JPH10269215A - Natural language analyzing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a natural language analyzing device that automatically generates, and adds and corrects an omitted element complementing rule for complementing omitted elements, by extracting omitted expressions of a source language and complementary elements of the omitted elements in a translated sentence of a target language. SOLUTION: An input sentence and an ideal translated sentence of the target language are read in through a translated sentence read-in part 1, an input sentence omitted element extraction part 52 is extracted omitted elements omitted from the input sentence, and a target language complementing element extraction part 53 extracts target language complementary elements complementing the omitted elements from the ideal translated sentence. On the basis of the extracted target language complementary elements, source language complementary element extraction part 54 extracts source language complementary elements and on the basis of omitted and complementary element pairs of the target language complementary elements and source language complementary elements, an omitted element complementing rule generation and alteration part 6 adds and alters the omitted element complementing rule.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a natural language analyzer for automatically learning and generating an abbreviation element complementing rule for complementing an abbreviation element such as a word or an expression abbreviated in a natural language.

[0002]

2. Description of the Related Art Conventionally, when constructing an abbreviation element complementing rule in a natural language analyzer, it is common to decode an input sentence to be analyzed and construct an abbreviated element complementing rule manually. In addition, the abbreviated element completion rule constructed in this way is based on the analysis result of the natural language analyzer, and when an inappropriate abbreviated element is complemented, the cause of the abbreviated element is complemented. Experts who understand the internal processing of the natural language analyzer well add and modify the rules manually

[0003]

As described above, in the conventional method of constructing an abbreviation element complementing rule in a natural language analysis device, in order to newly create or add a new rule, the internal processing of the natural language analysis device must be performed. It is necessary to create rules and verify the analysis results by well-understood experts. Therefore, there is a problem that a rule cannot be newly created or added or modified unless an expert is used. Also,
Since this work is performed manually, there is a problem that the time and economic cost for constructing the rules are increased.

[0004] The present invention has been made in view of the above,
Its purpose is to extract the abbreviation in the source language and the complement of the abbreviation in the target language bilingual sentence,
An object of the present invention is to provide a natural language analysis device that automatically creates and additionally corrects an omitted element complementing rule for complementing an omitted element.

[0005]

In order to achieve the above object, according to the present invention, an input sentence of a first natural language which is an original language is analyzed, and an original sentence which is an analysis structure of the input sentence is analyzed. A natural language analyzer that generates a linguistic analysis structure and generates and changes an abbreviation element completion rule for complementing an abbreviation element,
An input unit for inputting an ideal translation of a target language, which is a second natural language having a bilingual relationship with the input sentence; an input sentence omitting element extracting unit for extracting an abbreviated element from the input sentence; Target language supplementary element extracting means for extracting a target language complementary element complemented from the omission element, and a source language for extracting a source language complementary element based on the target language complementary element extracted by the target language complementary element extracting means A gist comprising: a supplementary element extracting unit; and an abbreviated element complementing rule generation changing unit that additionally changes the abbreviated element complementing rule based on an abbreviated complementing element pair that is a pair of the target language complementing element and the source language complementing element. I do.

According to the first aspect of the present invention, an input sentence and an ideal translation of a target language are input, an omission element is extracted from the input sentence, a target language complement element is extracted from the ideal translation, and the target language is extracted. The source language completion element is extracted based on the completion element, and the omitted element completion rule is additionally changed based on the omitted complementary element pair.

According to a second aspect of the present invention, an input sentence of a first natural language as a source language is analyzed, a source language analysis structure as an analysis structure of the input sentence is generated, and omitted elements are complemented. A natural language analysis device for generating and changing an abbreviation element complementing rule for performing the
Reading means for reading an ideal translation of a natural language, extracting an omitted element omitted from the input sentence from the input sentence and the ideal translated sentence, and extracting a target language complement element complemented by the omitted element from the ideal translated sentence. An abbreviation complement element pair extracting means for extracting and extracting an abbreviation complement element pair that is a pair of the abbreviation element and a source language complement element to be complemented with the abbreviation element based on the omission element and the target language complement element; The gist of the invention is to include an omitted element complementing rule generation changing unit that generates and changes the omitted element complementing rule based on the omitted complementary element pair extracted by the omitted complementary element pair extracting unit.

According to the second aspect of the present invention, an input sentence and an ideal translation are read, an omitted element in the input sentence is extracted from the input sentence and the ideal translated sentence, and a target language complemented by the omitted element from the ideal translated sentence. A complementary element is extracted, and an abbreviated element and an abbreviated complement element pair, which is a pair of a source language complement element to be complemented with the abbreviated element, are extracted based on the abbreviated element and the target language complement element. Generate and change the omission element completion rule based on.

Further, the present invention according to claim 3 provides the invention according to claim 2
In the described invention, the omitted complementary element pair extracting means includes:
The gist includes an input sentence omitted element extracting unit that extracts an omitted omitted element from the input sentence, and a target language complementary element extracting unit that extracts a target language complementary element that is complemented by the omitted element from the ideal translation. And

According to the third aspect of the present invention, the omitted elements omitted from the input sentence in the omitted complementary element pair extracting means are extracted by the input sentence omitted element extracting means, and the omitted elements are complemented from the ideal translation. The target language complement element is extracted by the target language complement element extracting means.

According to a fourth aspect of the present invention, in the second or third aspect of the present invention, the omission complementary element pair extracting means includes an input word expression in the input sentence from the input sentence and the ideal translation sentence, and the input word The gist of the present invention is to provide a bilingual expression pair extracting means for extracting an input sentence ideal translation word expression pair which is a pair of an ideal translation word expression in the ideal translation sentence having a bilingual relation with the expression.

According to the present invention, in the abbreviation complementary element pair extraction means, the input word expression in the input sentence and the ideal word in the ideal translation having a bilingual relationship with the input word expression from the input sentence and the ideal translation. An input sentence ideal translation word expression pair, which is a pair of translation word expressions, is extracted by a translation expression pair extraction unit.

The present invention described in claim 5 provides the present invention according to claim 3.
In the invention according to the fourth aspect, the omitted complementary element pair extracting means includes a source language complementary element extracting means for extracting the source language complementary element based on the target language complementary element extracted by the target language complementary element extracting means. It is the gist to have.

