JP2002091965A - Dictionary device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a dictionary device capable of enriching a dictionary, its structure, etc. SOLUTION: This dictionary device is provided in a natural language processor utilized by a plurality of users. The dictionary device has a dictionary main body that hierarchizes a plurality of technical term dictionaries of each category with a tree structure with a termini generales dictionary as a root node, a user dictionary registering means for setting a user dictionary while making the user dictionary correspond to an optional technical term dictionary, and an application dictionary deciding means for deciding the whole technical term dictionaries existing on the path of the tree structure from the technical term dictionary of a natural language processing object up to the termini generales dictionary and the whole user dictionaries of a processing request user, which is made to correspond to the technical term dictionaries as dictionaries to be applied, when the category is indicated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a dictionary device used in a natural language processing device. For example, the present invention relates to a machine translation device provided on a server of a network accessible by an unspecified number of users such as the Internet. It can be applied to dictionaries.

[0002]

2. Description of the Related Art With the spread of the Internet, overseas information can be easily accessed and information can be easily transmitted overseas. Therefore, there has been an increasing demand for machine translation of a foreign language document into Japanese or machine translation of a Japanese document into a foreign language.

In view of such demands, research and development of mounting a machine translation apparatus on a server of a network accessible by an unspecified number of users such as the Internet has been conducted.

In general, a machine translation device is provided with a general term dictionary storing words widely used in any field and a user dictionary in which a user arbitrarily registers words and the like. I have. Further, there is also a machine translation device having a technical term dictionary storing words widely used in a specific specialty field.

[0005] Since a machine translation device provided on a server on the Internet targets various documents, there are three types of dictionaries used for such a device: a general term dictionary, a user dictionary, and a technical term dictionary. It becomes a dictionary structure, and the weight of the technical term dictionary will be particularly large when the translation quality is considered.

[0006] As a method for obtaining a good translation result as a translation result of a document obtained from the Internet without taking as much time as possible, there is a method described in Japanese Patent Application Laid-Open No. Hei 10-21222.

According to the method described in this publication, a technical field is specified from document identification information such as a URL used when a document (Web page) to be translated is obtained, and a technical term dictionary is automatically set. Thus, it is intended to obtain a high-quality translation result without forcing the user to make complicated settings.

[0008]

However, since the document identification information is for uniquely representing the document and does not need to be a character representing the field or genre, the document identification information uses the field or genre of the document. Difficult to identify. In particular, since fields and genres of documents available from the Internet are various, it is difficult to specify all fields and genres.

Even if it can be specified, since the fields of documents available from the Internet are of various genres, a translation result that can be correctly understood cannot be obtained without a wide variety of technical term dictionaries.

[0010] It is practically difficult for a developer or a provider of a machine translation apparatus to prepare various technical term dictionaries.

For this reason, it is also conceivable to prepare a small number of technical terms in a large category as a technical term dictionary, and leave a part which cannot be handled by the technical term dictionary to a user dictionary.

However, in the case of a machine translator provided on a server on the Internet, the number of users is large, and if there are many parts left to a user dictionary, the storage capacity of the server becomes enormous. Also, many users may register the same word, which is wasteful as a system.

Further, it is conceivable that the technical term dictionary reflects the knowledge of a user who is rich in the field and genre. However, the contents registered and edited by a certain user may be unsatisfactory to other users. In an extreme case, there is a high possibility that the user will perform improper registration due to mischief or the like.

Therefore, there is a need for a dictionary device that can obtain good processing results in a natural language processing device used by many users.

[0015]

A dictionary device according to a first aspect of the present invention is a dictionary device provided in a natural language processing device used by a plurality of users. The dictionary device includes a plurality of specialized fields and genres. A dictionary body in which a technical term dictionary is hierarchized in a tree structure with a general term dictionary as a root node; and a user dictionary registering means for setting a user dictionary in association with any of the technical term dictionaries. I do.

A dictionary device according to a second aspect of the present invention is a dictionary device provided in a natural language processing device used by a plurality of users, and includes a user dictionary for each user that can be edited by a user, and a common dictionary common to all users. When the same dictionary information as the dictionary is stored in a number of the user dictionaries satisfying a predetermined condition,
A common dictionary enriching means for registering the dictionary information in the common dictionary.

According to a third aspect of the present invention, there is provided a dictionary device having an upper-level dictionary in a hierarchical structure, a plurality of lower-level dictionaries positioned below the upper-level dictionary, and a number of the same dictionary information satisfying a predetermined condition. A dictionary information integrating unit that registers the dictionary information in the upper dictionary when stored in the lower dictionary.

The fourth dictionary apparatus of the present invention is a dictionary apparatus provided in a natural language processing apparatus used by a plurality of users. The dictionary apparatus has a user dictionary that can be edited by each user and a common dictionary common to all users. It is characterized by having a dictionary and dictionary dividing / generating means for setting a lower dictionary below the common dictionary based on at least the stored contents of each user dictionary.

[0019]

DESCRIPTION OF THE PREFERRED EMBODIMENTS (A) First Embodiment A first embodiment in which a dictionary device according to the present invention is applied to a machine translation function provided on an Internet server will be described below with reference to the drawings. .

(A-1) Configuration of First Embodiment FIG. 2 is a block diagram showing a configuration of a machine translation network system to which the first embodiment is applied.

In FIG. 2, the machine translation network system 1 has a configuration in which a server 3 on the Internet 2 can connect to a user terminal 4 via the Internet 2. Although not shown in FIG. 2, the server 3 can be connected to another server or the like via the Internet 2.

The server 3 includes an HTTP daemon 10, a log analysis unit 11, an access log storage unit 12, and a Web server 1.
3. It has a machine translation system 14, a dictionary database 15, a dictionary converter 16, an HTML parser 17, and an input / output device 18.

