already registered in the unnecessary word dictionary in association with the output customer code, the process is terminated without performing any process. In step 96, if the output customer code is not registered in the unnecessary word dictionary, the logical index of the mismatched word and the output customer code are associated with each other and registered in the unnecessary word dictionary.

  On the other hand, if the determination in step 86 is affirmative, the process proceeds to step 88, where the logical index of the non-included word included in the information of the first logical index set stored in the memory is referred to, and By determining whether or not the logical index is included in the logical index set of the conversion target character string, it is determined whether or not the non-inclusive word is included in the conversion target character string. If the determination is affirmative, the process proceeds to step 89, and the customer code corresponding to the logical index set including the non-inclusive word among the logical index sets extracted by the search processing of step 68 to step 74 is converted. The customer code corresponding to the character string is output and the process proceeds to step 94.

  For example, in the example of FIG. 2, the conversion target character string is “HAYASHI ELECTRIC CO, .LTD” (logical index set is “105,106,107”) or “HAYASHI ELECTRIC CO, .LTD SHINZYUKU” (logical index set is “105,106,107,111”). If there is, in the search processing of step 68 to step 74, as a logical index set that matches the determination condition of step 70, a logical index set of a plurality of customers having an inclusion relationship, that is, a character string “HAYASHI representing the name of customer B” The logical index set “105,107” corresponding to “CO, .LTD” and the logical index set “105,106,107” corresponding to the character string “HAYASHI ELECTRIC CO, .LTD” representing the name of the customer C are respectively extracted from the name dictionary. Because the conversion target character string includes the non-inclusive word "ELECTRIC" in the above inclusion relationship, the conversion target character string includes the non-inclusive word It can be determined that represents the name of that customer C. In the customer code conversion process according to the present embodiment, the determination in steps 83 and 126 is affirmed in the above case, and then the determination in step 88 is affirmed and the process proceeds to step 89. As a customer code corresponding to, a customer code assigned to a customer whose name includes an uninclusive word (customer code “3450000” assigned to customer C in the above example) is output.

  In the above example, if the conversion target character string is “HAYASHI ELECTRIC CO, .LTD”, the logical index set of the conversion target character string completely matches the logical index set of the customer name corresponding to the output customer code. Therefore, the determination in step 94 is affirmed and the process is terminated. On the other hand, if the conversion target character string is “HAYASHI ELECTRIC CO, .LTD SHINZYUKU”, the conversion target character string for the character string “HAYASHI ELECTRIC CO, .LTD” that represents the customer name corresponding to the output customer code An extra word “SHINZYUKU” is added, and the logical index set of the conversion target character string does not completely match the logical index set of the customer name corresponding to the output customer code, so the determination in step 94 is negative. Then, the process proceeds to step 96. A mismatch word that is included in the conversion target character string and not included in the character string that represents the customer name corresponding to the output customer code (in the above example, the word “SHINZYUKU”) is included in the output customer code. If it is not registered as a corresponding unnecessary word, it is registered in the unnecessary word dictionary corresponding to the customer code output by the mismatched word (see also FIG. 2 (C)). Note that the above-described steps 94 and 96 correspond to the unnecessary word registering means described in claim 6, and each of the steps other than the above in the customer code determination process (including the multiple set extraction process described later) is the present invention. Corresponds to the conversion means.

  In addition, the customer name corresponding to the first logical index set has an inclusion relationship, and the character string to be converted does not include a non-inclusive word in the above inclusion relationship (for example, “HAYASHI CO, .LTD”)), or If the character string is a character string (for example, “HAYASHI AND CO, .LTD” or “HAYASHI CO, .LTD SHINZYUKU”) in which a word other than the non-inclusive word is added to the character string, the search processing of step 68 to step 74 Then, as a logical index set that matches the determination condition of step 70, a logical index set corresponding to a customer name that does not include a non-included word among a plurality of customer names having an inclusion relationship, that is, a character string “name of customer B” Only the logical index set “105,107” corresponding to “HAYASHI CO, .LTD” is extracted, but although the conversion target character string is likely to be a character string representing the name of customer B, customer B ’s There is a possibility that the specific word “ELECTRIC” in the name “HAYASHI ELECTRIC CO, .LTD” of the customer C inclusive relationship with the name is a missing character string for some reason (a character string representing the name of the customer C) .

