JP2007026155A - Distributed development system of word information - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a distributed development system of word information capable of providing a flexible dictionary development environment. <P>SOLUTION: A unique word ID is generated for each word, and with it as a key, information individually required for each application accompanying the word such as a reading, a translation and word constitution is managed. Thus, the word can be registered even without filling the entire information of the word. Also, since the word information can be shared through a network, the base of a distributed user dictionary development environment which can be utilized in common by the plurality of applications is provided. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a language analysis system that performs language analysis on an input document, processes and outputs an analysis result for a predetermined purpose, and connects to the language analysis system via a network. The present invention relates to a word information distributed development system including word information servers that provide information.

  In a computer system, there is a user dictionary as a mechanism for arranging and using language analysis processing for processing input documents input by a user.

  In this user dictionary, the user usually adds vocabulary according to personal preferences and usage patterns.

  However, there is a limit to the development of a dictionary alone, and for example, there is a desire for both user B and user C to use the vocabulary added by user A.

  In order to deal with these problems, the dictionary format is widely disclosed, and text data and the like in accordance with the format are disclosed on the network. For example, the so-called pdic format.

Further, in Patent Document 1, the development of a dictionary is carried out by a membership system, and the development is motivated by giving privileges according to the degree of activity of each member.
JP 2003-173409 A

  However, these conventional techniques only consider sharing dictionary information for a specific application. For example, dictionary vocabulary addition for application X can be used by application Y with a slight addition of information. It is difficult to deal with a special case. As a result, it was difficult to attract a wide range of independent dictionary developers to one dictionary development environment.

  The present invention has been made in view of such circumstances, and an object thereof is to provide a distributed development system for word information that can provide a flexible dictionary development environment.

  The present invention relates to a language analysis system that performs language analysis on an input document, processes and outputs an analysis result for a predetermined purpose, and connects to the language analysis system via a network. In a distributed development system of word information comprising word information servers that provide information, the language analysis system is used for morpheme analysis processing means for inputting document information and performing morpheme analysis, and for each morpheme analysis processing means. And word dictionary means for holding at least word ID, notation, part of speech information, and a plurality of word associated information, and notation acquired for each word using the word ID string received from the morpheme analysis processing means, part of speech information, And output information generation means for generating output information using word-accompanying information, and registration, deletion, and update of words in the word dictionary. A dictionary editing unit that performs a search, sends a search request that combines notation, part-of-speech information, word ID, and word incidental information to the word information server, and information on words that match the condition Is obtained from the word information server, the word editing unit is called from the word editing server, the word ID obtaining unit is obtained from the word information server, and is called from the dictionary editing unit. A word information registration unit that sends notation, part-of-speech information, and word-accompanying information in association with a word ID to the server, and the word information server includes word ID, notation, part-of-speech information, plural A word database means for storing the word-accompanying information and a word ID that is newly activated in response to a request from the language analysis system In response to a request from the language analysis system, the word database means, the notation using the word ID as a key, part-of-speech information, registration, deletion and update of word-associated information, and word ID, notation, part-of-speech information And a word database management means for performing a search using a combination of word accompanying information as a search condition.

  The language analysis system has a word ID of a synonym of a word indicated by a word ID as the word-accompanying information.

  Further, the language analysis system has a word ID of a word structure of a word indicated by a word ID as the word accompanying information.

  In addition, the language analysis system calls a creator ID list holding unit that holds a list of predetermined creator IDs, and calls the word information acquisition unit every predetermined time, and uses the creator ID list to store the word information. The system further comprises automatic communication means for obtaining the result and sending the result to the dictionary editing means, and the dictionary editing means registers the word information received from the automatic updating means in the word dictionary.

  The word information server further sets a creator ID for each of the notation, part-of-speech information, and a plurality of word-accompanying information held as information for each word, and newly responds to a request from the language analysis system. Creator ID giving means for issuing a creator ID is further provided.

  Therefore, according to the present invention, a unique word ID is generated for each word, and this is used as a key to manage information that is attached to the word, such as a phonetic, a translation, and a word structure, and is individually required for each application. Therefore, it is possible to register words without filling in all the word information, and since word information can be shared via a network, a distributed user dictionary development environment that can be used in common by multiple applications The effect that the base of can be provided is obtained.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

  FIG. 1 shows an example of a network system according to an embodiment of the present invention. This network system stores and manages a plurality of user terminal devices TM1 to TMn, a word information server device ST for managing word information, a search server device SK for managing search information, and a document to be searched. A document database device SD is connected via a local area network LAN.