Here, the Web server 13 functionally includes a communication tool 13a, a Web translation processing unit 13b, a dictionary editing unit 13c, a user registration / authentication unit 13
d and a community management unit 13e. The machine translation system 14 includes a translation engine 14a and a dictionary unit 1
4b. Further, the dictionary database 15 has a dictionary information section 15a, a user information section 15b, and a community information section 15c.

The user terminal 4 instructs the Internet 2 to search for a home page or a Web page (hereinafter, these are collectively referred to as a Web page) 5. Here, the user who has concluded the translation service provision contract with the operator of the server 3 is the retrieved We
Using the user terminal 4, the server 3 can be instructed to translate the page b into a predetermined language and transmit it. For example, the user can input a translation instruction input field on a homepage provided by the server 3, introduce a code for instructing translation into document identification information that specifies a Web page given to the server 3, or translate as link destination information. By designating the resulting document, it is possible to instruct that the searched Web page be translated into a predetermined language and transmitted.

In the server 3, the HTTP daemon 10
Is to transfer the searched Web page (HTML document) according to a predetermined transfer protocol.

The log analysis unit 11 obtains an access log from a Web page to be searched from the user terminal 4 and communication information defining the user terminal 4 and stores the access log in the access log storage unit 12. Log in and log out. When logging in, user authentication or the like is required.

The communication tool 13a in the Web server 13 has a communication function for searching for a Web page, a communication function with the user terminal 4, and the like.

The other functional units in the Web server 13, that is, the Web translation processing unit 13b, the dictionary editing unit 13c,
User registration / authentication unit 13d and community management unit 13
“e” is a functional unit related to translation of a Web page.

The Web translation processing unit 13b sends the retrieved W
HTML parser 17 when eb page needs translation
The retrieved web page is provided to the machine translation system 14 via the web browser and the translation of the web page is executed.

Here, the HTML parser 17 takes out only a portion (sentence portion) of the retrieved Web page (HTML document) which needs to be translated, based on the tag information and the like, and gives it to the machine translation system 14. The tag information is returned to the translation result, and is returned to a Web page (HTML document). Further, in the machine translation system 14, the translation engine 14a is a part that executes machine translation, and the dictionary unit 14b stores dictionary information used when the translation engine 14a performs machine translation. The dictionary information stored in the dictionary information unit 15a described later is converted by the dictionary converter 16 for the translation engine 14a and stored in the dictionary unit 14b.

Note that the translation activation for the retrieved Web page (HTML document) and the method of outputting the translated version are described in Japanese Patent Application Laid-Open No. 7-2027 by the same applicant.
No. 21 and Japanese Patent Application Laid-Open No. 7-202734 can be applied.

The first embodiment and each embodiment described below relate to the dictionary editing unit 13c, the user registration / authentication unit 13d, the community management unit 13e, the dictionary database 15, and the input / output device 18 of the Web server 13. It has features in the part to be performed.

The dictionary information section 15a in the dictionary database 15 stores various types of dictionary information.
The dictionary configuration of the dictionary information section 15a is a general term dictionary, a technical term dictionary, and a user dictionary.

In general, hierarchies of application at the time of translation are made in the order of user dictionary, technical term dictionary, and general term dictionary in descending order of priority. In the first embodiment, Hierarchy as shown in FIG. 1 is performed, and the first embodiment has a great feature in this hierarchization.

In FIG. 1, the dictionary configuration in the first embodiment basically has a hierarchical structure of a tree structure.

The general term dictionary D0 is used as a root node, and the next hierarchical level is provided with specialized term dictionaries D11 to D1x in a case where a specialized field or genre is largely classified. A technical term dictionary for further classifying specialized fields and genres of the higher layer is provided in the lower layers, and the number of branches to the terminal node (depth of the layer)
Need not be uniform.

In the example of FIG. 1, a computer technical term dictionary D111 and a computer software technical term dictionary D112 are provided at a lower level of the computer related term dictionary D11. On the other hand, a Japanese technical term dictionary D1x1, a Chinese technical term dictionary D1x2, and a Western technical term dictionary D1x3 are provided at a lower level of the cooking-related technical term dictionary D1x. Culinary Terminology Dictionary D
A 1x31 and an Italian cuisine technical term dictionary D1x32 are provided.

Note that only one technical term dictionary immediately below a technical term dictionary may exist. For example, only a golf dictionary may be provided as a dictionary in a lower hierarchy of the sports dictionary.

In the case of the first embodiment, the user dictionary also has a feature. Each user can provide a user dictionary in association with an arbitrary technical term dictionary, and the number of technical term dictionaries providing user dictionaries is not limited. Of course, the user registers words in the field or genre related to the associated technical term dictionary in the user dictionary.

In the example of FIG. 1, user A associates user dictionaries UA11, U11 with computer technical term dictionary D11 and computer hardware technical term dictionary D111.
A111 is provided.

It should be noted that a user dictionary can be provided in association with the general term dictionary D0 for user registration of words not related to the field or genre.

In the following, the tree structure portion (D11 to D1x3) of the technical term dictionary including the associated user dictionary will be described.
In the case of 2), as will be apparent from other embodiments described later, since the user's knowledge is reflected in the technical term dictionary, it will be referred to as a community dictionary below.

User information section 15 of dictionary database 15
In b, information on a user who has made a contract with the operator of the server 3 to use the server 3 is stored. Also,
The user information section 15b indicates whether the registered user can receive the machine translation service,
Also, information such as whether or not a user dictionary for the user has been set is stored.

The community information section 15c of the dictionary database 15 stores information on the structure related to the community dictionary portion in the dictionary structure shown in FIG.

The dictionary editing unit 13c of the Web server 13
This is for editing the dictionary information section 15a.

