JP4569178B2 - Classification code processor - Google Patents

Description

  The present invention relates to a classification code processing apparatus that supports an operation of assigning a classification code such as an international patent classification to a document.

The accumulated amount of literature materials such as scientific and technical literatures and patent literatures is increasing year by year, and there is a need for a technique for finding target literature materials from a huge amount of literature materials. For example, if it is a patent document, it is devised to make it easy to find a target document material by performing a search by limiting the technical field according to the International Patent Classification (IPC). Patent Document 1 discloses a technique for supporting classification assignment for document materials by referring to a classification table in which keywords and related fields are associated with each other.
JP-A-6-75995

  However, in the conventional technology for supporting classification, it is necessary to generate information on the distance between keywords in advance. For this reason, it is necessary to select a known keyword as a keyword, and perform effective classification processing in scientific and technical literature in which a variety of technical terms are created every day, or in documents that have inherent diversity in the words used. It was difficult.

  The present invention has been made in view of the above circumstances, and one of its objects is to provide a classification code processing apparatus capable of performing effective classification processing even for scientific and technical documents in which various words appear. To do.

  The present invention for solving the problems of the conventional example is connected to a document database that holds a plurality of documents associated with a classification code, and supports assignment of the classification code to a document to be processed. A classification code processing device, comprising: a keyword extracting unit that extracts at least one key character string from a document to be processed as a keyword; and a document group related to each of the extracted keywords from the document database. Document group extraction means for extracting and list generation for generating a list of classification codes for each keyword by referring to the document group extracted for each keyword and acquiring at least a part of the classification codes associated with the document group And the classification mark List is a feature to be subjected to treatment to assist the application of classification code for the processing target document.

  As described above, in the present embodiment, a document stored in the document database is directly searched for keywords without using information on a predetermined distance between keywords, and the document obtained as a result of the search is given. The classified code is used for processing that supports the assignment of the classification code to the document to be processed. Thereby, an effective classification process can be performed even for scientific and technical documents in which various words appear.

  In addition, the keyword extraction unit is characterized in that, in the document to be processed, a portion where a keyword is extracted is specified, and a character string that is a keyword is extracted from a part of the specified document.

  In addition, a dictionary database that stores a keyword and a classification code to be assigned to a document related to the keyword is stored in association with each other, and the classification is stored in the dictionary database in association with the keyword extracted by the keyword extraction unit. A search means for searching for a code, and if there is no classification code held in association with the keyword extracted by the keyword extraction means as a result of the search by the search means, the document group extraction means Extracting a document group related to each of the extracted keywords, and the list generation means refers to the document group extracted for each keyword and obtains at least a part of a classification code associated with the document group. And ki It is also possible to generate a list of classification codes for each word.

  Further, the document group extracting means, when at least one of a plurality of types of classification codes is selectively given to the document held in the document database, A group of documents to which a classification code is assigned may be extracted from the document database.

  Further, based on the extracted keyword, a central concept word related to the content of the document to be processed is acquired, and at least a part of a classification code related to the central concept word is acquired, and classification for each keyword is performed. And a second list generation means for generating a second list of codes, wherein the list of classification codes for each keyword and the second list are for processing that supports the assignment of classification codes to the document to be processed. It may be provided.

  At this time, at least a part of the extracted keywords may be selected according to a predetermined rule, and the selected keywords may be acquired as central concept words related to the contents of the document to be processed.

  In addition, a classification code processing method according to an aspect of the present invention uses a computer that is connected to a document database that holds a plurality of documents associated with a classification code. A classification code processing method for supporting assignment, wherein a keyword extraction step of extracting at least one character string as a keyword from a document to be processed as a keyword, and each of the extracted keywords from the document database A document group extraction step of extracting a document group to be performed, and a reference to the document group extracted for each keyword to obtain at least a part of a classification code associated with the document group, and a list of classification codes for each keyword is obtained. List generation process to be generated , Cause the computer to execute a list of the classification code is characterized in that it is subjected to treatment to assist the application of classification code for the processing target document.

  Here, in the keyword extracting step, a part from which the keyword is extracted is specified in the document to be processed, and a character string serving as a keyword may be extracted from a part of the specified document.