According to the fifth aspect of the present invention, the source language supplementary element extracting means extracts the source language supplementary element based on the target language supplementary element extracted by the target language supplementary element extracting means in the omitted complementing element pair extracting means. Extract with

[0015] Further, the present invention according to claim 6 provides the present invention as claimed in claim 2.
6. The invention according to any one of claims 1 to 5, further comprising input sentence analyzing means for analyzing the input sentence read by the bilingual sentence reading means to generate an input sentence analysis structure, and utilizing the input sentence analysis structure. The gist is to extract the omitted element, the target language complement element, and / or the source language complement element.

According to the present invention, the input sentence read by the bilingual sentence reading means is analyzed to generate an input sentence analysis structure by the input sentence analysis means, and the input sentence analysis structure is used. To extract the omitted element, the target language complement element, and / or the source language complement element.

According to a seventh aspect of the present invention, in the invention of the sixth aspect, the input sentence analyzing means has an input sentence omitted element complementing means for complementing an omitted element in the input sentence.

According to the present invention, in the input sentence analyzing means, the omitted elements in the input sentence are complemented by the input sentence omitted element complementing means.

The present invention described in claim 8 provides the present invention according to claim 2.
8. The invention according to any one of claims 7 to 7, further comprising an ideal translation analysis unit that analyzes the ideal translation read by the bilingual reading unit to generate an ideal translation analysis structure, and uses the ideal translation analysis structure. The gist is to extract the omitted element, the target language complement element, and / or the source language complement element.

According to the present invention, the ideal translation sentence read by the bilingual reading means is analyzed to generate an ideal translation analysis structure by the ideal translation analysis means, and the ideal translation analysis structure is omitted using the ideal translation analysis structure. Extract elements, target language complement elements and / or source language complement elements.

Further, the present invention described in claim 9 is based on claim 8.
In the invention described above, the gist of the invention is that the ideal translation analyzing unit has an ideal translation omitting element complementing unit that complements an omitting element in the ideal translation.

In the ninth aspect of the present invention, the omitted elements in the ideal translation are complemented by the ideal translation omitted element complementing means in the ideal translation analyzing means.

According to a tenth aspect of the present invention, in the invention of the eighth or ninth aspect, the ideal translation analyzing means determines the target of an anaphoric expression such as a pronoun in the ideal translation. It is the gist to have.

According to the tenth aspect of the present invention, the target of the anaphor expression such as a pronoun in the ideal translation is determined by the ideal translation anaphor analysis means.

According to the present invention as set forth in claim 11, in the invention as set forth in any of claims 6 to 10, the input sentence analyzing structure wherein the omitted element complementing rule generation changing means is obtained by the input sentence analyzing means. The gist of the invention is to generate and change the omitted element complementing rule by using the above.

According to the eleventh aspect of the present invention, the omitted element complementing rule is generated and changed by using the input sentence analysis structure obtained by the input sentence analyzing means.

Further, according to a twelfth aspect of the present invention, in the invention according to any one of the second to eleventh aspects, the omitted element complementing rule generation / change means is the omitted element extracted by the omitted element pair extracting means. An omission complement element pair storage means for recording a complement element pair, wherein the omission complement element pair storage means stored based on a set of pairs of the input sentence and the ideal translation sentence having a bilingual relationship to be learned. The gist is to generate and change the omitted element complementing rule using the set of the omitted complementary element pairs.

According to the twelfth aspect of the present invention, the omitted complementing element pair extracted by the omitted complementing element pair extracting means by the omitted element complementing rule generation changing means is recorded by the omitted complementing element pair storage means, and the learning target The abbreviation element complementing rule is generated and changed using the set of the abbreviation complement element pairs stored in the abbreviation complement element pair storage means accumulated based on the set of the input sentence and the ideal translation sentence having the bilingual relation.

According to a thirteenth aspect of the present invention, in the invention according to any one of the seventh to twelfth aspects, the input sentence omitted element complementing means complements an omitted element in the input sentence using the omitted element complementing rule. The gist is that the omitted element complementing rule learning method is repeatedly performed.

According to the thirteenth aspect of the present invention, the input sentence omitting element complementing means complements the omitting element in the input sentence using the omitting element complementing rule, and repeatedly executes the omitting element complementing rule learning method.

According to a fourteenth aspect of the present invention, the gist of the thirteenth aspect is that the omitted element complementing rule generation changing means has a learning end determining means for determining the end of learning.

According to the fourteenth aspect of the present invention, the end of learning is determined by the omission element complementing rule generation changing means by the learning end determining means.

Further, according to a fifteenth aspect of the present invention, in the fourteenth aspect of the present invention, the source language is a history of the source language analysis structure which is a result of analyzing the input sentence using the omitted element completion rule. It has a source language analysis structure history storage unit that records the analysis structure history, and the learning end determination unit ends the learning when the recorded content of the source language analysis structure history storage unit reaches a steady state. I do.

According to the fifteenth aspect of the present invention, the source language analysis structure history, which is the history of the source language analysis structure obtained as a result of analyzing the input sentence using the omitted element completion rule, is stored in the source language analysis structure history. The learning is terminated when the recorded content of the source language analysis structure history storage means is in a steady state.

[0035]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing a configuration of a natural language analyzer according to one embodiment of the present invention. The natural language analysis device shown in the figure is a natural language analysis system that automatically learns and generates an abbreviated element completion rule for complementing an abbreviated element such as a word or expression omitted in a natural language. To implement the rule learning method,
It is assumed that the first natural language is Japanese and the second natural language is English.

The natural language analyzing apparatus shown in FIG. 1 includes a bilingual sentence reading unit 1, a natural language analyzing system unit 2, a rule 3, an ideal translated sentence analyzing unit 4, an omitted complementary element pair extracting unit 5, an omitted element complementary rule generation changing unit. 6.

The bilingual sentence reading unit 1 inputs a pair (a bilingual sentence) of a first natural language sentence (input sentence) and a second natural language sentence (ideal translated sentence) having a bilingual relationship.

The natural language analysis system unit 2 analyzes the input sentence read by the bilingual sentence reading unit 1 using rule 3 to generate an input sentence analysis structure. Natural language analysis system 2
Is composed of an input sentence syntactic and semantic analyzing unit 21 and an input sentence omitting element complementing unit 22. Rule 3 is syntactic and semantic analysis rule 3.
1. It is composed of an omitted element complementing rule 32.