User registration / authentication unit 1 of Web server 13
3d performs user registration, authenticates whether a person who accesses the server 3 is a legitimate user, confirms whether a machine translation service can be provided when a machine translation service is requested, and checks a user dictionary. For example, to confirm whether the person can execute the operation.

The community management unit 1 of the Web server 13
3e manages information of the community information section 15c. For example, when the field or genre of the Web page to be provided for translation is determined, the community information unit 15c
The dictionary used in the machine translation is determined based on the above information.

More specifically, if the field and genre of the Web page for which a translation request is made by the user A is computer hardware, the user dictionary UA111, the computer hardware technical term dictionary D111, the user dictionary UA11, and the computer The terminology dictionary D11 is determined as a dictionary to be used, and the application priority is set to be higher in the order described above (the general term dictionary D0 is always a target).

That is, all the technical term dictionaries on the path of the tree structure from the technical term dictionary relating to the field and genre of the Web page to the general term dictionary, and all the user dictionaries associated with the technical term dictionaries are obtained. Decide which dictionary to use.

The input / output device 18 is used when the operator of the server 3 starts editing a dictionary or edits a dictionary.

The machine translation network system 1 to which the first embodiment is applied is capable of responding to simultaneous translation requests from a plurality of users. For example, one set of the machine translation system 14 and the HTML parser 17 may be used in a time-division manner to respond to a plurality of translation requests, and the machine translation system 14 and the HTML parser 1 may be used.
7 may be prepared in advance to respond to a plurality of translation requests. In the latter case, an original machine translation program or the like may be loaded into a main memory or the like of a plurality of CPUs prepared each time a translation is requested, so that a plurality of translation requests can be simultaneously handled.

When a set such as the machine translation system 14 or the HTML parser 17 is set for each user who has requested translation, the dictionary unit 14b of the machine translation system 14
Only dictionary information determined to be used according to the field and genre of page b may be set from the dictionary database 15.

(A-2) Operation of First Embodiment (A-2-1) User Dictionary Registration Next, the operation of the server 3 when the user registers a word in the user dictionary will be described with reference to the flowchart of FIG. This will be described with reference to FIG. At the time of this processing, the information exchanged between the user terminal 4 and the server 3 is information in the form of an HTML document.

For example, the user uses the user terminal 4 to
A predetermined Web page provided by the server 3 is displayed, and a word user registration mode is instructed.

At this time, the server 3 starts the user dictionary registration process shown in FIG.
The authentication unit 13d) determines whether or not the user is a person who can perform word user registration (step S1).

If the user is not a person who can execute the user registration of the word, the server 3 returns that fact and terminates the user dictionary registration processing (step S2).

On the other hand, if the user is a person who can execute user registration of words, the server 3 (community management unit 13
e) extracts the tree structure of the general term dictionary and the technical term dictionary from the community information section 13e, and displays a user dictionary registration input screen including the information of the tree structure on the user terminal 4
(Step S3), and waits for registered word information from the user terminal 4 (step S4).

When the user dictionary registration input screen is displayed, the user inputs word information to be registered, and based on the displayed general term dictionary and technical term dictionary tree structure, the registered word information is displayed. And the information of the dictionary to be associated is transmitted to the server 3. Note that a plurality of registered word information may be transmitted together. In the following, it is assumed that one piece of registered word information is transmitted for simplification of the description.

Server 3 (user registration / authentication unit 13d)
Given the registered word information, based on the general term dictionary or technical term dictionary included therein, and referring to the storage information in the user information section 15b and the community information section 15c, It is determined whether a user dictionary (user dictionary associated with a general term dictionary or a technical term dictionary) related to the registered word information has already been set (step S5).

If no user dictionary has been set,
After setting such a user dictionary (step S6),
If it has already been set, the registered word information provided from the user terminal 4 is immediately registered (step S7), and the series of user dictionary registration processing ends.

Here, the dictionary information section 15a may have a storage area which is clearly divided for each dictionary type. However, considering many user dictionaries, the dictionary information includes a dictionary type identification code. In this case, setting of the user dictionary means assignment of an identification code.

(A-2-2) Machine Translation Processing Next, machine translation processing of a Web page will be described with reference to the flowchart of FIG.

The server 3 (Web translation processing unit 13b)
When the need to translate the Web page arises, the machine translation process shown in FIG. 4 is started.

When the necessity of translating a web page arises, for example, (1) when the user instructs to translate and deliver the retrieved web page, (2) the user displays the web page once displayed (3) When the user wants a Web page in another language to be transmitted via the Internet, for example.

The server 3 (Web translation processing unit 13b)
When the machine translation process shown in FIG. 4 is started, first, the machine translation system 14 and the like to be used for the current machine translation are allocated (including determination of a time slot in time division; step S10). The field and genre of the Web page are recognized (step S11).

Here, the user may specify the field or genre of the Web page from the user terminal 4,
Further, the server 3 (Web translation processing unit 13b) may automatically recognize the field or genre of the Web page. As a method of automatically recognizing a field or genre, not only the method described in the gazette described in the section of the related art can be applied, but also one or a plurality of keywords defining the field or genre can be specified in advance for each field or genre. Various existing methods can be applied, such as a method that is determined and determined from the number of appearances of keywords on a Web page. When a plurality of candidates for the recognition field and the genre occur, for example, the lowest rank in the community dictionary is set.

As described above, when recognizing the field and genre of the Web page to be translated, the server 3 (Web translation processing unit 13b) determines the dictionary type to be used in the machine translation, and the dictionary of that type is used. A state is set (step S12). As described above, all the technical term dictionaries on the path of the tree structure of the community dictionary from the technical term dictionary relating to the field and genre of the Web page to the general term dictionary, and the user associated with the technical term dictionary If all of the dictionaries are determined to be used, and the information of the determined dictionaries is transferred from the dictionary information unit 13a to the dictionary unit 14b, or if all types of dictionary information are set in the dictionary unit 14b, , So that only the information of the determined type of dictionary can be accessed.