  Each of the user terminal devices TM1 to TMn receives a search client for requesting a search service from the search server device SK, and search target information (document information, search conditions, etc.) input by the user to the search client. The search server device SK includes a language analysis system for converting into information applied at the time of search, and the language analysis system acquires various information from the word information server device ST.

  FIG. 2 shows an example of the configuration of the language analysis system.

  In the figure, an input document buffer 1 stores a character string input by a user to a search client and outputs it to a morpheme analysis processing unit 2. The morpheme analysis processing unit 2 refers to the word dictionary unit 3 and divides the input document information into words constituting the document information and converts them into word ID strings. Here, a well-known morphological analysis method can be applied. For example, it is possible to apply a method that employs a word division result that minimizes the number of independent words included in the target character string. Since each word string obtained as a result of this word division has a word ID, the morphological analysis result (word string) can be replaced with a word ID string.

  The word dictionary unit 3 includes at least a word ID (a number uniquely assigned to each word), notation, part-of-speech information, and a plurality of word-accompanying information (information associated with the word such as a furigana and a corresponding translation) Is to hold.

  The output information generation unit 4 is for generating output information using the notation, part-of-speech information, and word associated information acquired for each word using the word ID string received from the morphological analysis processing unit 2. The output information is output to the search client via the output information buffer 5.

  The dictionary editing unit 6 is for registering, deleting, and updating words in the word dictionary unit 3, and the word ID acquisition unit 7 is called from the dictionary editing unit 6 and is connected via the local area network LAN. Thus, the word ID is obtained from the word information server device ST.

  The word information registration unit 8 is called from the dictionary editing unit 3 and transmits notation, part-of-speech information, and word associated information to the word information server device ST in association with the word ID via the local area network LAN. Is.

  The word information acquisition unit 9 is activated by the automatic update unit 10 and acquires word information in which creator IDs included in the specified creator ID list are registered from the word information server device ST via the local area network LAN. belongs to. Further, the word information acquisition unit 9 is for issuing a new word ID to the word information server device ST via the local area network LAN and acquiring the word ID obtained from the word information server device ST. .

  The automatic updating unit 10 acquires information on a word having a creator ID held in the creator ID list holding unit 11 by using the word information acquisition unit 9 at a predetermined time period. In addition, of the word information, word information that holds information on the word structure and synonyms is stored in the word dictionary unit 3.

  The creator ID list holding unit 11 is for holding a list of creator IDs uniquely assigned to each user, and the user can input any creator ID to the creator ID list holding unit 11 via the user interface. Can be saved.

  The communication control unit 12 is for the word ID acquisition unit 7, the word information registration unit 8, and the word information acquisition unit 9 to manage communication with the word information server device ST performed via the local area network LAN. It is.

  FIG. 3 shows an example of the configuration of the word information server device ST.

  In the figure, a word database unit 21 is for holding a word ID, notation, part-of-speech information, and a plurality of word-accompanying information as information for each word in a predetermined format. In response to a request from the language analysis system, the word database unit 21 is registered with the word ID as a key, part-of-speech information, and registration, deletion and update of word-accompanying information, and word ID, notation, part-of-speech information, In addition, the search is performed using a combination of word-accompanying information as a search condition.

  The word ID assigning unit 23 receives a word ID issue request through the network, creates a new word ID, and responds to the issue request source. For example, when a request is received, a word ID is generated by counting up from the initial value 1 in order.

  The creator ID assigning unit 24 receives a request for issuing a creator ID through the network, creates a new creator ID, and responds to the issue request source. For example, when a request is received, the creator ID is formed by counting up from the initial value 1 in order.

  The word database management unit manages data consisting of four sets of word IDs, creator IDs, property names, and property values as shown in Table 2, and functions for deleting / updating / retrieving database contents in the same way as normal database management I will provide a. This operation is the same as a general database.

  The communication control unit 26 is for the word database management unit 22, the word ID assigning unit 23, and the creator ID assigning unit 24 to manage communication with other terminals performed via the local area network LAN. is there.

  Registration of user information in the user information management unit 25 is performed by the creator ID adding unit 24 and can be performed by a command from an administrator user (not shown).

  FIG. 4 shows an example of the configuration of the word dictionary. This word dictionary is registered in the word dictionary unit 3 and the word database unit 21. Usually, only the information about the word registered by the user using the user terminal devices TM1 to TMn including the language analysis system is registered in the word registration unit 3, and the word database unit 21 of the word information server device ST. The word information registered by all users is registered.