  In the above case, the determination in step 88 is affirmed and the process proceeds to step 98, and whether or not the logical index set of the character string to be converted completely matches the first logical index set (the number of mismatched words is 0). To determine. If the determination is affirmative (for example, if the conversion target character string is “HAYASHI CO, .LTD”), the process proceeds to step 108 without performing any processing, but if the determination is negative (for example, the conversion target character string is In the case of “HAYASHI AND CO, .LTD” or “HAYASHI CO, .LTD SHINZYUKU”, the process proceeds to step 100, and the conversion is performed by comparing the logical index set of the conversion target character string with the first logical index set. A logical index of a word (non-matching word) that exists in the target character string but does not exist in the character string corresponding to the first logical index set is extracted. For example, when the conversion target character string is “HAYASHI AND CO, .LTD” and the first logical index set is the logical index set of the customer B, the mismatch word is “AND”, so the logical index “AND” 108 "is extracted.

  In the next step 102, an unnecessary word dictionary is searched using the customer code corresponding to the first logical index set as a key. The search for the unnecessary word dictionary may be performed after the entire unnecessary word dictionary is read from the HDD 22 and expanded in the memory, or the single customer information is read from the HDD 22 and stored in the memory repeatedly. Is also possible. In the next step 104, it is determined whether or not the corresponding information is extracted by the search in step 102. If the determination is affirmative, the process proceeds to step 106, and it is determined whether or not the information extracted by the search in step 102 includes the logical index of the mismatch word extracted in the previous step 100.

  If the determination in step 106 is affirmative, the mismatch word from which the logical index has been extracted in the previous step 100 is registered in the unnecessary word dictionary as an unnecessary word for the customer name corresponding to the first logical index set. Therefore, when the customer name corresponding to the first logical index set is represented, the character string to which the above mismatch word is added exists in the character string determined in the past (for example, as described above) In the example, if the name of the customer B is represented, the character string to which the above “AND” is added exists in the character string determined in the past), and the conversion target character string is the first character string. It can be determined that the probability of representing the customer name corresponding to the logical index set is very high. For this reason, when the determination in step 106 is affirmed, the process proceeds to step 107, the customer code corresponding to the first logical index set is output as the customer code corresponding to the character string to be converted, and the process is terminated.

  In the previous step 100, a plurality of words may be extracted as mismatch words, but the extracted plurality of mismatch words are associated with the customer code corresponding to the first logical index set and the unnecessary word dictionary. The determination in step 106 may be affirmed only when all are registered in the list, or only one of the plurality of extracted mismatch words corresponds to the first logical index set. The determination in step 106 may also be affirmed even if it is registered in the unnecessary word dictionary in association with the customer code (in this case, the process proceeds to step 96 where the unregistered unmatched word is assigned to the first word. Preferably, it is registered as an unnecessary word in the unnecessary word dictionary in association with the customer code corresponding to the logical index set).

  On the other hand, if the determination in step 104 or step 106 is affirmative, the logical character in the previous step 100 is included in the character string determined in the past as representing the customer name corresponding to the first logical index set. It is determined that there is no character string to which the mismatch word from which the index has been extracted is added, and the conversion target character string represents the customer name corresponding to the first logical index set based on the mismatch word. It is difficult. Therefore, if the determination in step 104 or step 106 is affirmed, the process proceeds to step 108, the name dictionary is searched using the inclusion code added to the first logical index set as a key, and the first logical index Information of the second logical index set representing the customer name having an inclusion relationship with the customer name corresponding to the set is extracted from the name dictionary.

  In the next step 110, by comparing the logical index set of the conversion target character string with the second logical index set, the character string corresponding to the second logical index set exists in the conversion target character string. Extracts the logical index of non-matching mismatched words. For example, when the conversion target character string is “HAYASHI CO, .LTD SHINZYUKU” and the second logical index set is the logical index set of the customer C, the logical index “118” of the mismatch word “SHINZYUKU” is extracted. In the next step 112, an unnecessary word dictionary is searched using the customer code corresponding to the second logical index set as a key. In step 114, it is determined whether or not the corresponding information is extracted by the search in step 112. If the determination is affirmative, the process proceeds to step 116, and it is determined whether or not the information extracted by the search in step 112 includes the logical index of the unmatched word extracted in the previous step 110.

  If the determination in step 116 is affirmative, the mismatch word from which the logical index has been extracted in the previous step 110 is registered in the unnecessary word dictionary as an unnecessary word for the customer name corresponding to the second logical index set. Therefore, when the customer name corresponding to the second logical index set is represented, there is a character string to which the above mismatch word is added in the character string determined in the past. It can be determined that the conversion target character string has a high probability of representing the customer name corresponding to the second logical index set.