Thereafter, the server 3 (HTML parser 17)
Separates only a translatable sentence portion from the Web page (step S13). Then, the server 3 (the translation engine 14a) executes the translation (step S14), and the server 3 (the HTML parser 17) returns the tag and the like to the obtained translation result (step S15), and the translation result W
The eb page is obtained, and a series of machine translation processing ends.

When a web page created by a registered user of the server 3 is searched for translation by another person (translation at the time of transmitting the web page), a user dictionary relating to the registered user is used for translation. You may do it.

(A-3) Effects of the First Embodiment As described above, according to the first embodiment, a plurality of types of specialty with different classification sizes such as large classification, middle classification,. The term dictionary is prepared in a hierarchical structure with a tree structure with the general term dictionary as the top node, and a user dictionary can be set in association with each technical term dictionary. Depending on the classification and genre of the document to be translated, the field, Machine translation is performed by using all the technical term dictionaries on the tree-structured path from the technical term dictionary relating to the genre to the general term dictionary and all the user dictionaries associated with those technical term dictionaries. Therefore, regardless of the classification and genre of the translation target document, a translation result with good translation quality can be obtained without taking much time on average.

That is, if the specialty or the specialty of the genre is high, the technical term dictionary of the lower hierarchy corresponding thereto and the information of the user dictionary associated therewith are used for translation, and the translation quality can be improved. Conversely, if the specialty or genre is low, only the information in the higher-level technical term dictionary and the user dictionary associated with it is used for translation. It is possible to obtain a translation result of a high quality.

There are a wide variety of documents available on the Internet, ranging from those with a high degree of specialty and genre to low-specificity. It works effectively.

(B) Second Embodiment Next, a second embodiment in which a dictionary device according to the present invention is applied to a machine translation function provided on a server on the Internet will be described with reference to the drawings.

(B-1) Configuration of the Second Embodiment A machine translation network system to which the second embodiment is applied can also be represented by the block diagram of FIG. 2 according to the first embodiment. .

However, the functional configuration of the second embodiment will be described below with reference to FIG. Note that FIG.
FIG. 9 is a functional block diagram showing functions arranged in the above manner and incorporating characteristic functions in the second embodiment.

In FIG. 5, a machine translation network system 21 according to the second embodiment
The above search / translation server 23 is connected to the browsing / input means 24 via the Internet 22 and is constructed.

The browsing / input means 24 corresponds to the user terminal 4 according to the first embodiment. The browsing / input means 24 requests the Internet 22 to search for a document, request translation of a search document, search for a search document, It is used to display a document, a translated document obtained by translating the document, request a registration in a user dictionary, and the like.

The search / translation server 23 searches for a document,
It executes machine translation of a document, and has a communication control unit 31, a machine translation unit 32, a dictionary management unit 33, a dictionary database 34, a technical term dictionary enrichment unit 35, and the like.

The communication control means 31 controls communication with an external Internet element (document storage means), not shown, and with the browsing / input means. Communication with an external Internet element (document storage means) is communication for extracting a predetermined document.

The machine translation means 32 controls which dictionary is used when machine translation is required for the search document, and translates the search document into another language.

The dictionary management means 33 stores the dictionary database 3
4 for creating a dictionary, extracting and editing dictionary word information, and the like. For example, the dictionary management unit 33 extracts word information of the dictionary specified by the machine translation unit 32 from the dictionary database 34 and gives the word information to the machine translation unit 32. Also, for example, the dictionary management unit 33 takes out the word information of the dictionary requested from the technical term dictionary enriching unit 35 from the dictionary database 34 and gives the word information, and determines that the word information is to be added to a certain dictionary. Is added to the corresponding dictionary in the dictionary database 34.

The dictionary database 34 is a database that stores various dictionaries. Also in the second embodiment, the plurality of dictionaries stored in the dictionary database 34 are the same as those in the tree structure illustrated in FIG. 1 described in the first embodiment. In other words, a general term dictionary is used as a root node, and a technical term dictionary for a case where a specialty field or genre is roughly classified is provided in the next layer, and a special field or genre of a higher layer is provided in a lower layer. Is provided, and the number of branches (depth of the hierarchy) reaching the terminal node is not uniform.
A user dictionary is provided in association with each technical term dictionary.

In FIG. 5, in order to make it easier to understand the features of the second embodiment, a certain technical term dictionary Dm
And a plurality of user dictionaries Dm1 to DmN associated with this technical term dictionary Dm.

The technical term dictionary enriching means 35 automatically extracts, from the word information of the user dictionaries Dm1 to DmN, information that is considered better to be registered in the technical term dictionary Dm and registers the information in the technical term dictionary Dm. Things.

Here, the processing of the technical term dictionary enrichment means 35 may be periodically started, for example, once a day (for example, at 2:00 am). May be started using an input / output device (not shown) (see input / output device 18 in FIG. 2). Furthermore, every time a word is registered in any of the user dictionaries,
It may be activated.

(B-2) Operation of the Second Embodiment Next, the operation of the technical term dictionary enriching means 35 constituting the feature of the second embodiment will be described with reference to the flowchart of FIG. .

FIG. 6 shows a case where the program is started periodically.
Or, it corresponds to the case where the server is started by the server operator. FIG. 6 shows a process when a certain technical term dictionary is targeted, and the process shown in FIG. 6 is executed for all the technical term dictionaries.

When the processing shown in FIG. 6 is started, the technical term dictionary enriching means 35 extracts information of all the user dictionaries from the dictionary database 34 while summing up the same word information for each related technical term dictionary. Buffering (step S31).