  Further, the computer is connected to a dictionary database that stores a keyword and a classification code to be assigned to a document related to the keyword in association with each other. The dictionary database is associated with the keyword extracted by the keyword extraction unit. And when there is no classification code held in association with the extracted keyword as a result of the search, in the document group extraction step, from the document database, A document group related to each of the extracted keywords is extracted, and at the list generation step, the document group extracted for each keyword is referred to, and at least a part of the classification code associated with the document group is acquired. Keyword It is also possible to generate a list of classification codes.

  In the document group extraction step, when at least one of a plurality of types of classification codes is selectively given to the document held in the document database, the type given to the document to be processed It is also possible to extract a group of documents to which the classification code is assigned from the document database.

  Further, based on the extracted keyword, a central concept word related to the content of the document to be processed is acquired, at least a part of a classification code related to the central concept word is acquired, and classification for each keyword is performed. The computer further executes a second list generation step of generating a second list of codes, and the list of classification codes for each keyword and the second list support the assignment of classification codes to the document to be processed. It is good also as being used for the process to do.

  At this time, at least a part of the extracted keywords may be selected according to a predetermined rule, and the selected keywords may be acquired as central concept words related to the contents of the document to be processed.

  The program according to another aspect of the present invention supports assignment of a classification code to a document to be processed to a computer that is connected to a document database that holds a plurality of documents associated with the classification code. A keyword extraction procedure for extracting at least one key character string as a keyword from a document to be processed, and a document related to each of the extracted keywords from the document database A document group extraction procedure for extracting a group and the document group extracted for each keyword are referred to, and at least a part of the classification codes associated with the document group is acquired to generate a list of classification codes for each keyword. List generator When, cause the computer to execute a list of the classification code is characterized in that it is subjected to treatment to assist the application of classification code for the processing target document.

  Embodiments of the present invention will be described with reference to the drawings. As shown in FIG. 1, the classification code processing apparatus according to the embodiment of the present invention includes a control unit 11, a storage unit 12, a storage unit 13, an operation unit 14, and a display unit 15. ing.

  The control unit 11 is realized by a CPU (Central Processing Unit) or the like, and operates according to a program stored in the storage unit 12. In the present embodiment, the control unit 11 uses a keyword extraction process for extracting a keyword from the document (hereinafter referred to as a “processing target document” for distinction) using the document to which the classification code is assigned as a processing target. A search process for searching for a classification code associated with the extracted keyword in advance, a classification code list generation process for extracting a document group related to each of the keywords extracted from the document database, and generating a list of classification codes , Selecting a central concept word representing the content of the processing target document from the extracted keywords, obtaining a classification code related to the selected central concept word, and using at least one classification code obtained through these processes Processing to be presented to the person. Details of these specific processes will be described in detail later.

  The storage unit 12 includes a memory element such as a RAM (Random Access Memory), and stores a program executed by the control unit 11. The storage unit 12 also operates as a work memory that holds various data used in the process of the control unit 11.

  The storage unit 13 includes a computer-readable recording medium such as a hard disk device, and stores a program executed by the control unit 11. When processing is performed by the control unit 11, the program is read from the storage unit 13, stored in the storage unit 12, and used for processing. The storage unit 13 stores a document database that holds a plurality of documents to which classification codes are assigned in advance. Furthermore, the storage unit 13 holds a dictionary database that stores a keyword and a classification code to be assigned to a document related to the keyword in association with each other.

  Specifically, the document database held by the storage unit 13 holds a plurality of documents and a classification code associated with each document (FIG. 2). In the description of the present embodiment, as a specific example, a case where this document is a patent document and a scientific and technical document will be described as an example. In addition, each document has a type of a predetermined edition (for example, the seventh edition if it is later than January 2000) in the international patent classification if it is a patent document, for example, depending on the issue date, issuing organization, etc. A classification code is assigned. Here, a plurality of types of classification codes may be assigned to the document. For example, codes such as international patent classification and FI classification may be given.

  Further, as shown in FIG. 3, the storage unit 13 may hold a dictionary database in which a keyword is associated with a classification code to be assigned to a document related to the keyword. This dictionary database may be updated, for example, when a new document is added to the document database. This update process is executed by the control unit 11. The contents of this update process will be described later in detail.

  The operation unit 14 is a mouse, a keyboard, or the like, and outputs the content of the user's instruction operation to the control unit 11. The display unit 15 is a display or the like, and presents information to the user in accordance with an instruction input from the control unit 11.