The input sentence syntactic and semantic analysis unit 21 of the natural language analysis system unit 2 performs a syntactic and semantic analysis of the input sentence read by the bilingual sentence reading unit 1 by using the syntactic and semantic analysis rule 31, and obtains the input sentence syntactic and semantic structure. Generate The input sentence abbreviation element complementing unit 22 complements an omitted element in the input sentence based on the input sentence syntax / semantic structure generated by the input sentence syntax / semantic analysis unit 21 using the omitted element complementing rule 32.

The ideal translated sentence analyzing unit 4 is a bilingual sentence reading unit 1
Analyzes the ideal translated sentence read in step (1) to generate an ideal translated sentence analysis structure. The ideal translated sentence analyzing unit 4 includes an ideal translated sentence syntactic and semantic analyzing unit 41, an ideal translated sentence omitting element complementing unit 42, and an ideal translated sentence analytic analyzing unit 43.

The ideal translated sentence syntactic and semantic analyzing unit 41 of the ideal translated sentence analyzing unit 4 performs a syntactic and semantic analysis of the ideal translated sentence read by the bilingual sentence reading unit 1 and generates an ideal translated sentence syntactic and semantic structure. The ideal translated sentence omitted element complementing section 42 complements the omitted element in the input sentence based on the input sentence syntax / semantic structure generated by the ideal translated sentence syntax / semantic analyzing section 41. Ideal translation sentence analysis unit 43
Determines the target of an anaphoric expression such as a pronoun in an ideal translation based on the ideal translation syntax / semantic structure generated by the ideal translation syntax / semantic analysis unit 41.

The omitted complementary element pair extraction unit 5 extracts an omitted element from the input sentence and extracts the omitted element from the ideal translated sentence, for the input sentence and the ideal translated sentence read by the bilingual sentence reading unit 1. Extract the target language complement elements to be complemented, extract the source language complement elements that are in parallel with the target language complement elements based on the extracted omitted elements and target language complement elements, and extract the extracted omitted language elements and target language complement. Extract an abbreviated complement element pair that is a pair of an element and a source language complement element. The omitted complementary element pair extraction unit 5 includes a bilingual expression pair extraction unit 51, an input sentence omitted element extraction unit 52, a target language complement element extraction unit 53, and a source language complement element extraction unit 54.

The bilingual expression pair extraction unit 51 of the abbreviation complementary element pair extraction unit 5 converts the input sentence and ideal translation sentence read by the bilingual sentence reading unit 1 into an input word expression in the input sentence and realizes the input word. An input sentence ideal translation word expression pair, which is an ideal translation word expression pair in an ideal translation having a bilingual relationship, is extracted. The input sentence omitting element extraction unit 52 is extracted from the input sentence by the input sentence analysis structure generated by the natural language analysis system unit 2 and the ideal translated sentence analysis structure generated by the ideal translated sentence analysis unit 4 and the bilingual expression pair extraction unit 51. The abbreviated element which is the abbreviated element is extracted using the input sentence ideal translation word expression pair. The target language supplementary element extraction unit 53 is extracted by the natural language analysis system 2 from the ideal translation and the ideal translation analysis structure generated by the ideal translation analysis unit 4 and extracted by the bilingual expression pair extraction unit 51. A target language complement element to be complemented with the omitted element is extracted using the input sentence ideal translation word expression pair. The source language supplementary element extracting unit 54 extracts a source language supplementary element based on the target language supplementary element extracted by the target language supplementary element extracting unit 53.

The omitted element complementing rule generation changing unit 6 generates and changes the omitted element complementing rule 32 in the rule 3 based on the omitted complementary element pair extracted by the omitted complementary element pair extracting unit 5.

FIG. 2 shows an example of a basic block configuration of a natural language analysis apparatus using an abbreviation element complementing rule which is additionally changed in the present invention. The natural language analysis device shown in FIG. 1 includes a first natural language input unit 7 for inputting a sentence of a first natural language to be analyzed, a natural language analysis system unit 2, a rule 3 used in the natural language system unit, a natural language analysis. The first natural language analysis structure output unit 8 outputs the first natural language analysis structure generated by the system unit 2.

The natural language analysis system unit 2 comprises an input sentence syntax / meaning analysis unit 21 and an input sentence omitting element complementing unit 22. Rule 3 includes a syntactic and semantic analysis rule 31 and an omission element complementing rule 32. The syntactic and semantic analysis rule 31 includes a syntactic and semantic analysis dictionary 311, and a syntactic and semantic analysis rule 312. The input sentence syntactic and semantic analysis unit 21 of the natural language analysis system unit 2 performs a syntactic and semantic analysis of the first natural language sentence read by the first natural language input unit 7 using the syntactic and semantic analysis rule 31 and inputs the sentence. Generate a sentence syntax semantic structure. The input sentence abbreviation element complementing unit 22 complements the omitted element of the first natural language sentence based on the input sentence syntax / semantic structure generated by the input sentence syntax / semantic analysis unit 21 using the omitted element complementing rule 32.

FIG. 3 shows an example of a syntactic and semantic analysis dictionary 311 according to an embodiment of the present invention. The syntactic and semantic analysis dictionary 311 is
For example, there is a word syntax and semantic dictionary, and the standard form, part of speech, and semantic attributes for each word of Japanese as the first natural language are entered in the word syntax and semantic dictionary. For example, it is described that the standard form of "car" is also "car", the part of speech is "noun", and the semantic attribute is "concrete". In addition, the standard form of "wo" is "wo", the part of speech is "particle", and the semantic attribute is "purpose". The standard forms of “want to buy” are “buy” and “tai”.
For "buy", the part of speech is "verb" and the semantic attribute is "purchase". For "tai", the part of speech is "modal" and the semantic attribute is Is described as "hope".