FIG. 7 shows an example in which data is taken out while counting. From the left, information (dictionary data ID) numbered for each different dictionary data, headwords, translations, and registered numbers are shown. It is assumed that the dictionary data IDs are sequentially numbered from 0.

When the retrieval while summing up is completed, the technical term dictionary enriching means 35 sets the dictionary data ID to 0 (initial value) (step S32).

Then, the processing loop of steps S33 to S37 is repeated until the enrichment processing is completed for all dictionary data.

The technical term dictionary enriching means 35 determines whether or not there is word information whose registered number is a majority of the number of user dictionaries associated with the technical term dictionary (steps S33, S34). ).

If there is at least one such word information, all the word information is additionally registered in the technical term dictionary (step S35). In this case, it is confirmed that the word information does not exist in the technical term dictionary and registered.

If the additional registration of such word information in the terminology dictionary is completed, or if there is no word information whose registration number is a majority of the number of user dictionaries, the dictionary data ID is incremented by one. (Step S36), it is determined whether or not unprocessed dictionary data remains (Step S37).

If unprocessed dictionary data remains, the process returns to step S33, and if the enrichment process has been completed for all dictionary data, a series of processes shown in FIG. 6 is completed.

In the example of FIG. 7, if the number of user dictionaries associated with the technical term dictionary is five, the word information “pencil; pencil” has three user dictionary registrations, which is more than a majority. Through the processing in FIG. 6, the information is registered in the target technical term dictionary.

In the above description, the condition for registration in the technical term dictionary is set to a majority or more of the number of user dictionaries. However, a registration requirement using another ratio as a threshold may be used. Further, the threshold value that defines the registration requirement may be different depending on the number of user dictionaries.

In the above description, the word information remains in the user dictionary even after it is additionally registered in the technical term dictionary. However, the word information additionally registered in the technical term dictionary is deleted from the user dictionary. May be.

Further, in the above description, the registration in the technical term dictionary is determined based on the registration status in all the user dictionaries associated with the technical term dictionary. Whether to register in the technical term dictionary may be determined based on the registration status. When provided in a server on the Internet, the number of user dictionaries may be large, and in order to simplify the processing, the registration status of some (for example, 100) user dictionaries is changed to a technical term dictionary. May be determined. Here, as a part of the user dictionaries, a predetermined number of user dictionaries may be determined from a user having a large number of registered users, or a user dictionary having a predetermined number or more of word information registered by the user may be used. A plurality of user dictionaries may be determined using random numbers or the like, or may be determined in combination.

Further, in the above description, the registration in the technical term dictionary is completely automated. However, after the server operator or the like displays and confirms the word information to be additionally registered in the technical term dictionary, You may make it register additionally in a technical term dictionary.

If a user dictionary exists in association with the general term dictionary, the general term dictionary may be expanded in the same manner as the above processing.

(B-3) Effect of Second Embodiment As described above, according to the second embodiment, technical terms used by many users are automatically imported into the technical term dictionary. Therefore, the technical term dictionary can be enriched and the translation quality can be improved. In particular, those who have not performed user registration benefit greatly from the technical term enrichment function of the second embodiment.

Also, from the viewpoint of the system operator (server operator), it is enormous in time, man-hour, and cost to create a perfect technical term dictionary from the beginning. As a prerequisite, the perfection of the technical term dictionary at the start of the system is relaxed, and the dictionary creation time, man-hour, and cost can be reduced to some extent.

(C) Third Embodiment Next, a third embodiment in which the dictionary device according to the present invention is applied to a machine translation function provided on a server on the Internet will be described with reference to the drawings.

(C-1) Configuration of Third Embodiment A machine translation network system to which the third embodiment is applied can also be represented by the block diagram of FIG. 2 according to the first embodiment. .

However, the functional configuration of the third embodiment will be described below with reference to FIG. Note that FIG.
FIG. 14 is a functional block diagram showing functions arranged in the above manner and incorporating characteristic functions in the third embodiment.
The same and corresponding parts as those in FIG. 5 according to the embodiment are denoted by the same reference numerals.

Referring to FIG. 8, the machine translation network system 21A according to the third embodiment also has the Internet 2
Search and translation server 23A on the Internet 22
And is connected to the browsing / input means 24 via the.

The third embodiment is slightly different from the second embodiment in the functional configuration of the search / translation server 23A. That is, in the third embodiment, a dictionary information integration unit 36 is provided instead of the technical term enrichment unit 27 of the second embodiment.

Although the dictionary database 34 is the same as that of the second embodiment, FIG. 8 clarifies the hierarchies of the technical term dictionary in relation to the features of the third embodiment. ing.

When the same word information is present in a plurality of technical term dictionaries immediately below a certain technical term dictionary, the dictionary information integrating means 36 constituting the feature of the third embodiment ranks the word information higher. Is registered in the terminology dictionary. For example, when the same word information is registered in the political dictionary Dn1 and the economic dictionary Dn2 in the notation of the dictionary database 34 in FIG.
Is registered in the word information.

Here, the processing of the dictionary information integrating means 36 is also
It may be activated periodically, for example, once a day (for example, at 2:00 am).
The input / output device not shown by the operator (input / output device 1 in FIG. 2)
8)). Further, as in the second embodiment, each time an additional registration is made in the technical term dictionary, the program may be started.

(C-2) Operation of the Third Embodiment Next, the operation of the dictionary information integrating means 36 constituting a feature of the third embodiment will be described with reference to the flowchart of FIG.

FIG. 9 shows a case where the program is started periodically.
Or, it corresponds to the case where the server is started by the server operator. FIG. 9 shows a process in the case where a certain technical term dictionary is targeted, and the process shown in FIG. 9 is executed for all the technical term dictionaries.

When the processing shown in FIG. 9 is started, the dictionary information integrating means 36 first extracts word information registered in all technical term dictionaries below the subject technical term dictionary (step S41). The dictionary data ID is set to 0 (initial value) (step S42).