  For example, the conversion target character string is “HAYASHI CO, .LTD SHINZYUKU”, and the word “SHINZYUKU” is extracted as a mismatch word of the first logical index set, but corresponds to the first logical index set. While it was not registered in the unnecessary word dictionary as an unnecessary word for customer B, it is also extracted as a mismatched word in the second logical index set and is unnecessary as an unnecessary word for customer C corresponding to the second logical index set If it is registered in the word dictionary, the character string to which the above “SHINZYUKU” is added exists in the character string determined in the past as representing the name of the customer C. The character string to be converted can be determined to have a high probability of being a character string representing the name of the customer C although the word “ELECTRIC” is missing. For this reason, when the determination in step 116 is affirmed, the process proceeds to step 118, where the customer code corresponding to the second logical index set is output as the customer code corresponding to the character string to be converted, and the process ends.

  In the previous step 110, a plurality of words may be extracted as mismatched words, but the extracted plurality of mismatched words are unnecessary words associated with the customer code corresponding to the second logical index set. The determination in step 116 may be affirmed only when all of the words are registered in the dictionary, or only one of the extracted mismatch words is included in the second logical index set. The determination in step 116 may also be affirmed even if it is registered in the unnecessary word dictionary in association with the corresponding customer code (also in this case, the process proceeds to step 96, where the unregistered unmatched word is assigned to the second word. It is preferable to register it as an unnecessary word in the unnecessary word dictionary in association with the customer code corresponding to the logical index set).

  On the other hand, if the determination in step 114 or step 116 is affirmative, the logical character in the previous step 110 is included in the character string determined in the past as representing the customer name corresponding to the second logical index set. Since there is no character string to which the inconsistent word from which the index is extracted is added, either the customer name corresponding to the first logical index set or the customer name corresponding to the second logical index set is the character string to be converted. It is difficult to determine whether or not For this reason, when the determination in step 114 or step 116 is affirmed, the process proceeds to step 120 to display the conversion target character string on the display 16 and to the character string corresponding to the first and second logical index sets, the customer Information such as a code is displayed on the display 16 and a predetermined message is displayed on the display 16 to request the operator to determine the customer name represented by the character string to be converted.

  In the next step 122, it is determined whether or not a determination result by the operator has been input, and step 122 is repeated until the determination is affirmed. When the determination of the customer name is requested, the operator determines the customer name represented by the character string to be converted by performing an operation such as making an inquiry to the second financial institution that sent the message. Then, information indicating the determination result is input via the keyboard 20. As a result, the determination in step 122 is affirmed and the process proceeds to step 124. After the customer code corresponding to the input determination result is output as the customer code corresponding to the conversion target character string, the process proceeds to step 94. Steps 120 to 124 correspond to the conversion means described in claim 5.

  Therefore, if the character string to be converted is a character string in which an extra word is added to the customer name determined by the operator, the determination in step 94 is affirmed and the process proceeds to step 96, where the character string to be converted is converted. Since unnecessary words in the list are registered in the unnecessary word dictionary as unnecessary words for the determined customer, the same word (unnecessary word) is added as a character string representing the same customer after the next time. When the generated character string appears, the determination of step 106 or step 116 is affirmed based on this unnecessary word, so that the customer name represented by the conversion target character string is automatically determined without bothering the operator again. Can be determined.

  Subsequently, when a plurality of logical index sets are extracted by the search processing of step 68 to step 74, and there is no inclusion relationship between customer names corresponding to the extracted logical index sets (extracted logical index sets Is composed of only logical index sets corresponding to customer names having no inclusion relationship, or logical index sets corresponding to customer names having no inclusion relationship and logical index sets corresponding to customer names having inclusion relationship are mixed. In the case), the multiple-set extraction process executed in step 128 will be described with reference to the flowchart of FIG.

  In step 140, the search result information corresponding to each of the plurality of logical index sets extracted by the search process is sequentially referred to, and it is determined whether or not the search result information including the inclusion code exists. In the plurality of logical index sets extracted by the search process, it is determined whether or not there is a logical index set in which the corresponding customer name has an inclusion relationship with other customer names. If the determination is negative, the process proceeds to step 142, and after only a plurality of logical index sets extracted by the search process are set as processing targets, the process proceeds to step 148. On the other hand, if the determination in step 140 is affirmative, the process proceeds to step 144, and a logical index in which an inclusion code exists in the corresponding search result information among the plurality of logical index sets extracted by the search process. For a set (logical index set in which the corresponding customer name has an inclusion relationship with other customer names), the name dictionary is searched using the inclusion code as a key, and the customer name having an inclusion relationship with the customer name represented by the logical index set Is extracted from the name dictionary. In the next step 146, the plurality of logical index sets extracted in the search process and the logical index set extracted in step 144 are set as processing targets, and the process proceeds to step 148.