FIG. 4 shows an example of the syntactic and semantic analysis rule 312 according to the embodiment of the present invention. Syntax and semantic analysis rule 312
For example, there are verb syntactic and semantic analysis rules, and the verb syntactic and semantic analysis rules include not only Japanese verbs but also
The verb particles are attached as patterns, that is,
For example, for the verb “buy”, the particle pattern “S” (subject, noun) “ga” (particle) “O” (object, noun) “wo” (particle) ”is entered. A search is made to search for a syntactic and semantic analysis structure centered on the verb as a key including the verb, and further, a selection condition indicating the meaning of the case element of the verb is described.
For the verb "buy", the particle pattern (S is O
)), The syntactic and semantic analysis indicates that the verb "buy" has a modality "hope" and that S has a nominative relation and O has a relation with the object case. In the case of generating a structure, it indicates that there is a restriction to apply the syntactic and semantic analysis structure when S has the main “meaning” and O has the meaning of “concrete”.

FIG. 5 shows an example of the omitted element complementing rule 32 according to an embodiment of the present invention. In the omitted element complementing rule 32, for example, a syntactic and semantic analysis structure condition and a complementary element corresponding to the syntactic and semantic analysis structure condition are described. When the syntactic and semantic analysis structure matches the syntactic and semantic analysis structure condition, the complement element is complemented. For example, if an arbitrary (*) verb is accompanied by modal "prohibition", elements having a nominative relation are omitted, and elements having a target case are omitted if not omitted. It indicates that the nominative complements you.

Next, the processing procedure of the omission element complementing rule learning system for a natural language analysis system having the basic block configuration shown in FIG. 1 will be described. Here, the procedure of this processing will also be described by taking the bilingual sentence shown in FIG. 6 as an example.

First, a bilingual sentence reading unit 1 inputs a pair (a bilingual sentence) of a first natural language sentence and a second natural language sentence having a bilingual relationship. The bilingual sentence is used for automatic learning of the omitted element complementing rule for the natural language analysis system. By selecting a set of bilingual sentences (bilingual sentences) for learning of this abbreviation element complementing rule, a natural language analysis system omits the natural language analysis system in the same field sentences where similar omission phenomena frequently occur as the learned bilingual sentences. It is possible to tune the complementary processing. In the example of FIG. 6, the first natural language sentence in Japanese (input Japanese sentence) “I want to buy a car”
, An English sentence (ideal English translated sentence), which is a second natural language having a bilingual relationship, is input as a translated sentence.

The input Japanese sentence, which is the first natural language sentence, of the input bilingual sentence is input Japanese sentence which is the first natural language analysis structure by the natural language analysis system unit 2 using rule 3. Generate an analysis structure. The processing procedure of this natural language analysis system will be described using a natural language analysis system having a basic block configuration shown in FIG.

First, an input Japanese sentence as a first natural language sentence is, for example, a syntactic and semantic analysis dictionary 311 such as a word syntactic and semantic analysis dictionary shown in FIG. 3 or a verb syntactic and semantic analysis rule such as shown in FIG. Input sentence syntactic and semantic analysis unit 2 using syntactic and semantic analysis dictionary 31 including syntactic and semantic analysis rule 312
1 to generate a Japanese syntactic and semantic analysis structure. The generated Japanese syntactic and semantic analysis structure is, for example, a Japanese sentence whose abbreviation elements are complemented by the input sentence abbreviation element complementing unit 22 using the abbreviation element complementation rules 32 such as the nominative abbreviation element complementation rules shown in FIG. Generate an analysis structure. The generated Japanese sentence analysis structure is output by the first natural language analysis structure output unit 8.

The analysis structures and rules referred to and generated at each stage of the natural language analysis system are stored with information so that they can be referred to in the omitted element complementing rule automatic learning process. For example, individual syntactic and semantic analysis structures and selection conditions obtained when the syntactic and semantic analysis rule 312 in FIG. 4 is searched are also stored.

In the natural language analysis system as described above, the syntactic and semantic analysis dictionary 311 shown in FIG. 3, the syntactic and semantic analysis rule 312 shown in FIG. 4, and the omitted element complementing rule 32 shown in FIG. Thus, the input sentence analysis structure shown in FIG. 7 is generated for the input Japanese sentence shown in FIG.
In these input sentence analysis structures, words in the input sentence are described according to the relation between verbs and each element and the function. Although the nominative element in the input Japanese sentence of FIG. 6 is omitted, the abbreviation element complementing rule 32 of FIG. 5 does not include an abbreviation element complementing rule for complementing the nominative of the input Japanese sentence. The complementing unit 22 does not perform complementation.

Here, the relation of "subject" indicates that the noun phrase functions as the subject of the verb, and the relation of "object" indicates that the noun phrase functions as the object of the verb. Show.

Next, the ideal translation analyzing section 4 shown in FIG. 1 will be described. The ideal translation analysis unit 4 automatically generates an ideal translation analysis structure by analyzing the sentence of the second natural language (ideal English translation) read by the bilingual sentence reading unit 1.
The procedure of the process of the ideal translation analysis unit 4 will be described. First, the ideal English translation sentence, which is a sentence of the second natural language, is analyzed by the ideal translation sentence syntactic and semantic analysis unit 41 to generate an English syntactic and semantic analysis structure. The generated English syntactic and semantic analysis structure generates an English analysis structure in which omitted elements are complemented by the ideal translation sentence omitted element complementing unit 42. The English analysis structure in which the omitted elements are supplemented generates an English analysis structure to which the information on the target of the anaphor expression is added by the ideal translation anaphor analysis unit 43.

For the ideal English translation shown in FIG. 6, an ideal translation analysis structure shown in FIG.
As in the input sentence analysis structure, the words in the ideal translation are described according to the relationship between verbs and each element and the functions in these ideal sentence analysis structures. Since there is no omission in the ideal English translation of FIG. 6,
Although the ideal translation omitted element complementing unit 42 does not perform complementation,
Since “my” is an anaphoric expression in the ideal translation sentence anaphor analysis unit 43, the information of “I” to be designated is added to the analysis structure of “my”.

The input sentence analysis structure and the ideal translated sentence analysis structure obtained as described above are converted by the bilingual expression pair extraction unit 51 into a word expression in the input sentence analysis structure and a word expression in the ideal translated sentence analysis structure which are in a bilingual relationship. , An input sentence ideal translation word expression pair is extracted. Such an input sentence ideal translation sentence word expression pair is extracted by referring to, for example, a Japanese-English bilingual dictionary holding translation element pairs of Japanese as the first natural language and English as the second natural language. You. A Japanese-English bilingual expression pair extracted from the input sentence analysis structure and the ideal translation analysis structure shown in FIG. 7 is extracted by the bilingual expression pair extraction unit 51, and FIG.
An input sentence word expression pair as shown in FIG.