  Here, the content of the process performed by the control part 11 is demonstrated. First, the keyword extraction process will be described. The control unit 11 first receives an input of a processing target document. Here, the processing target document may be received from a server on a network via a communication unit (not shown), or may be read from a portable recording medium (not shown). Further, text input from the operation unit 14 can also be handled as a processing target document in the present embodiment.

  The control unit 11 extracts keywords from the processing target document. Here, the keyword is a character string serving as a key for searching for a document in the document database. Specifically, the control unit 11 delimits the text included in the processing target document by a unit of subject predicate relation (generally a unit larger than a word), and stores each character string obtained as a result of the division in the storage unit 12. . For example, in the case of Japanese, it may be separated by a particle (case particle / corresponding particle) such as so-called “Tenihaha”. It should be noted that unnecessary particles such as particles are removed from each character string group obtained by the division. Further, an unnecessary word database in which unnecessary words (stop words) are listed in advance is held in the storage unit 13, and the listed unnecessary words are removed from the storage unit 12.

  For example, “The interior of the teapot is divided into two by extending the partition to the vicinity of the opening, and one spout is provided in each chamber obtained by dividing the partition into two tea bowls in one teapot. By dividing the text “Improved convenience so that it can be poured into”, a character string such as “teapot”, “inside”, etc. is obtained. Then, parts such as particles are removed, and “teapot”, “inside”, and so on. Next, character strings such as “inside” and “two minutes” are removed as predetermined unnecessary words (character strings that can appear in any document), and eventually “teapot” and “near the opening” , “Partition”, “each room”, “spout”, “tea pot”, “tea bowl”, “pour”, and the like. The control unit 11 selects these character string groups as keywords.

  Further, the control unit 11 may perform a process of extracting the keyword from the entire processing target document, but may select a part of the processing target document as a target part and extract the keyword from the selected target part. . Specifically, in the case of patent documents, it is also possible to selectively specify a column for claims, a column for describing embodiments and examples as a processing target, and extract a keyword from the specified processing target part. Good.

  Note that here, keywords are extracted by dividing them in units of subject predicates, but after dividing each word by morphological analysis, for example, a character string obtained by concatenating adjacent kanji words May be extracted as a keyword. As a result, for example, a term such as “ultraviolet irradiation device” is subdivided into “ultraviolet light”, “irradiation”, and “device”. Documents related to what to do are included in the search results).

  Next, the control unit 11 searches for a classification code associated in advance with these extracted keywords. That is, the control unit 11 searches each keyword from the dictionary database. If a classification code (a plurality of classification codes) related to the keyword is stored in association with the dictionary database, the list of classification codes is stored in the storage unit 12 in association with the keyword. At this time, when a part of the classification code is registered as an important code in the dictionary database, only the important code may be selectively extracted to form a classification code list.

  If the keyword is not stored in the dictionary database as a result of the search (if there is no classification code associated with the keyword), processing for extracting a document group is performed for the keyword.

  In the classification code list generation process, the control unit 11 extracts a document group related to each extracted keyword from the document database by a full-text search process or the like. Thereby, for each keyword, a document including each keyword is extracted from the document database. The control unit 11 takes out at least a part of the classification code given to each of the documents extracted by the search. For example, when a plurality of classification codes are assigned to the document, the first one is taken out. In addition, when at least one of a plurality of types of classification codes is selectively given, if there is a type of classification code to be given to the processing target document, at least a part thereof is taken out and processed If there is no classification code of the type to be given to the document, the classification code is not extracted from the extracted document.

  As a result, a group of documents to which the type of classification code to be given to the processing target document is extracted from the document database, and the classification code is extracted from the document group.

  Specifically, it is a case where both the document to which the sixth edition of the international patent classification is granted and the document to which the seventh edition is granted are stored in the document database. When assigning the classification code of the international patent classification 7th edition to the target document, the classification code of the international patent classification 7th edition of the document extracted from the document database is assigned using the keyword extracted from the processing target document as a key. Is selected, and the classification code assigned to the selected document is extracted.

  The control unit 11 obtains a list of classification codes for each keyword by this process, and stores each list in the storage unit 12 in association with the keyword. Specifically, as in the above example, each result of “teapot”, “near the opening”, “partition”, “each room”, “spout”, “teapot”, “tea bowl”, “pour” As shown in Fig. 4, a list containing "A23F 3/06", "A45C 11/20", etc. of the seventh edition of the international patent classification is stored for "teapot" as shown in Fig. 4 A list including “A11C 11/02” and the like is stored. This list is illustrated as including overlapping classification codes, but after sorting in a predetermined order such as alphabetical order, the process of removing duplicate lines (UNIX ( (Registered trademark) command, processing equivalent to the uniq command) may be performed. This makes it possible to generate a list excluding duplicates.