  In the next step 148, a single unprocessed logical index set is extracted from the logical index set set as the processing target. In the next step 150, the logical index set of the conversion target character string is completely identical to the extracted logical index set. Determine if you are doing it. If the determination is affirmative, the process proceeds to step 160 and the number of unnecessary words (unnecessary among mismatched words that are present in the conversion target character string and not present in the character string corresponding to the extracted logical index set) The number of unnecessary words registered as unnecessary words in the word dictionary is set to 0, the number of unnecessary words is stored in the memory in association with the customer code corresponding to the extracted logical index set, and the process proceeds to step 162. .

  On the other hand, if the determination in step 150 is negative, the process proceeds to step 152, where the logical index set of the conversion target character string is compared with the extracted logical index set, and is present in the conversion target character string. Then, all logical indexes of unmatched words that do not exist in the character string corresponding to the extracted logical index set are extracted. In step 153, the unnecessary word dictionary is searched using the customer code corresponding to the extracted logical index set as a key. In the next step 154, it is determined whether or not the corresponding information is extracted by the search in step 152. When negative determination is affirmed, it transfers to step 106, and as above-mentioned, the number of unnecessary words = 0 is matched with a customer code, and is memorize | stored in memory. If the determination in step 154 is affirmed, the process proceeds to step 156, and the logical index of the mismatched word extracted in step 152 is included in the logical index of unnecessary words included in the information extracted by the search in step 153. Determine if it exists.

  If the determination in step 156 is negative, the process proceeds to step 106, and the number of unnecessary words = 0 is stored in the memory in association with the customer code. However, if the determination in step 156 is affirmative, the process proceeds to step 158. Of the unmatched words from which the logical index is extracted in step 152, the number of unmatched words (number of unnecessary words) registered in the unnecessary word dictionary as unnecessary words corresponding to the extracted logical index set is counted, and the count result is used as an unnecessary word. The number is set and stored in the memory in association with the customer code corresponding to the extracted logical index set, and the logical index of the unmatched word registered as an unnecessary word is also stored in the memory in association with the customer code. The process proceeds to step 162.

  In step 162, it is determined whether or not all logical index sets set as processing targets have been extracted. If the determination is negative, the process returns to step 148, and steps 148 to 162 are repeated until the determination of step 162 is affirmed. As a result, the number of unnecessary words is stored in the memory for all logical index sets set as processing targets, and the logic of mismatched words registered as unnecessary words for the logical index set with the number of unnecessary words ≧ 1. Each index is also stored in the memory. If the determination at step 162 is affirmative, the routine proceeds to step 164. In step 164, the number of logical index sets in which the number of unnecessary words is 1 or more is counted while sequentially referring to the number of unnecessary words in each logical index set set as a processing target, and the process branches according to the count result. .

  Here, when the number of logical index sets in which the number of unnecessary words is 1 or more is 0 (any word that does not match the character string corresponding to each logical index to be processed in the conversion target character string is an unnecessary word dictionary) In the case where the logical index is not registered in the previous step 153, the logical index is included in the character string previously determined to represent one of the customer names corresponding to each logical index set to be processed. Since there is no character string to which the extracted mismatched word is added, it is determined whether the conversion target character string represents a customer name corresponding to each logical index set to be processed. It is difficult. For this reason, when the number of logical index sets having one or more unnecessary words is zero, the process proceeds to step 174 to display the conversion target character string on the display 16 and to correspond to each logical index set to be processed. Information such as a character string and a customer code is displayed on the display 16 and a predetermined message is further displayed on the display 16 to request the operator to determine the customer name represented by the character string to be converted.

  In the next step 176, it is determined whether or not a determination result by the operator has been input, and step 176 is repeated until the determination is affirmed. When the operator who has determined the customer name according to the request inputs information representing the determination result via the keyboard 20, the determination in step 176 is affirmed and the process proceeds to step 178, where the customer code corresponding to the character string to be converted is input. After outputting the customer code corresponding to the determination result, the process proceeds to step 94 (see FIG. 4). Steps 174 to 178 correspond to the conversion means described in claim 5. Also in this case, if the conversion target character string is a character string in which an extra word (mismatch word) is added to the customer name determined by the operator, the determination in step 94 is affirmed and the process proceeds to step 96. Since the mismatch word is registered in the unnecessary word dictionary as an unnecessary word for the determined customer, the same word (unnecessary word) is added as a character string representing the same customer from the next time onward. When the written character string appears, the customer name represented by the character string to be converted can be automatically determined based on the unnecessary word without trouble of the operator again.

  Further, when the number of logical index sets whose number of unnecessary words is 1 or more is 1, a single logical index set whose number of unnecessary words is 1 or more among the individual logical index sets to be processed. Only, since one or more mismatched words extracted in the previous step 153 are registered as unnecessary words in the unnecessary word dictionary, it corresponds to a logical index set in which the number of unnecessary words is 1 or more. The character string to which the above mismatched word registered as an unnecessary word was added in the character string determined in the past as representing the customer name to be converted, and the conversion target character string is It can be determined that the probability of representing the customer name corresponding to the logical index set in which the number of unnecessary words is 1 or more is very high.