FIG. 8 shows an analysis structure obtained from the input sentence analysis structure corresponding to the input sentence word expression "buy" and an ideal translation sentence analysis structure corresponding to the ideal translated word expression "want to buy". Analysis structure obtained from the input sentence analysis structure corresponding to the word expression pair of the analysis structure, the input sentence word expression "car", and the analysis structure obtained from the ideal translation sentence analysis structure corresponding to the ideal translation word expression "car" Are extracted by the bilingual expression pair extraction unit 51. In addition,
In the present embodiment, since the system includes the natural language analysis system unit 2 for analyzing the input sentence and the ideal translated sentence analyzing unit 4 for analyzing the ideal translated sentence, the input sentence ideal translated word expression pair has an analysis structure of the input sentence and the ideal translated sentence. Although extracted, a method of holding a pair of each word expression itself is also conceivable.

Using the input sentence ideal translation sentence word expression pair, the input sentence analysis structure, and the ideal translation sentence analysis structure obtained as described above, the input sentence omission element extraction unit 52 extracts the omitted element in the input sentence.

FIG. 9 is a flowchart showing processing relating to extraction of input sentence omitting elements according to an embodiment of the present invention.

In step S5201, the bilingual expression pair extracting unit 52 of the omitted complementary element pair extracting unit 5 extracts an input sentence ideal translation word expression pair from the result of the bilingual expression pair extracting unit 51. In the example of FIG. 8, “want to buy” and “want to buy”
A pair of an analysis structure corresponding to "" and a pair of an analysis structure corresponding to "car" and "car" are extracted.

In the next step S5202, the input sentence analysis structure is extracted from the result of the natural language analysis system unit 2.
The input sentence analysis structure shown in FIG. 7 is extracted from the input sentence shown in FIG.

In step S5203, an ideal translation analysis structure is extracted from the result of the ideal translation analysis unit 4. The ideal translation analysis structure shown in FIG. 7 is extracted from the ideal translation shown in FIG.

In step S5204, input sentence substructures are sequentially extracted from the input sentence analysis structure extracted in step S5202. In the example of the input sentence analysis structure in FIG.
The partial structure corresponding to "want to buy", the partial structure corresponding to the element with the nominative omission flag set, and the partial structure corresponding to the target case "car" are extracted.

In step S5205, step S52
It is determined whether the input sentence substructure extracted in step 04 has an omission flag. If it is determined that the omission flag is present (YES), the process proceeds to step S5206; otherwise (NO), the process proceeds to step S5207. In the above example, since the omission flag is set for the input sentence substructure corresponding to the nominative element, the determination is YES, and step S
The process moves to 5206. In addition, since the omission flag is not set for both the input sentence partial structure corresponding to “want to buy” and the partial structure corresponding to “car”, the determination is NO.
It moves to step S5207.

In step S5206, the input sentence partial structure with the omission flag set is extracted as an omission element. In the above example, the partial structure corresponding to the element with the nominative omission flag is extracted as the omission element.

In step S5207, for the input sentence substructure, it is determined whether or not there is an ideal translated sentence substructure having a bilingual relationship with the input partial structure by referring to the input sentence ideal translated word expression pair. If it exists (YES), the process shifts to step S5208; otherwise (NO), step S5208
It moves to 5211. In the above example, the input sentence substructure corresponding to “want to buy” and the input sentence substructure corresponding to “car” are the ideal translated sentence substructure corresponding to “want to buy” and the ideal corresponding to “car”, respectively. Since both the translated sentence substructures exist, the answer is YES in step S5208.
Move to The input sentence partial structure corresponding to the element with the nominative omission flag is set to NO because there is no ideal translation sentence partial structure in a bilingual relationship.
Move to 1.

In step S5208, for the input sentence substructure, an adjacent input sentence substructure which is an input sentence substructure adjacent to the input sentence substructure is extracted, and an ideal translation sentence substructure having a bilingual relation with the input sentence substructure is extracted. An adjacent ideal translation partial structure that is an adjacent ideal translation partial structure is extracted. Here, the partial structure adjacent to a certain partial structure indicates, for example, a partial structure immediately below the partial structure that is directly related to the partial structure.
In the above example, the adjacent input sentence substructure of the input sentence substructure corresponding to “want to buy” corresponds to the input sentence substructure corresponding to the element with the nominative omission flag set and the target case “car”. This is an input sentence substructure, and there is no adjacent input sentence substructure of the input sentence substructure corresponding to “car”. Also,
The adjacent ideal translation partial structure corresponding to "want to buy" which is in parallel with "want to buy" is an ideal translation partial structure equivalent to the nominative "I" and a target car "car"
Is the ideal translation substructure corresponding to "car" and the ideal translation substructure adjacent to the ideal translation substructure corresponding to "car" in parallel with "car" is the ideal translation substructure corresponding to the possessive case "my" is there. Here, the partial structure immediately below the partial structure that is directly related to the partial structure is defined as an adjacent partial structure. However, a method of defining a partial structure other than this definition as an adjacent partial structure is also conceivable.

In step S5209, step S52
In the input sentence substructure extracted in step 08, the ideal translated sentence substructure having a bilingual relationship with the input sentence substructure, and its adjacent ideal translated sentence substructure, the adjacent input sentence substructure does not exist. It is determined whether an input sentence substructure that exists at one location and has a bilingual relationship with its adjacent substructure does not exist in the input sentence ideal translation word expression pair.
If this condition is satisfied (YES), the flow shifts to step S5210, and if not (NO), step S5211.
Move to In the above example, there is no adjacent input sentence substructure in the input sentence substructure corresponding to “car”, and the adjacent ideal translation portion of the ideal translation substructure corresponding to “car” in parallel with “car” Since there is one ideal translated sentence partial structure corresponding to the possessive case “my”, and according to the input sentence ideal translated word expression pair shown in FIG. 8, there is no input sentence partial structure having a parallel translation relationship with “my”. YES, and step S5210
Move to Further, since there are two adjacent input sentence substructures in the input sentence substructure corresponding to “want to buy”, N
It becomes O, and it transfers to step S5211.