Then, it is determined whether or not there is word information registered in a majority of the number of lower level technical term dictionaries (steps S43, S44).

If there is at least one such word information, all the word information is additionally registered in the higher-level technical term dictionary, and the word information is deleted from the lower-level technical term dictionary (step S35). ). At this time, additional registration may be performed after confirming that the word information does not exist in the higher-level technical term dictionary.

After the above processing is completed, the dictionary data ID
Is incremented by 1 (step S46), and it is determined whether or not unprocessed dictionary data remains (step S4).
7).

If unprocessed dictionary data remains, the process returns to step S42, and if the integration process has been completed for all dictionary data, a series of processes shown in FIG. 9 ends.

Here, the siphoning process (integration process) is performed in the order from the technical term dictionary with the lower hierarchy to the technical term dictionary with the higher hierarchy, so that the word information that has just been siphoned is further higher in the specialty dictionary. It is sometimes picked up in the term dictionary.

In the above description, the condition for registration in the higher-level technical term dictionary is registered with a majority of the number of lower-level technical term dictionaries. However, a registration requirement using another ratio as a threshold may be used. . Also, depending on the number of subordinate terminology dictionaries,
The threshold value that defines the registration requirement for higher ranks may be different. In addition, each term dictionary manages the usage frequency and usage rate of word information, and multiple terminology term dictionaries,
The fact that the word information is used to some extent may be included in the registration requirements for the higher-level technical term dictionary.

Also, in the above description, the word information is deleted from the lower-level technical term dictionary when it is additionally registered in the higher-level technical term dictionary. May be left in the terminology dictionary.

Further, in the above description, the registration in the higher-level technical term dictionary is completely automated, but the server operator or the like displays and confirms word information to be additionally registered in the higher-level technical term dictionary. Above, you may make it register additionally in a high-level technical term dictionary.

Note that word information registered in a plurality of technical terms dictionaries of a large category may be registered in a general term dictionary.

(C-3) Effect of Third Embodiment As described above, according to the third embodiment, a function of siphoning word information from a lower-level technical term dictionary to a higher-level technical term dictionary is provided. Therefore, a higher-level technical term dictionary is enriched, and improvement in the translation quality of documents that do not belong to a subdivided field or genre can be expected to improve the quality.

(D) Fourth Embodiment Next, a fourth embodiment in which the dictionary device according to the present invention is applied to a machine translation function provided on a server on the Internet will be described with reference to the drawings.

(D-1) Configuration of the Fourth Embodiment A machine translation network system to which the fourth embodiment is applied can also be represented by the block diagram of FIG. 2 according to the first embodiment. .

However, in the following, referring to FIG.
The functional configuration of the embodiment will be described. In addition, FIG.
FIG. 12 is a functional block diagram showing functions arranged in FIG. 2 and above and incorporating characteristic functions in the fourth embodiment, and is the same as FIG. 5 and FIG. 8 according to the second and third embodiments described above. , Corresponding parts are denoted by the same reference numerals.

In FIG. 10, a machine translation network system 21B according to the fourth embodiment also has a search / translation server 23B on the
2 and is connected to the browsing / input means 24 via the network.

The fourth embodiment differs slightly from the second and third embodiments in the functional configuration of the search / translation server 23B. That is, in the fourth embodiment, a dictionary division / generation unit 37 is provided instead of the terminology enrichment unit 27 of the second embodiment and the dictionary information integration unit 36 of the third embodiment.

The dictionary database 34 is the same as in the second and third embodiments, but is related to the technical term dictionary in FIG. 10 because of the relationship with the features of the fourth embodiment. The hierarchy with the user dictionary is clearly shown.

The dictionary dividing / generating means 37 constituting the feature of the fourth embodiment is based on the word information of a plurality of user dictionaries associated with the technical term dictionary, and adds the technical term dictionary at a lower level to the technical term dictionary. , Or demote word information to a lower level technical term dictionary.

For example, in the notation of the dictionary database 34 of FIG. 10, the golf dictionary D
s1 and a baseball dictionary Ds2 are generated.

Here, the processing of the dictionary division / generation means 37 may be started periodically, such as once a month (for example, at the beginning of the month).
3 may be started by using an input / output device (not shown) (see the input / output device 18 in FIG. 2).

(D-2) Operation of the Fourth Embodiment Next, the dictionary division / composition which is a feature of the fourth embodiment will be described.
The operation of the generation unit 37 will be described with reference to the flowchart of FIG. Note that the process of FIG. 11 shows a process for one technical term dictionary, and the process of FIG. 11 is executed for all the technical term dictionaries.

When the process shown in FIG. 11 is started, the dictionary dividing / generating means 37 extracts and buffers related user dictionary information from the dictionary database 34 while counting the same word information (step S51). .

FIG. 12 shows an example in which data is taken out from the user dictionary related to the sports dictionary while counting. From the left, the dictionary data ID, the headword (key), the translation, and the number registered in the user dictionary are shown.

When the retrieval while counting is completed, the dictionary dividing / generating means 37 sets the dictionary data ID to 0 (initial value) (step S52).

The processing loop of steps S53 to S58 is repeated for all dictionary data.

The dictionary dividing / generating means 37 checks whether or not there is word information having the same headword (key) but different translation (value) in the registered word information of the user dictionary associated with the technical term dictionary. In some cases, whether the number of user dictionaries in which each word information is registered is equal to or more than 20% of the total number of user dictionaries associated with the technical term dictionary (the ratio is not limited to this). It is determined whether or not it is (steps S53, S54).

If there is user-registered word information with the same headword (key) but a different translated word (value), and the registration ratio of each word information is 20% or more, the dictionary division / generation means 3
7 determines whether there is a lower-level technical term dictionary with respect to the technical term dictionary (step S55).