  For example, when the conversion target character string is “AKASAKA BANK CO., LTD SHINZYUKU” and the information shown in FIG. 2B is registered in the name dictionary, the customer A (name “AKASAKA BANK CO ., LTD ") and customer E (named" SHINZYUKU BANK CO., LTD ") logical index sets are extracted respectively. The mismatch word between the conversion target character string and the name of customer A is “SHINZYUKU” (logical index is “111”), and the mismatch word between the conversion target character string and the name of customer E is “AKASAKA” (logical index is “100”). However, as is clear from FIG. 2C, “SHINZYUKU” is registered as an unnecessary word of the customer A in the unnecessary word dictionary, and the logical index set of the customer A is associated with this. Whereas the number of unnecessary words = 1, the unnecessary word of customer E is not registered in the unnecessary word dictionary, and the number of unnecessary words = 0 for the logical index set of customer E. Therefore, the character string to which the mismatch word “SHINZYUKU” is added exists in the character string determined in the past as representing the name of the customer A, while the name of the customer E is represented. The character string to which the mismatch word “AKASAKA” is added does not exist in the character strings determined in the past, and the conversion target character string has a very high probability of representing the name of the customer A. It can be judged.

  For this reason, when the number of logical index sets in which the number of unnecessary words is 1 or more is 1, the process moves from step 164 to step 172, and the number of unnecessary words is 1 or more as the customer code corresponding to the conversion target character string. After outputting the customer code corresponding to the logical index set, the process proceeds to step 94 (see FIG. 4). Also in this case, if the character string to be converted is a character string in which an extra word (unmatched word) not registered as an unnecessary word is added to the customer name corresponding to the output customer code, the determination in step 94 is made. Is affirmed and the process proceeds to step 96, and the mismatch word is registered in the unnecessary word dictionary as an unnecessary word for the customer corresponding to the output customer code.

  By the way, for example, when the conversion target character string is “NAKADA AND AKASAKA BANK CO., LTD SHINZYUKU BRANCH”, if the information shown in FIG. 2B is registered in the name dictionary, the customer A (“ AKASAKA BANK CO., LTD "), customer E (" SHINZYUKU BANK CO., LTD ") and customer F (" NAKADA AND AKASAKA CO., LTD ") are extracted. The inconsistent words between the conversion target character string and the name of customer A are “NAKADA”, “AND”, “AKASAKA”, “SHINZYUKU”, and “BRANCH”, but as is apparent from FIG. Since only “AND” and “SHINZYUKU” are registered as unnecessary words of customer A, the number of unnecessary words is 2 for customer A, and the logical index of these two unnecessary words is stored in the memory. The For customer E, “NAKADA”, “AND”, “AKASAKA”, and “BRANCH” are mismatched words, but since the unnecessary word of customer E is not registered in the unnecessary word dictionary, the number of unnecessary words for customer E = 0. It becomes. Furthermore, for customer F, “BANK”, “SHINZYUKU”, and “BRANCH” are mismatched words, and these mismatched words are registered in the unnecessary word dictionary as unnecessary words of customer F, so the number of unnecessary words for customer F = 3 and the logical indexes of the three unnecessary words are stored in the memory. In this example, the number of logical index sets in which the number of unnecessary words is 1 or more is “2”.

  Here, for the determination of the customer name represented by the conversion target character string, the number of unnecessary words in each logical index set to be processed may be simply compared (this invention is also included in the scope of rights). When an unnecessary word in a certain logical index set is an unnecessary word in another logical index set (when unnecessary words are duplicated), the conversion target character string is represented based on this unnecessary word. Since it is difficult to specify the customer name, it is expected that the determination accuracy is improved by removing this unnecessary word.

  For this reason, if the number of logical index sets whose number of unnecessary words is 1 or more is larger than 1 (a plurality of relevant logical index sets exist), the process proceeds from step 164 to step 166 to perform processing. By comparing the logical indexes of unnecessary words stored in the memory for each target logical index set with each other, it is determined whether or not there is an unnecessary word duplication for each logical index set (corresponding customer code). To do. If the determination is negative, the process proceeds to step 170 without performing any processing. If the determination is affirmative, the process proceeds to step 168, and a logical index set in which unnecessary words are duplicated (corresponding customer code). After subtracting the number of unnecessary words by the number of redundant unnecessary words, the process proceeds to step 170.