  Then, the control unit 11 registers the keyword obtained here and a list of classification codes related thereto in the dictionary database. As a result, a list of classification codes for keywords that have not been registered in the dictionary database can be registered in the dictionary database. That is, it can be said that the dictionary database is a cache of the result of the classification code list generation process of the control unit 11.

  Note that at least one of the classification codes included in the list of classification codes may be registered as an important code so as to be distinguishable from other classification codes. Specifically, before removing duplicates, the appearance frequency of each classification code is checked, and those whose appearance frequency is equal to or higher than a predetermined threshold are registered as important codes. Further, the control unit 11 may register a more specific classification code as an important code among the classification codes whose appearance frequency is equal to or higher than a predetermined threshold. Here, the specificity means a classification code that cannot be obtained from the result of searching the document database using another keyword as a key (even if it is obtained, the appearance frequency is less than a predetermined frequency).

  The control unit 11 refers to the results obtained by the search process and the classification code list generation process, and extracts the classification codes that exist between the keywords. Specifically, for each keyword, the control unit 11 generates a check table in which the column of the classification code obtained thereby is checked (FIG. 5). In FIG. 5, a symbol “◯” is shown in the checked column. After generating the check table, the control unit 11 next checks the number of checks for each classification code. For example, when there are five related keywords for a specific classification code, five checks are made in the column of the classification code on the check table.

  Then, the control unit 11 selects at least one classification code having the highest number of checks as a first assignment candidate classification code. Specifically, as in the above example, each result of “teapot”, “near the opening”, “partition”, “each room”, “spout”, “teapot”, “tea bowl”, “pour” If there is, the “A47G 19/22” is obtained with the classification code of the 7th edition of the international patent classification as corresponding to five of these seven keywords (excluding duplication).

  Here, the classification code having the largest number of checks is selected. However, for example, a condition that there are more checks than a predetermined ratio compared to the number of keywords, a predetermined threshold value, and the like. The first grant candidate classification code may be selected under conditions such as those with the above number of checks.

  For example, under the condition that there is a check of a ratio of 50% or more, 50% of 7 keywords, that is, "3.5" or more (however, since the number is always an integer, "4" or more In other words, “A47J 31/06” corresponding to the four keywords is obtained.

  Further, the control unit 11 selects a part of the keywords extracted from the processing target document as a central concept word, and selects a second grant candidate classification code based on the central concept word. Here, the central concept word is a keyword representing the content of the document to be processed, and is selected based on a predetermined condition from the keywords extracted by the keyword extraction process.

  Here, the predetermined condition is, for example, as follows. That is, the control unit 11 reads out a list of classification codes obtained by the search process and the classification code list generation process from the storage unit 12, and generates a logical sum of a set of classification codes included in the list. Thereby, a group of classification codes extracted in association with the extracted keyword is obtained. For each keyword, the control unit 11 calculates a ratio (classification code distribution ratio) that the number of classification codes extracted in association with the keyword occupies with respect to the number of classification codes included in the generated logical sum. For example, when the number of classification codes included in the logical sum is 100 and the number of classification codes obtained in association with the keyword “teapot” is 30, the classification code distribution ratio is 30%.

  The control unit 11 further examines the appearance frequency of each keyword from the extracted keyword group (the group before elimination of duplication). Specifically, as in the above example, each result of “teapot”, “near the opening”, “partition”, “each room”, “spout”, “teapot”, “tea bowl”, “pour” If there is, the frequency is “2” for “teapot” and the frequency is “1” for other keywords. Therefore, the control unit 11 selects a keyword of interest in the order of appearance frequency, and checks whether or not the classification code distribution ratio calculated for the keyword of interest is below a predetermined ratio threshold (for example, 40%). Here, if it is below the predetermined ratio threshold value, the attention keyword is selected as the central concept word. If the ratio threshold value is not below, the next keyword of interest is selected. Note that a keyword whose appearance frequency in the keyword group is smaller than the predetermined frequency threshold may not be selected as the keyword of interest.

  Here, “teapot”, which has the highest frequency, is the keyword of interest, and the classification code distribution ratio for the “teapot” is 30%, so that 30% is below the predetermined ratio threshold. For example, this “teapot” is selected as a central concept word.