  One word information registered in the word dictionary includes a word ID for identifying each word, a creator ID representing a user who has registered the word, a notation representing a character string constituting the word, and a part of speech of the word. If there is a synonym for the word, the synonym word ID that represents the word ID of the synonym, and if the word is a combination of notations of a plurality of other words, constitute the word It consists of word constituent word IDs in which word IDs of other words are arranged in the order of description.

  With the above configuration, when a user of any one of the user terminal devices TM1 to TMn searches for a document registered in the document database device SD using the search client of the own terminal, for example, the user indicates When the search condition is “pro-wrestling” and “Tokyo”, the search client passes “pro-wrestling” to the language analysis system, and obtains two word strings {pro, wrestling} and {pro-wrestling}. Also, “Tokyo” is passed to the language analysis system to obtain {Tokyo}.

  Here, as operators that can be used in the search server device SK, an AND operator, an OR operator, a NOT operator, and “?” That matches any one word or zero word can be used.

In this case, the search client uses the above word string,
((Pro ?? Wrestling) OR Pro Wrestling) A search expression for AND Tokyo is generated and passed to the search server device SK.

  As a result, the search server device SK searches the document database device SD based on the passed search expression, and hits a document such as “This is a professional wrestling in Tokyo Dome”.

  Here, in the language analysis system, when “pro-wrestling” is stored in the input document buffer 1, the morpheme analysis processing unit 2 performs the three word division results of “pro / sling”, “pro-wrestling / ring”, and “pro-wrestling”. And select “Pro Wrestling” from the viewpoint of minimizing the number of independent words. Accordingly, the word ID “1234567” (see FIG. 4) is passed to the output information generation unit 4.

  The output information generation unit 4 first generates possible synonym combinations as a sequence of a plurality of word IDs using the passed sequence of word IDs. If a word ID sequence {1234567, 1234567} obtained by analyzing “pro-wrestling pro-wrestling” is passed to the output information generation unit 4, {1234567, 1234567}, {1234567, 2345673, }, {2345673, 1234567}, {2345673, 2345673} are obtained respectively.

  Next, for each word, what is replaced with the word structure is replaced with the word structure. In this case, the above example is as follows. {1274567, 0006712, 1274567, 0006712}, {127567, 0006712, 2345673}, {2345673, 1274567, 0006712}, {2345673, 2345673}.

  Finally, the output information generation unit 4 acquires notation information using each word ID, and stores the acquired notation string in the output information buffer 5.

  An example of processing in this case is shown in FIG.

  First, the user moves the word W (start). . . W (end) and operator O (start). . . When a search condition consisting of O (end) is input to the search client (process 101), the search client uses the word W (start). . . W (i). . . Each W (end) is sent to the language analysis system.

  The language analysis system performs morphological analysis on individual words W (i), and generates a word string M (i, start). . M (i, j). . . M (i, end) is obtained (processing 102, 103). Further, M (i, j) is replaced with a plurality of notation groups SW (i, j, n, m) by the replacement process with the synonym notation group (process 104).

  The language analysis system sends this SW (i, j, n, m) back to the search client, and the search client uses this notation group to change W (i) constituting the search condition {(SW (i, j, start (start, start) ??? SW (i, j, start, 1) ???) OR (SW (i, j, 1, start) ??? SW (i, j, 1,1) ??? .) OR. . . (SW (i, j, end, start) ?? SW (i, j, end, end))} is rewritten (process 105).

  This replaces the proximity operation by concatenating the constituent words of the word W (i) with the character string “??”, and synonyms are formed by concatenating the synonyms with “OR” for each constituent word. The subordinate constituent words are also concatenated with “??”.

  The search client performs this rewriting on all the words (NO loop of decision 106), and then creates a search expression based on the rewriting result (process 107). The retrieval formula in this case is (RW (start) O (start) RW (1) O (1)... O (end) RW (end)).

  Then, the search expression is passed to the search server device SK (process 108), the search result is acquired from the search server device SK (process 109), and the search result is presented to the user (process 110).

  FIG. 6 shows an example of processing for expanding M (i, j) to SW (i, j, n, m).

  First, a word ID: X indicating M (i, j) is received (process 201), and a plurality of synonym IDs of X: Y (start). . . Y (end) is acquired from the word dictionary (process 202).

  Next, for each Y (n), the word structure SY (n, start). . . SY (n, end) is acquired from the word dictionary (NO loop of processes 203 and 204 and determination 205).