  For example, in the above example, the unnecessary words for customer A are “AND” and “SHINZYUKU”, and the unnecessary words for customer F are “BANK”, “SHINZYUKU”, and “BRANCH”, so the unnecessary word “SHINZYUKU” is duplicated. In the process of step 168, one is subtracted from the number of unnecessary words of customer A and customer F, respectively. In this example, the number of unnecessary words of the customer A = 1 and the number of unnecessary words of the customer F = 2, and as a result, the size relationship of the number of unnecessary words does not change. For example, the number of unnecessary words in a certain logical index set is large. If a large number of words overlaps with unnecessary words in other logical index sets, the number of unnecessary words of the customer is reduced by the processing in step 168, so that the size relationship of the number of unnecessary words changes, and the conversion target The number of unnecessary words in the logical index set corresponding to the customer name presumed to have a high probability represented by the character string is relatively increased.

  In step 170, the number of unnecessary words stored in the memory corresponding to each logical index set set as a processing target is sequentially referenced to extract the maximum number of unnecessary words and the number of unnecessary words becomes the maximum value. After counting the number of existing logical index sets, whether there are multiple logical index sets with the maximum number of unnecessary words in each logical index set set as the processing target based on the count result Judge whether or not. When there is only one logical index set having the maximum number of unnecessary words, it can be determined that the conversion target character string has a high probability of representing the customer name corresponding to this logical index set. Therefore, if the determination in step 170 is negative, the process proceeds to step 172, and after outputting the customer code corresponding to the logical index set having the largest number of unnecessary words as the customer code corresponding to the character string to be converted, step 94 is performed. (See FIG. 4). For example, in the above-described example, it is determined that the conversion target character string represents the name of the customer F having the maximum number of unnecessary words (the number of unnecessary words = 2), and the customer code of the customer F is output. become.

  On the other hand, when there are a plurality of logical index sets with the maximum number of unnecessary words, it is determined which of the customer names corresponding to the plurality of logical index sets with the maximum number of unnecessary words is the character string to be converted. It is difficult to judge. For this reason, when the determination at step 170 is affirmed, the routine proceeds to step 174. In this case, as described above, the operator determines the customer name represented by the conversion target character string.

  As described above, in the present embodiment, when a plurality of logical index sets are extracted by the search process when the customer name (customer code) represented by the conversion target character string is determined, the conversion is performed by the above-described multiple set extraction process. Whether or not a mismatch word that exists in the target character string and does not exist in the character string corresponding to the extracted logical index set is registered as an unnecessary word in the unnecessary word dictionary, and is unnecessary among the mismatch words Since the logical index set (customer code) corresponding to the conversion target character string is determined based on the number of words registered as words (the number of unnecessary words) and the presence / absence of unnecessary word duplication in each logical index Improves the probability that the customer name (customer code) represented by the character string to be converted can be automatically determined without bothering the operator It can be.

  In the above, when a plurality of logical index sets are extracted by the search process and there are a plurality of logical index sets with the number of unnecessary words ≧ 1, conversion target characters are determined based on the number of unnecessary words and the presence or absence of unnecessary words. Although the example which judges the customer name (customer code) which a row | line | column represents was demonstrated, this invention is not limited to this, For example, conversion object is based on either one of the number of unnecessary words and the presence or absence of unnecessary word duplication Alternatively, the customer name (customer code) represented by the character string may be determined. Alternatively, for each unnecessary word, the number of times (or frequency) that appears in the character string representing the corresponding customer name is determined as the unnecessary word. When there are a plurality of logical index sets with the number of unnecessary words ≧ 1 stored in the dictionary, the frequency of appearance of each unnecessary word in the character string representing the corresponding customer name To determine the customer name (customer code) estimated to be represented by the conversion target character string (for example, calculating an evaluation value by weighting each unnecessary word according to the appearance frequency). Also good.

  The frequency of appearance of unnecessary words is, for example, stored in the unnecessary word dictionary for each individual customer name the number of times a character string representing each individual customer name is input, and a character string representing the corresponding customer name for each unnecessary word Is stored in the unnecessary word dictionary (the number of input times and the number of appearances correspond to the frequency information described in claim 4), and the character string to be converted represents a certain customer name. Each time it is judged, the number of times of input of the character string corresponding to the customer name is incremented, and the number of times of appearance of the unnecessary word included in the conversion target character string is repeated to calculate the appearance frequency of the specific unnecessary word. In this case, the number of appearances of the specific unnecessary word can be determined by dividing the number of appearances of the character string representing the corresponding customer name.