  Note that the control unit 11 may not select the central concept word and not select the second assignment candidate classification code based on the central concept word if there is nothing that matches the condition. Moreover, the number of central concept words is not necessarily one.

  The control unit 11 calculates the appearance frequency of each classification code appearing on the classification code list from the classification code list obtained for the central concept word. Then, a classification code whose appearance frequency exceeds a predetermined appearance ratio threshold value is selected as the second assignment candidate classification code. For example, in the classification code list obtained for the central concept word “teapot”, there are 85 classification codes, of which 35 (37.5%) are “A47G 19/14” and 24 (28%) Is “A47J 31/06”, and so on, if the appearance ratio threshold is set to 35%, “A47G 19/14” is selected as the second grant candidate classification code.

  If the above flow is arranged, the control unit 11 performs an operation as shown in FIG. First, the control unit 11 extracts keywords from the processing target document (S1). Then, for each extracted keyword, a list of classification codes assigned in advance to the document in the document database related to each keyword is acquired (S2). In the search of this process S2, for keywords that are cached and registered with a classification code list related to the dictionary database, the classification code list is obtained with reference to the dictionary database. The document database is searched using the keyword as a key, and as a result of the search, at least part of the classification codes assigned to the documents included in the document group is extracted from the obtained document group, and the classification code list is obtained. .

  The control unit 11 checks how many keywords are associated with each classification code included in the list of classification codes acquired here. Then, for example, a classification code that is commonly associated with a predetermined number of keywords with respect to all keywords is extracted, and is used as a first assignment candidate classification code (corresponding to a list of classification codes for each keyword of the present invention). Select (S3).

  Next, the control unit 11 selects a central concept word from the keywords extracted in the process S1 (S4). The central concept word is selected according to the frequency of appearance of each keyword in the extracted keyword group and the distribution of classification codes extracted in relation to each keyword (related to each keyword in the logical sum of the classification code list). This is performed according to a predetermined rule based on the ratio of the extracted classification code.

  Then, from the classification code list related to the central concept word, based on the appearance frequency, for example, a classification code having an appearance frequency exceeding a predetermined appearance ratio threshold value corresponds to a second grant candidate classification code (corresponding to the second list of the present invention) ) Is selected (S5). In addition, in this process S5, if there is no classification code exceeding the appearance ratio threshold value, it is not always necessary to select the second addition candidate classification code.

  The control unit 11 further calculates a logical sum of the first grant candidate classification code selected in process S3 and the second grant candidate classification code selected in process S5, and stores the result as the grant candidate classification code. The data is stored in the unit 12 (S6), and the process ends. The control unit 11 displays the granted candidate classification code thus acquired on the display unit 15 and presents it to the user.

  In addition, the control unit 11 may associate classification code information including at least a part of the grant candidate classification code with the processing target document and store it in the document database of the storage unit 13. Specifically, the classification code selected and operated by the user as the classification code to be actually assigned from the displayed grant candidate classification codes is stored in the document database of the storage unit 13 in association with the processing target document. At this time, if the user inputs another classification code that is not included in the grant candidate classification code, the input classification code is also associated with the processing target document and stored in the document database of the storage unit 13. Also good.

  Next, dictionary database update processing by the control unit 11 will be described. When a new document is added to the document database, the control unit 11 may perform the following process as the dictionary database update process.

  That is, the control unit 11 generates a list of keywords already registered in the dictionary database. Then, each keyword included in the list is sequentially selected as a key, and the document database is searched using the selected key. Then, a document including the selected key is extracted from the document database. At least a part of the classification codes associated with the extracted document is taken out, a classification code list is generated, the keyword selected as a key is associated with the generated classification code list, and stored in the dictionary database. Store. The existing information stored in association with the keyword is overwritten. Thereby, the dictionary database is updated.

  Since the classification code processing apparatus of the present embodiment operates as described above, for example, even a scientific and technical literature including various technical terms uses a full-text search for an existing document and characterizes the processing target document. The classification process can be performed using the central concept word.

  In the description so far, for example, when a classification code of the seventh edition of the international patent classification is given to the processing target document, a classification code of the seventh edition of the international patent classification is given among the documents stored in the document database. In the above example, the candidate candidates are selected by referring to the classification code assigned to the extracted document. Good. That is, a correspondence table (so-called concordance table) is set between each type of classification code, and a document to which a classification code of International Patent Classification Sixth Edition is assigned is referred to this table. The existing classification code is converted into a seventh classification code. Then, the candidate classification code may be selected with reference to the converted classification code.