  The notation is acquired from the word dictionary using these acquired SY (n, m) (process 206), and this is used as SW (n, m) to respond to the upper process (process 207).

  In the present embodiment, the users of the user terminal devices TM1 to TMn can appropriately edit the word dictionary registered in the word dictionary unit 3. The editing result is reflected in the contents of the word database unit 21 of the word information server device ST.

  First, the operation of the dictionary editing unit 6 will be described by taking as an example the addition of new words, addition of word information, and automatic updating by the user.

  Suppose the user wants to add the word “ultimate wrestling”. The user operates the dictionary editing unit 6 to input the general noun “ultimate wrestling” and the word composition information “ultimate” “wrestling” as registered words.

  First, the dictionary editing unit 6 passes “ultimate wrestling”, “ultimate”, and “wrestling” to the word information acquisition unit 9. Thereby, the word information acquisition part 9 acquires each word information from a word information server apparatus. In this case, “ultimate wrestling—no corresponding word”, “wrestling-0006712”, and “ultimate-44445566” are obtained.

  Next, the dictionary editing unit 6 calls the word information acquisition unit 9 using the word IDs: 0006712, 4445566 in order, and presents this to the user. The user uses this to determine whether or not “Ultimate-44444566”, which is the division result of “Ultimate Wrestling”, is a correct constituent word. If it is correct, “OK” is determined. Send to department. If there is no next candidate, it will be handled as new word registration described later.

  When “ultimate” is determined, the same procedure is applied to “wrestling” to acquire all word configurations. Here, it is assumed that the above-mentioned word structure 444556-0006712 was correct.

  Next, the dictionary editing unit 6 further passes the word IDs 0006712 and 4445566 of the word configuration information to the word information acquisition unit 9 and searches for words that hold these as word configuration word IDs. Here, the two words described above are used. There is no word that holds ID as a word structure at the same time.

  As a result, the dictionary editing unit 6 determines that there is no word that may have a synonym such as “ultimate wrestling” or “wrestling ultimate”, and treats “ultimate wrestling” as a completely new word. Specifically, the word ID acquisition unit 7 is called to acquire a new word ID from the word information server device ST. Then, the obtained word ID (for example, 9934121) is assigned to the ultimate wrestling.

  Further, the dictionary editing unit 6 passes 9934121 and the notation “ultimate wrestling”, the part of speech “general noun”, and the word configuration “0006712, 4455566” to the word information registration unit, and the word information registration unit passes this to the word information on the network. Pass it to the server as a word registration request. The word information registration unit also manages a unique creator ID, and the initial value of this creator ID is “−1”.

  When the creator ID is “−1”, the word information registration unit 8 issues a creator ID issuance request to the word information server device ST, and replaces the initial value with the acquired creator ID. A creator ID is always attached to a word registration request. If the creator ID is “C11243”, the word registration request is as shown in FIG.

  If the user wants to add the word “Ultimate”, which is a registered word, and the word “Ultimate”, which is a registered word, the user wants to add “Ultimate” and synonym information “ Enter "Ultimate" as additional information.

  When the word information of both “ultimate” and “ultimate” can be acquired, the dictionary editing unit 6 issues a word information addition request as shown in FIG. In this case, it is assumed that different creators add the new creator ID.

  On the other hand, when the word dictionary unit 3 is automatically updated, the automatic update unit 10 uses the creator ID list holding unit 11.

  The user registers in advance in the creator ID list holding unit 11 the creator ID of the creator (user) who has the same field of interest as that of himself or the creator ID of the creator who creates reliable word information.

  The creator ID list held in the creator ID list holding unit 11 is simply a list of creator IDs, and has, for example, the contents {C11243, C17212, C00001, C00007}.

  At the time of automatic update, the automatic update unit 10 uses the word information acquisition unit 9 for word information having the creator ID “C11243”, “C17212”, “C00001”, or “C00007” held in the creator ID list holding unit 11. Then, from the word information server device ST, only word IDs having “word structure” and “synonyms” as property names are selected from the obtained word ID group, and the word information is edited as a search result. Send to part 3. It is assumed that the value “NULL (null value)” is set for words having no synonyms or word structures.

  In the dictionary editing unit 6, the word information is obtained in a state where all the information including the word ID is complete, and this is registered in the word dictionary unit 3. However, duplicate word information is not registered.

  As a result, in the system using the creator ID list described above, the vocabulary “ultimate wrestling” as shown in FIG. 7B and the synonym relationship between “ultimate” and “ultimate” can be used.