  In the above description, an example in which only the logical index of unnecessary words is associated with the customer code and registered in the unnecessary word dictionary has been described as an example, but the present invention is not limited to this. For example, when registering unnecessary words, The position of the unnecessary word in the character string including the unnecessary word is also registered as information, and the conversion is performed in consideration of whether the position of the unnecessary word in the conversion target character string matches the position registered in the unnecessary word dictionary. The customer name represented by the target character string may be determined. As described above, the appearance frequency or the number of appearances of each unnecessary word is registered in the unnecessary word dictionary, and the appearance frequency or the number of appearances is appropriately updated. Only unnecessary words whose appearance frequency or number of appearances is greater than or equal to a predetermined value may be used for determining the customer name represented by the conversion target character string.

  In addition, in the above, when a word extracted from the conversion target character string is not registered in the basic word dictionary, a logical index is assigned to the word and newly registered in the basic word dictionary, thereby performing conversion Although an example of converting all the words constituting a character string to a logical index has been described, the present invention is not limited to this, and words that are not registered in the basic word dictionary among the words constituting the character string to be converted In this case, the subsequent processing may be performed as it is without converting to the logical index. In this case, there is a possibility that words registered as logical indexes and words registered as text data may coexist in the unnecessary word dictionary and the auxiliary dictionary, but the present invention includes such an aspect within the scope of rights.

  In the above description, the character string representing the customer name is described as an example of the character string according to the present invention. However, the present invention is not limited to this, and represents an arbitrary object formed by combining a plurality of words. Applicable to strings.

It is a block diagram which shows schematic structure of the computer system which concerns on this embodiment. (A) is an example of a basic word dictionary, (B) is an example of a name dictionary, and (C) is an image diagram showing an example of an unnecessary word dictionary. It is a flowchart which shows the content of the customer code conversion process. It is a flowchart which shows the content of the customer code determination process. It is a flowchart which shows the content of the process at the time of multiple set extraction.

Explanation of symbols

10 Computer System 12 Computer 16 Display 18 Mouse 20 Keyboard 22 HDD
28 network 30 computer

Claims (8)