It is a block diagram showing the structural example of the classification code processing apparatus which concerns on embodiment of this invention. It is explanatory drawing showing the example of the content of a document database. It is explanatory drawing showing the example of the content of a dictionary database. It is explanatory drawing showing the example of the list | wrist of the classification code for every keyword. It is explanatory drawing showing an example of a check table. It is a flowchart figure showing the process example of the classification code processing apparatus which concerns on embodiment of this invention.

Explanation of symbols

11 control unit, 12 storage unit, 13 storage unit, 14 operation unit, 15 display unit.

Claims (5)

A classification code processing apparatus that is connected to a document database that holds a plurality of documents associated with a classification code and supports the assignment of a classification code to a document to be processed,
A keyword extracting means for extracting, as a keyword, a character string that is at least one key from a document to be processed;
A document group extracting means for extracting a document group related to each of the extracted keywords from the document database;
A list generation unit that refers to the document group extracted for each keyword, obtains at least a part of the classification code associated with the document group, and generates a list of classification codes for each keyword;
Means for counting the number of related keywords for each classification code from the list of classification codes for each keyword, and selecting a first grant candidate classification code based on this number;
Obtaining a logical sum of classification codes included in the classification code list for each keyword, and a ratio of the number of classification codes included in the classification code list for each keyword to the number of classification codes included in the logical sum It was calculated as the distribution ratio of the classification code for each keyword from among the keywords the keyword extracting means has extracted, based on the distribution ratio of the classification code for each keyword that the operation to get the keywords centered concept words, the keyword Means for selecting a second grant candidate classification code from the classification code list for each keyword based on the appearance frequency of the classification code related to the central concept word from the list of classification codes for each
Including
A classification code processing apparatus that performs a process of supporting the assignment of a classification code to the processing target document based on the first grant candidate classification code and the second grant candidate classification code . The classification code processing device according to claim 1,
The keyword extracting means includes
A classification code processing apparatus characterized by identifying a portion from which a keyword is extracted from the document to be processed and extracting a character string as a keyword from a part of the identified document. The classification code processing device according to claim 1 or 2 ,
A dictionary database that stores the keywords and the classification codes to be assigned to the documents related to the keywords in association with each other;
Search means for searching for a classification code held in the dictionary database in association with the keyword extracted by the keyword extraction means;
As a result of the search by the search means, when there is no classification code held in association with the keyword extracted by the keyword extraction means, the document group extraction means relates to each of the extracted keywords from the document database. Means for extracting the document group to be selected and selecting the first grant candidate classification code by referring to the document group extracted for each keyword and obtaining at least a part of the classification code associated with the document group A classification code processing device that selects a first assignment candidate classification code from among classification codes for each keyword. The classification code processing device according to any one of claims 1 to 3 ,
The document group extraction means, when at least one of a plurality of kinds of classification codes is selectively given to the document held in the document database, the kind of classification given to the document to be processed A classification code processing apparatus, wherein a group of documents to which codes are assigned is extracted from the document database. A program that is connected to a document database that holds a plurality of documents associated with a classification code, and that executes processing for supporting the assignment of a classification code to a document to be processed.
A keyword extracting means for extracting, as a keyword, a character string that is at least one key from a document to be processed;
A document group extracting means for extracting a document group related to each of the extracted keywords from the document database;
A list generation unit that refers to the document group extracted for each keyword, obtains at least a part of the classification code associated with the document group, and generates a list of classification codes for each keyword;
Means for counting the number of related keywords for each classification code from the list of classification codes for each keyword, and selecting a first grant candidate classification code based on this number;
Obtaining a logical sum of classification codes included in the classification code list for each keyword, and a ratio of the number of classification codes included in the classification code list for each keyword to the number of classification codes included in the logical sum It was calculated as the distribution ratio of the classification code for each keyword from among the keywords the keyword extracting means has extracted, based on the distribution ratio of the classification code for each keyword that the operation to get the keywords centered concept words, the keyword Means for selecting a second grant candidate classification code from the classification code list for each keyword based on the appearance frequency of the classification code related to the central concept word from the list of classification codes for each
Function as
Wherein the first grant candidate classification code, based on the second grant candidate classification code, program characterized by causing the processing to assist the application of classification code for the processing target document.

Images