In step S5210, step S52
The adjacent input sentence partial structure that satisfies the condition 09 is extracted as an omitted element. In the above example, the adjacent input sentence partial structure of the input sentence partial structure of “car” is extracted as an omitted element. In this case, the relationship between the input sentence partial structure of “car” and its adjacent input sentence partial structure is undecided.

In step S5211, it is determined whether there is any input sentence substructure that has not been checked yet.
Finally, when the inspection is completed for all the input sentence substructures in the input sentence analysis structure generated by the natural language analysis system unit 2, the processing of the input sentence omitting element extraction unit 52 ends.

As for the input sentence analysis structure of FIG. 7, the input sentence omitted element shown in FIG. In the example of FIG. 10, the input sentence omitting element is extracted as a pair of the input sentence partial structure of the input sentence omitting element and the immediately preceding input sentence partial structure.

Next, the processing of the target language supplementary element extraction unit 53 will be described. FIG. 11 is a flowchart showing the target language supplementary element extraction processing according to one embodiment of the present invention.

In the process of FIG. 11, first, in step S530
In 1, the target language supplementary element extraction unit 53 of the omitted supplementary element pair extraction unit 5 extracts the input sentence ideal translation sentence word expression pair from the result of the bilingual expression pair extraction unit 51. In the example of FIG. 8, a pair of analysis structures corresponding to “want to buy” and “want to buy” and a pair of analysis structures corresponding to “car” and “car” are extracted.

In step S5302, the input sentence analysis structure is extracted from the result of the natural language analysis system unit 2. FIG.
The input sentence analysis structure shown in FIG. In step S5303, an ideal translation analysis structure is extracted from the result of the ideal translation analysis unit 4. The ideal translation analysis structure shown in FIG. 7 is extracted from the ideal translation shown in FIG. In step S5304, the input sentence omitting element extraction unit 5
The input sentence omitting element is extracted from the result of 2. For the input sentence of FIG. 6, the input sentence omitting element of FIG. 10 is extracted.

In step S5305, based on the input sentence analysis structure extracted in step S5302, the ideal translated sentence partial structure in which no bilingual input sentence partial structure exists in the input sentence ideal translated word expression pair extracted in step S5301. In order. In the example of the ideal translation analysis structure in FIG. 7, a partial structure corresponding to "I" and a partial structure corresponding to "my" are extracted.

In step S5306, it is determined whether the ideal translation partial structure extracted in step S5305 is a pronoun expression or a definite article expression. If it is either expression (YES), the process proceeds to step S5307; otherwise (NO), the process proceeds to step S5312. In the above example, since both “I” and “my” are pronoun expressions, the determination is YES, and the flow shifts to step S5307. In this embodiment, a pronoun expression and a definite article expression are used as conditions, but other conditions may be used.

In step S5307, the next higher ideal translation partial structure that is directly related to the ideal translation partial structure that is a pronoun expression or a definite article expression is extracted. In the above example, "I"
For, an ideal translation partial structure corresponding to “want to buy” is extracted, and for “my”, an ideal translation partial structure corresponding to “car” is extracted.

At step S5308, at step S53
In step 07, the input sentence substructure that has a bilingual relationship with the immediately above ideal translation partial structure extracted in step 07 is extracted from the input sentence ideal translation word expression pair. In the above example, using the input sentence ideal translated word expression pair shown in FIG. 8, “I want to buy” from the partial structure corresponding to “want to buy” which is the ideal translated sentence partial structure immediately above “I”. The input sentence substructure corresponding to is extracted, and “my”
An input sentence partial structure corresponding to "car" is extracted from the partial structure of "car" which is the ideal translated sentence partial structure immediately above.

In step S5309, step S53
In step S5304, it is determined whether the input sentence partial structure extracted in step 08 includes an abbreviation in the input sentence abbreviation extracted in step S5304. If there is an abbreviation that includes the input sentence substructure (YE
S), and proceeds to step S5310, otherwise (NO), proceeds to step S5312. In the above example, since there is an abbreviated expression including an input sentence substructure corresponding to “want to buy” and an abbreviated expression including an input sentence substructure corresponding to “car”, the determination is YES, and the determination is YES.
Move to 10.

In step S5310, the structure of the omitted element including the input sentence partial structure extracted in step S5308 determined in step S5309 is the ideal translated sentence portion which is a pronoun expression or a definite article expression extracted in step S5305. Determine if it conflicts with the structure. In the case of contradiction (YES), the flow shifts to step S5312,
If there is no contradiction (NO), the flow shifts to step S5311.

In the above example, the ideal translated sentence partial structure of “I” has the meaning of “subject” and has a nominative relationship with “wantto buy”, while a parallel translation with “want to buy” Since the structure of the omitted element including the partial structure of "want to buy" has an omitted element having the meaning of "subject" which has a nominative relationship with "want to buy", the two are not inconsistent.
It is determined to be O, and the flow shifts to step S5311. In addition, the ideal translation partial structure of “my” has the meaning of “subject” and its referent is “I”, which has a relation of possession with “car”, whereas it has a bilingual relation with “car”. "Cars"
In the structure of the omitted element including the partial structure of, there is an omitted element having an undetermined meaning that has an undetermined relationship with “car”, and therefore the two are not inconsistent.
It moves to 311.

In step S5311, the ideal translation partial structure determined as not inconsistent with the structure of the omitted element in step S5310 is extracted as the target language complement element of the omitted element. In the above example, both the ideal translation substructures of “I” and “my” are extracted as the target language complement elements of the omitted elements that do not contradict each other.

In step S5312, it is determined whether or not an ideal translated sentence substructure that has not been checked remains. Finally, when the inspection is completed for all the ideal translated sentence substructures in the ideal translated sentence analysis structure generated by the ideal translated sentence analysis unit 4, the processing of the target language supplementary element extraction unit 53 is completed.

With respect to the input sentence analysis structure shown in FIG. 7, finally, the target language supplementary element shown in FIG. It should be noted that the relationship between the undetermined omitted element and the "car" in the input sentence omitted element that complements "my" in FIG. 10 is recognized as possessive, which is the relationship between "my" and "car". .

Next, the processing of the source language supplementary element extraction unit 54 will be described. The source language supplementary element extracting unit 54 converts the target language supplementary element extracted by the target language supplementary element extracting unit 53 to generate a source language supplementary element. For conversion from the target language complement element to the source language complement element, a method using a bilingual dictionary of the source language and the target language can be considered.