  In this way, the user can register word information as appropriate, or correct or update the registered word information as appropriate.

  In addition, a unique word ID is generated for each word, and this is used as a key to accompany the word such as phonetics, translations, word structure, and so on. Words can be registered without filling in all information, and word information can be shared via a network, thereby providing a base for a distributed user dictionary development environment that can be shared by a plurality of applications.

  In addition, it is possible to assign a unique word ID for each word and to manage synonym information by this, so that it is possible to specify an accurate word for the synonym.

  In addition, by assigning a unique word ID for each word and enabling the search of word configuration information thereby, it is possible to specify an exact word for a synonym.

  In addition, by making it possible to handle creator IDs, it is possible to specify a group of creators that are biased to a specific field or have a certain level of skill, thereby obtaining word information that reflects user preferences and requests Can be made possible.

  Further, the word dictionary can be automatically updated by maintaining a list of creator IDs and enabling update using the list.

  It should be noted that the present invention can be similarly applied to a search system and a translation system.

The block diagram which showed an example of the network system concerning one Example of this invention. The block diagram which showed an example of the structure of the language analysis system. The block diagram which showed an example of the structure of word information server apparatus ST. Schematic which showed an example of the structure of a word dictionary. The flowchart which showed an example of the process of a user terminal device. The flowchart which showed an example of the process which expand | deploys M (i, j) to SW (i, j, n, m). Schematic which showed an example of update information.

Explanation of symbols

TM1 to TMn User terminal device ST Word server device SK Search server device SD Document database device 1 Input document buffer 2 Morphological analysis processing unit 3 Word dictionary unit 4 Output information generation unit 5 Output information buffer 6 Dictionary editing unit 7 Word ID acquisition unit 8 Word information registration part 9 Word information acquisition part 10 Automatic update part 11 Creator ID list holding part 12, 26 Communication control part 21 Word database part 22 Word database management part 23 Word ID provision part 24 Creator ID provision part

Claims (5)

A language analysis system that analyzes the input document, processes and outputs the analysis result for a predetermined purpose, and connects to the language analysis system via a network to provide word information to the language analysis system In the distributed development system of word information consisting of word information servers,
The language analysis system includes:
Morphological analysis processing means for inputting document information and performing morphological analysis;
Word dictionary means that is used in the morphological analysis processing means and holds at least a word ID, a notation, a part-of-speech information, and a plurality of word-accompanying information for each word;
Output information generating means for generating output information using notation, part-of-speech information acquired for each word using the word ID string received from the morphological analysis processing means, and word accompanying information;
A dictionary editing means for registering, deleting, and updating words in the word dictionary;
It is called from the dictionary editing means, sends a search request that combines notation, part-of-speech information, word ID, and word incidental information to the word information server, and information on the word that matches the condition is sent from the word information server. Word information acquisition means to acquire;
A word ID acquisition unit that is called from the dictionary editing unit and acquires a word ID from the word information server;
A word information registration unit that is called from the dictionary editing unit and sends the notation, part-of-speech information, and word incidental information in association with the word ID to the word information server;
The word information server
Word database means for holding word ID, notation, part-of-speech information, and a plurality of word-accompanying information as information for each word;
Word ID giving means for newly issuing a word ID in response to a request from the language analysis system;
In response to a request from the language analysis system, to the word database means, notation using word ID as a key, part of speech information, registration of word incidental information, deletion and update, and word ID, notation, part of speech information, and A word information distributed development system comprising word database management means for performing a search using a combination of word accompanying information as a search condition.   2. The distributed development system of word information according to claim 1, wherein the language analysis system has a word ID of a synonym of a word indicated by a word ID as the word accompanying information. 2. The distributed development system of word information according to claim 1, wherein the language analysis system has a word ID of a word structure of a word indicated by a word ID as the word accompanying information. The language analysis system includes:
Creator ID list holding means for holding a list of predetermined creator IDs;
Calling the word information acquisition means at predetermined time intervals, obtaining word information using a creator ID list, further comprising an automatic communication means for sending the result to the dictionary editing means,
4. The distributed development system for word information according to claim 1, wherein said dictionary editing means registers word information received from said automatic updating means in a word dictionary. The word information server
Further, a creator ID is set for each of the notation, part-of-speech information, and plural word-accompanying information held as information for each word,
5. The word information according to claim 1, further comprising creator ID giving means for newly issuing a creator ID in response to a request from the language analysis system. Distributed development system.

Images