If there is no lower-level terminology dictionary, the dictionary dividing / generating means 37 outputs the same headword (key).
Then, a terminology dictionary lower in the number of types of word information having different translation words (values) and a registration ratio of 20% or more is created, and each word information is registered (step S56). Here, the user dictionary including such word information is a dictionary data ID.
May be directly associated with the terminology dictionary specified by
You may make it associate with the newly created technical term dictionary in which the registered word information of the self is registered.

The creation of the lower-level technical term dictionary here may be a temporary creation instead of a complete creation.

On the other hand, if there is a lower level technical term dictionary, the dictionary dividing / generating means 37 uses the same headword (key) but a different translation word (value) and a word whose registration ratio is 20% or more. The information is moved to a lower level technical term dictionary (step S57). Here, the user dictionary including such word information may be directly associated with the technical term dictionary specified by the dictionary data ID, or may be associated with the moved lower-level technical term dictionary.

The determination of the type of the lower-level terminology dictionary for moving the word information is performed, for example, by recognizing the field or genre with reference to the contents of the backward dictionary database having the translated word as a headword. In addition, for example, a part or all of the character string in the translated word of the current word information is moved to a lower-level terminology dictionary on the side where the word information including the translated word character string exists.

Here, the movement of the word information to the lower level technical term dictionary may be not a complete movement but a temporary movement.

The dictionary dividing / generating means 37 completes the creation of the lower-level technical term dictionary and the transfer of the predetermined word information to the lower-level technical term dictionary as described above, or the translation of the terminology with the same headword (key). If there is no word information of user registration with different (value), or if there is, if the registration ratio of any word information is less than 20%, the dictionary data ID
Is incremented by 1 (step S58), and it is determined whether or not dictionary data remains (step S5).
9).

If dictionary data remains, the process returns to step S53, and if dictionary division / generation processing has been completed for all dictionary data, a series of processing shown in FIG. 11 ends.

In the dictionary dividing / generating process, it is difficult to completely automate the process, and it is necessary to perform post-processing by an operator on the server operator side (hereinafter, referred to as a system operator) at the end of the process in FIG. For example, naming of a field or genre to a newly created technical term dictionary, determination of whether or not this creation is necessary if the creation in step 56 is provisional creation, and determination of whether or not the creation in step 57 is temporary movement It is necessary to determine whether or not this movement is necessary, and to perform a process of moving other word information from a higher-level technical term dictionary when a lower-level technical term dictionary is newly created.

For example, in the example of FIG. 12, if the number of user dictionaries associated with the sports dictionary is ten, two types of word information, the headword being “Pitcher” and the translated words being “pitcher” and “seventh iron”. Since both have three user registrations, they satisfy the requirement of newly creating a lower-level technical term dictionary or the requirement of moving word information of the lower-level technical term requirement.

In this case, if there is no lower-level technical term dictionary in the sports dictionary, two lower-level technical term dictionaries are created, and one of the word information “Pitch”
er; 7 iron "is registered, and the other word information" P
"itcher;pitcher" is registered. In the post-processing by the system operator, the former will be given the name of a golf dictionary and the latter will be named a baseball dictionary.

On the other hand, if the golf dictionary and the baseball dictionary exist below the sports dictionary, the word information “P
"itcher; 7 iron" is registered in the golf dictionary, and the word information "Pitcher;pitcher" is registered in the baseball dictionary. This means that, for example, if there is another word information whose translated word is “iron” in the golf dictionary,
This is because there is a common character string with the translated word "7 iron". Further, for example, if there is another word information whose translated word is "right hand" in the baseball dictionary, there is a common character string with the translated word "pitcher" this time.

FIG. 13 shows the state of the change of the community dictionary when a golf dictionary and a baseball dictionary are newly created under the sports dictionary through the processing of FIG. 11 from a state where only such a sports dictionary is prepared. It is shown.

In the above description, the determination of the presence / absence of a lower level technical term dictionary is made only from the registered information of the user dictionary. However, based on the registered information of the user dictionary and the registered information of the technical term dictionary, It may be determined whether or not a lower-level technical term dictionary is created. For example, when word information of user registration that is the same as the registered information of the technical term dictionary but has a different translation is registered in a user dictionary of a predetermined number or more or a predetermined ratio or more, creation of a lower level technical term dictionary, etc. May be performed.

In the above description, a case where a lower-level technical term dictionary is created only when a lower-level technical term dictionary does not exist has been described. A technical term dictionary at a lower level of the same hierarchy may be created. That is, irrespective of the presence or absence of the lower-level technical term dictionary, if the condition is satisfied, another lower-level technical term dictionary may be created (in other words, the process of moving to a lower-level technical term dictionary may be eliminated). For example, even if there are a judo dictionary and a land dictionary below the sports dictionary, as described above, the “Pitche
r; 7-iron "and"Pitcher; pitcher "may be registered as a user. In this case, a golf dictionary or a baseball dictionary (naming will be performed later) is created as a lower-level terminology dictionary.

(D) Effects of Fourth Embodiment As described above, according to the fourth embodiment, a new technical term dictionary is created according to the registration status of the user dictionary. Various technical term dictionaries corresponding to various fields and genres can be provided, and appropriate technical term dictionaries for the fields and genres of the translation target Web page can be used for machine translation. Regardless of field or genre, translation quality can be improved.

(E) Other Embodiments In the description of each of the above-described embodiments, various modified embodiments have been referred to, but further modified embodiments as exemplified below can be cited.

The thresholds such as the majority and 20% shown in the above embodiments can be freely changed, and not only the ratio threshold but also an absolute number such as 5 or more may be used as the threshold.