For each target language complementary element in FIG. 12, a source language complementary element shown in FIG.

As described above, the omitted complementary element pair extraction unit 5
The omitted element complement rule, which is a pair of the input sentence omitted element, the target language complement element, and the source language complement element obtained in the above, is generated by the omitted element complement rule generation changing unit 6 in FIG. The new rule is added to the element complementing rule 32, or the omitted element complementing rule already existing in the omitted element complementing rule 32 is changed. For the omitted complementary element pair of FIG. 13, a rule entry is finally added to the omitted element complementary rule of FIG. 5, as shown in FIG. In the example of FIG. 14, the Japanese verb "buy" has the appearance of "hope", the nominative case of "buy" is omitted, the purpose case of "buy" is "car",
When the ownership of "car" is omitted, a rule has been added that complements "me" to the nomination of "buy" and complements "my" to the ownership of "car".

Note that the omitted element complementing rule generation changing unit 6
Input sentence omitted element extracted by the omitted complementing element pair extraction unit 5,
The abbreviation completion extracted from the set of pairs of the input sentence and the ideal translation sentence that are the target of learning the abbreviation element completion rule, which stores the abbreviation completion element pair that is the pair of the target language completion element and the source language completion element A method of extracting the most consistent and effective omitted element complementing rule using a set of element pairs is also conceivable.

As described above, one process of the omission element complementing rule learning system for the natural language analysis system having the basic block configuration of FIG. 1 is completed. However, when multiple problems with the same sentence are mixed, not all of them can be extracted in one run, and the dictionary or rule that has been changed may be worse than before the addition was changed. There is also. Therefore, it is necessary to set the learning end condition and repeat the omitted element pair extraction unit 5 and the omitted element complementing rule generation changing unit 6 in FIG. 1 until the end condition is satisfied. 1 of learning end judgment
As a proposal, a source language analysis structure history storage unit that holds the history of the source language analysis structure generated by the natural language analysis system unit is provided, and it is checked whether the source language analysis structure history storage unit is in a steady state.

The procedure of the repeatedly executed learning system will be described with reference to the flowchart of FIG. At the first time, learning is performed in the same manner as described above. At that point, FIG.
Step S9 of reading the bilingual sentence of step S9, and step S9 of additionally changing the omitted element complementing rule 32 by the omitted complementary element pair extracting unit 5 and the omitted element complementing rule generating unit 6 shown in FIG.
2 is over. Next, in step S93, the natural language analysis system unit 2 analyzes again using the omitted element complementing rule 32 that has been added and changed in step S92. Next, it is determined whether or not the recorded contents of the analysis structure stored in step S94 as a learning end condition are in a steady state.
At first, it can not be said that it is in a steady state because only the first run has been made. Therefore, the determination is NO, and step S92
Proceed to. Steps S92 and S9 after the second time
3 is omitted until the content of the analysis structure reaches a steady state in step S94.
Two additional modifications are made. As described above, if the omitted complementary element pair extracting unit 5 and the omitted element complementary rule generation changing unit 6 are repeatedly performed, there is no omitted element complementary rule to be extracted from the bilingual sentence, and the analysis structure is in a steady state. And a final omitted element complementing rule is extracted.

The present invention is not limited to the above embodiments, but can be variously modified or applied within the scope of the claims.

To summarize the above, in this embodiment,
First translations that are bilingual with each other by the bilingual sentence reading unit 1
By reading the natural language sentence (input sentence) and the second natural language sentence (ideal translation sentence), and performing the following processing, a new omitted element complementing rule can be automatically created and additionally corrected. .

That is, the omitted complementing element pair extracting unit 5 extracts the omitted word and / or the omitted element which is the expression from the input sentence, and uses the word and / or expression complemented to the omitted element from the ideal translation. An abbreviation, which is a pair of a source language complement element that is a first natural language word and / or expression to be complemented to an abbreviation element and an abbreviation element based on the abbreviation element and the target language complement element, by extracting a certain target language complement element Extract complementary element pairs. The omitted element complementing rule generation changing unit 6
Based on the omitted complementary element pairs extracted by the omitted complementary element pair extraction unit 5, an omitted element complementing rule is generated and changed. This makes it possible to automatically create and change an abbreviation element complementing rule for analyzing an abbreviation in an input sentence in a bilingual sentence with a natural language analysis system.

The omitted supplementary element pair extracting section 5 includes an input sentence omitted element extracting section 52 for extracting an omitted element, which is an element omitted from the input sentence, and a word and / or expression to be complemented from the ideal translation sentence to the omitted element. And a target language complementary element extraction unit 53 that extracts the target language complementary element. As a result, elements that tend to translate abbreviated elements in the source language sentence into sentences in the target language, such as techniques for extracting abbreviated elements from the source language sentence and fixed expressions such as pronoun expressions from the target language sentence, can be used. The extraction technique can be used in the omitted complementary element pair extraction unit 5.

Further, the abbreviation complementary element pair extraction unit 5 inputs words and / or expressions in the input sentence from the input sentence and the ideal translated sentence, and words and / or words in the ideal translated sentence having a bilingual relationship with the input word expression. A translation expression pair extraction unit 51 that extracts an input sentence ideal translation word expression pair that is an ideal translation word expression pair can be provided. This makes it possible to classify words and expressions in the ideal translation having a bilingual relationship with the input sentence into those having a bilingual relationship in the input sentence and those having no bilingual expression in the input sentence. The determination as to whether or not the element should be performed can be made more accurately.
Further, when there is an element in the input sentence that has a bilingual relationship with the element to be complemented by the omitted element in the ideal translation, it is possible to extract an omission phenomenon in which the complementary element is present in the same sentence.

Further, the omitted language supplementary element pair extraction unit 5 is provided with a source language complement element extraction unit 54 for extracting a complement element in the source language based on the target language complement element extracted by the target language complement element extraction unit 53. be able to. This makes it possible to extract the position of the omitted element and the element to be supplemented by the omitted element only in the first natural language that is the source language.