Each time a character string composed of a plurality of words combined to represent a predetermined object is input, the input character string is presumed to be represented by the input character string. An information conversion device for converting into identification information given to the target of
For each object to which identification information is assigned, a name dictionary registered by associating a combination of identification information given to each object and a word constituting a character string representing each object is stored. First storage means;
A single or a plurality of specific words for a combination of words that have been input in the past and that are registered in the name dictionary in association with identification information assigned to the specific target The single or plural specific words in the character string configured by the combination of the added words are associated with the identification information given to the specific target and are registered as unnecessary words of the specific target Second storage means for storing a word dictionary;
By comparing each combination of words constituting the input character string with each combination of words registered in the name dictionary, each of the constituent words is present in the input character string. A search means for searching for word combinations;
Based on a search result by the search means, a target estimated to be represented by the input character string is determined, and the input character string is converted into identification information given to the determined target. A combination of words excluding a single or a plurality of first specific words from a plurality of words constituting the input character string in a combination of words extracted by a search by the search means; A combination of words excluding the first specific word when a combination of words excluding a single or plural second specific words is included from a plurality of words constituting the input character string; Search whether or not the first identification information associated and registered in the name dictionary is associated with at least one of the single or plural first specific words and registered in the unnecessary word dictionary Then Second identification information associated with a combination of words excluding the second specific word and registered in the name dictionary is associated with at least one of the single or plural second specific words Whether or not it is registered in the unnecessary word dictionary and only one of the first and second identification information is at least one of the single or plural first specific words, or the single or When the unnecessary character dictionary is associated with at least one of the plurality of second specific words and registered in the unnecessary word dictionary, the input character string is included in the unnecessary word dictionary of the first and second identification information. Conversion means for determining that the object corresponds to the registered identification information;
An information conversion device comprising:   The conversion means searches whether the first identification information is registered in the unnecessary word dictionary in association with at least one of the single or plural first specific words, and the second As a result of searching whether the identification information is associated with at least one of the single or plural second specific words and registered in the unnecessary word dictionary, the first and second identification information are the single Or the first and second when each of the first specific word is registered in the unnecessary word dictionary in association with at least one of the plurality of first specific words or at least one of the single or plural second specific words. Among the identification information, among the words included in the first specific word or the second specific word, other than the words registered in the unnecessary word dictionary associated with the first and second identification information, respectively. Associated with a word Determining the identification information are information converter according to claim 1, wherein said input character string is and determines that represent the object corresponding to the determined identification information.   The conversion means searches whether the first identification information is registered in the unnecessary word dictionary in association with at least one of the single or plural first specific words, and the second As a result of searching whether the identification information is associated with at least one of the single or plural second specific words and registered in the unnecessary word dictionary, the first and second identification information are the single Or the first and second when each of the first specific word is registered in the unnecessary word dictionary in association with at least one of the first specific words or at least one of the single or plural second specific words. Among the identification information, the identification information that is included in the input character string as the first specific word or the second specific word and is associated with a larger number of words and registered in the unnecessary word dictionary is determined, The input Character string information converter according to claim 1, characterized in that it is determined that represents the object corresponding to the determined identification information. In the unnecessary word dictionary stored in the second storage unit, in the character string input in the past, frequency information indicating the frequency of appearance of a specific target unnecessary word represented by the character string is stored in the unnecessary word dictionary. Each registered unnecessary word is registered,
Each time the conversion means determines a target estimated to be represented by the input character string based on a search result by the search means, a specific character string that the character string represents in the input character string Based on whether or not the target unnecessary word is included, the frequency information of the unnecessary word of the specific target registered in the unnecessary word dictionary is updated, and the word extracted by the search by the search unit is updated. In a combination, a combination of words excluding a single or a plurality of first specific words from a plurality of words constituting the inputted character string, and a plurality of words constituting the inputted character string A combination of words excluding single or plural second specific words is included, and the first identification information is associated with at least one of the single or plural first specific words and is unnecessary. Add to word dictionary As a result of searching whether or not the second identification information is associated with at least one of the single or plural second specific words and registered in the unnecessary word dictionary. The first and second identification information are associated with at least one of the single or plural first specific words or at least one of the single or plural second specific words in the unnecessary word dictionary. When each is registered, it is included in the input character string as the first specific word or the second specific word, and is registered in the unnecessary word dictionary in association with the first and second identification information. The information conversion apparatus according to claim 1, wherein an object represented by the input character string is determined based on frequency information of each unnecessary word.   The conversion means searches whether the first identification information is registered in the unnecessary word dictionary in association with at least one of the single or plural first specific words, and the second As a result of searching whether the identification information is associated with at least one of the single or plural second specific words and registered in the unnecessary word dictionary, both the first and second identification information are the When the input character string is not registered in the unnecessary word dictionary in association with the first specific word or the second specific word, the input character string is displayed on the display device. 2. The information variable according to claim 1, wherein the determination of the object to be represented is entrusted to an operator, and the object to be represented by the input character string is determined based on a determination result input by the operator via the input device. Apparatus.   The combination of words extracted by the search by the search means includes a combination of words excluding a single or a plurality of specific words from a plurality of words constituting the input character string, And when the said conversion means judges that the said input character string represents the specific object corresponding to the combination of the word except the said specific word, the said specific word is made into the said specific object. 2. The information conversion apparatus according to claim 1, further comprising an unnecessary word registering unit that associates with the given identification information and registers the unnecessary word as the specific target unnecessary word in the unnecessary word dictionary. Third storage means for storing a basic word dictionary registered in advance by associating a character string corresponding to each word constituting the input character string and a logical index assigned to each word;
By searching the basic word dictionary using character strings corresponding to individual words constituting the inputted character string as keys, the inputted character string is divided into individual words constituting the character string. A word conversion means for converting into a logical index as
Further comprising
In the name dictionary, the logical index combination is registered as a combination of the words, and in the unnecessary word dictionary, a logical index is registered as the unnecessary word,
The search means and the conversion means perform each processing using a combination of logical indexes obtained by converting the input character string in units of words by the word conversion means as the input character string. The information conversion device according to claim 1, wherein the information conversion device is performed. Each time a character string composed of a plurality of words combined to represent a predetermined object is input, it is estimated that the input character string represents the input character string. An information conversion program for functioning as an information conversion device for converting into identification information given to the predetermined target,
For each object to which identification information is assigned, the computer registers each of the identification information assigned to each object and a combination of words constituting a character string representing each object, and registered A first storage means for storing a dictionary and a combination of words registered in the name dictionary in association with identification information given in the past and representing a specific target. The single or plural specific words in a character string composed of a combination of words to which a single or plural specific words are added are associated with the identification information given to the specific object, and the specific A second storage means for storing an unnecessary word dictionary registered as an unnecessary word of the target,
The computer,
By comparing each combination of words constituting the input character string with each combination of words registered in the name dictionary, each of the constituent words is present in the input character string. Search means for searching for word combinations,
And based on the search result by the search means, a target estimated to be represented by the input character string is determined, and the input character string is converted into identification information given to the determined target In addition, the combination of words extracted by the search by the search means includes a combination of words excluding a single word or a plurality of specific words from a plurality of words constituting the input character string. If the specific identification information registered in the name dictionary in association with a combination of words excluding the single or plural specific words is at least one of the single or plural specific words Whether the specific identification information is associated with at least one of the single or plural specific words and stored in the unnecessary word dictionary. An information conversion program that, when registered, causes the input character string to function as a conversion unit that determines that the character string represents an object to which the specific identification information is assigned.

Images