In the second and third embodiments, the word information of a dictionary is automatically registered in another dictionary (technical term dictionary). It may be used in part of the processing. For example, in the second embodiment, when the number of registered users is large and word information to be registered in the technical term dictionary is detected, the user or the system operator may be allowed to determine whether or not to register in the technical term dictionary. . In the case where the user makes a judgment, the decision may be made by a majority decision of the users who provide the user dictionary in association with the technical term dictionary. Examples of the majority information collection method include sending and collecting e-mails to the user, voluntary voting of the user for notification using a bulletin board, and the like.

In the fourth embodiment, the user's judgment can be partially used.

Further, the user may be allowed to directly operate the technical term dictionary. Note that, of course, the system operator can directly operate the technical term dictionary.

Further, it goes without saying that the features of the above embodiments may be appropriately combined.

In each of the above embodiments, the case where the dictionary device of the present invention is used for the machine translation function provided in the server on the Internet has been described, but the application target of the dictionary device of the present invention is not limited to this. It is not done.

For example, the present invention can be applied to a dictionary device when one machine translation device is used by a plurality of persons. Also, the dictionary device of the present invention can be applied to dictionaries other than the dictionary for machine translation. For example, the dictionary device of the present invention can be applied to a kana-kanji conversion dictionary used for voice recognition or the like.

[0164]

As described above, according to the dictionary device of the present invention, the dictionary and its structure can be easily enhanced.
The quality of the natural language processing result obtained by using the dictionary device of the present invention can be improved.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram of hierarchies of various dictionaries according to the dictionary device of the first embodiment.

FIG. 2 is a block diagram showing a schematic configuration of a machine translation network system to which the dictionary device of each embodiment is applied.

FIG. 3 is a flowchart showing an operation of registering in a user dictionary related to the technical term dictionary of the first embodiment.

FIG. 4 is a flowchart illustrating a machine translation operation using a community dictionary according to the first embodiment.

FIG. 5 is a block diagram illustrating a functional configuration according to a second embodiment.

FIG. 6 is a flowchart illustrating an operation of enriching a technical term dictionary according to the second embodiment.

FIG. 7 is an explanatory diagram of user-registered word information registered in a technical term dictionary in the second embodiment.

FIG. 8 is a block diagram illustrating a functional configuration according to a third embodiment.

FIG. 9 is a flowchart illustrating an operation of integrating a higher-level terminology dictionary according to the third embodiment.

FIG. 10 is a block diagram illustrating a functional configuration according to a fourth embodiment.

FIG. 11 is a flowchart showing a division / generation operation of a lower-level terminology dictionary in the fourth embodiment.

FIG. 12 is an explanatory diagram of word information of a user registration in which a lower-level technical term dictionary is generated in the fourth embodiment.

FIG. 13 is an explanatory diagram showing a dictionary structure before and after a lower-level terminology dictionary is generated in the fourth embodiment.

[Explanation of symbols]

21, 21A, 21B: Machine translation network system, 22: Internet, 23, 23A, 23B: Search / translation server, 24: Browsing / input means, 31: Communication control means, 32: Machine translation means, 33: Dictionary management means , 3
4 ... Dictionary database, 35 ... Technical term dictionary enrichment means, 36 ... Dictionary information integration means, 37 ... Dictionary division / generation means.

 ────────────────────────────────────────────────── ─── Continued on the front page (72) Inventor Yasuhiro Kawakita 1-7-12 Toranomon, Minato-ku, Tokyo Oki Electric Industry Co., Ltd. (72) Inventor Satoshi Nakagawa 1-7-12 Toranomon, Minato-ku, Tokyo Offshore Inside Electric Industry Co., Ltd. (72) Inventor Toshihiko Matsunaga 1-7-12 Toranomon, Minato-ku, Tokyo Oki Electric Industry Co., Ltd. F-term (reference) 5B091 CC02 CC14 CC15 CC17 CD03 EA17

Claims (6)

[Claims] 1. A dictionary device provided in a natural language processing device used by a plurality of users, wherein a plurality of technical term dictionaries for each category in which specialized fields and genres are classified are formed in a tree structure using a general term dictionary as a root node. A dictionary device comprising: a hierarchical dictionary main body; and a user dictionary registration unit for setting a user dictionary in association with an arbitrary technical term dictionary. 2. When a category to be subjected to natural language processing is designated, all the technical term dictionaries on a tree-structured path from a technical term dictionary of the category to a general term dictionary are associated with the technical term dictionaries. 2. The dictionary apparatus according to claim 1, further comprising an applied dictionary determining unit that determines a dictionary to which all of the user dictionaries of the process requesting user are applied. 3. A dictionary device provided in a natural language processing device used by a plurality of users, wherein a user dictionary for each user editable by the user, a common dictionary common to all users, and the same dictionary information are provided under a predetermined condition. And a common dictionary enrichment means for registering the dictionary information in the common dictionary when the number of the user dictionaries satisfying the condition is satisfied. 4. An upper-level dictionary in a hierarchical structure, a plurality of lower-level dictionaries located below the upper-level dictionary, and when the same dictionary information is stored in a number of the lower-level dictionaries satisfying a predetermined condition. And a dictionary information integrating means for registering the dictionary information in the upper dictionary. 5. A dictionary device provided in a natural language processing device used by a plurality of users, a user dictionary for each user editable by the user, a common dictionary common to all users, and at least storage of each of the user dictionaries. A dictionary device comprising: a dictionary division / generation unit configured to set a lower dictionary below the common dictionary based on contents. 6. The dictionary dividing / generating means, wherein the same dictionary information is stored in a number of the user dictionaries satisfying a predetermined condition, and the other dictionary information having the same key and a different value as the dictionary information is: 6. The dictionary device according to claim 5, wherein when the number of user dictionaries satisfying a predetermined condition is stored, lower dictionaries each storing the dictionary information are set below the common dictionary.