The input sentence analysis unit analyzes the input sentence to generate an input sentence analysis structure, and extracts an omitted element and / or a target language complement element and / or a source language complement element using the input sentence analysis structure. be able to. This makes it possible to extract omitted elements that can be determined for the first time by analyzing the input sentence, and to more accurately use target language complement elements and source language complement elements by using the constraints on elements that can be complemented by these omitted elements that were found as a result of the analysis. Elements can be extracted.

Further, the input sentence analyzing section may include an input sentence omitting element complementing section 22 for complementing an omitted element in the input sentence. As a result, whether the abbreviation element in each input sentence is an abbreviation element that has not been completed yet and requires a new completion rule, or is an abbreviation element that has already been completed but an erroneous element has been completed Thus, it is possible to distinguish whether or not the omitted element that has already been complemented and the correct element has been complemented, and a detailed omitted element complementing rule can be extracted.

In the input sentence analysis unit, a natural language analysis system that uses the omitted element complementing rules extracted by the present learning method for analyzing the input sentence can be used. As a result, the analysis tendency of the natural language analysis system can be directly used for extracting the omitted element and / or the target language complementary element and / or the source language complementary element.

The ideal translated sentence analyzing unit 4 analyzes the ideal translated sentence to generate an ideal translated sentence analyzing structure, and extracts the omitted element and / or the target language complementary element and / or the source language complementary element using the ideal translated sentence analyzing structure. can do. This makes it possible to more accurately extract, from the ideal translation, the candidate for the target language complement element, which is an element that is likely to be a parallel translation element with the omitted element in the input sentence, and the candidate for the source language complement element that has a parallel translation with it. , The omitted element in the input sentence can be extracted more accurately.

In the ideal translated sentence analyzing section 4, an ideal translated sentence omitting element complementing section 42 for complementing an omitted element in the ideal translated sentence is provided.
Can be included. Thereby, even if the element to be complemented to the omitted element in the input sentence is also omitted in the ideal translation, if the ideal translation sentence omitting element complementing section 42 complements the element, the complemented element is omitted from the input sentence. Can be extracted as an element to be complemented.

The ideal translated sentence analyzing unit 4 can include an ideal translated sentence anaphor analyzing unit 43 for determining a target of an anaphor expression such as a pronoun in an ideal translated sentence. Thus, if an element to be supplemented to the omitted element in the input sentence exists in the ideal translation and is an anaphoric expression, the ideal translation analyzer 4
The information of the pointing target of the anaphor expression determined in the above can also be extracted, and a more detailed omitted element complementing rule can be extracted. Also, an element to be supplemented to the omitted element in the input sentence exists in the ideal translation, which is an anaphor expression, and the target in the ideal translation determined by the ideal translation sentence analytic analysis unit 43 is in the same sentence or a sentence. If an element in the input sentence that is present in another sentence and is in parallel with the referent in this ideal translation is determined, the element in the input sentence that is in parallel with this referent is omitted. It can be determined that the element should be complemented, and the in-sentence complement rules and inter-sentence complement rules for the omitted elements can be extracted.

Further, the omitted element complementing rule generation changing unit 6
By using the input sentence analysis structure obtained by the input sentence analysis unit, the omitted element complementing rule can be generated and corrected. As a result, it is possible to generate and change a detailed omitted element complementing rule according to the analysis structure of the input sentence. Furthermore, by using a natural language analysis system that uses the abbreviation element completion rules extracted by this learning method for analysis in this input sentence analysis unit,
By directly utilizing the analysis structure output by the natural language analysis system, it is possible to efficiently generate the omitted element completion rules.

Further, the omission element complementing rule generation / change unit 6 includes an omission element complementation pair recording process of recording the omission complement element pair extracted by the omission complement element pair extraction unit 5 into the omission complement element pair storage unit. Generating and changing an abbreviation element complementing rule using a set of abbreviated complementation element pairs stored in an abbreviated complementation element pair storage unit based on a set of pairs of the input sentence and the ideal translation sentence having a target bilingual relationship. be able to. As a result, taking into account the results of various omitted complementary element pairs extracted from the set of pairs of the input sentence and the ideal translated sentence that are in the bilingual relationship to be learned, the most effective and matching It is possible to extract a good omitted element complement rule.

The input sentence is analyzed using the omitted element completion rule obtained by the omitted element completion rule learning method, and the generated and modified omitted element completion rule can be verified by repeating the omitted element completion rule learning method. Further, it is possible to extract the omitted element complementing rules that cannot be obtained in one learning process.

In the termination condition determination process, the learning termination condition is automatically determined. As a result, the learning is automatically terminated without the person viewing the learning result.

In the source language analysis structure recording processing, the source language analysis structure history, which is the history of the source language analysis structure obtained as a result of analyzing the input sentence using the above omitted element completion rule, is recorded in the source language analysis structure history storage unit. I do. Thus, when the contents of the source language analysis structure history storage unit obtained by the source language analysis structure recording process have reached a steady state, the learning is automatically terminated, and the optimal dictionary and rules are registered.

[0112]

As described above, according to the present invention,
The input sentence and the ideal translation are read, the omitted elements in the input sentence are extracted from the input sentence and the ideal translation, the target language complement element complemented by the omitted element is extracted from the ideal translated sentence, and the target language complement element is extracted based on the omitted element and the target language complement element. The abbreviation element pair, which is a pair of the abbreviation element and the source language complement element to be complemented with the abbreviation element, is extracted, and the abbreviation element completion rule is generated and changed based on the extracted abbreviation complement element pair. As described above, it is possible to automatically accumulate the omitted element complementing rules suitable for complementing the omitted element in the input sentence without manual operation.

Further, by repeating the omitted element complementing rule learning method using the omitted element complementing rule obtained by the omitted element complementary rule generation changing process, the registered omitted element complementing rule can be verified, and one more learning can be performed. It is also possible to extract omitted element complement rules that were not obtained in the process.

Further, the source language analysis structure obtained by the input sentence analysis processing is recorded in the source language analysis structure storage means, so that the contents of the source language analysis structure storage means obtained by the source language analysis structure storage processing are recorded. Is automatically terminated when the steady state is reached, and an optimal omitted element complementing rule can be registered.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a natural language analysis device according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a basic configuration of a natural language analysis apparatus using an omitted and omitted element complementing rule according to an embodiment of the present invention.

FIG. 3 is a diagram showing a configuration of a syntactic and semantic analysis dictionary used in the natural language analysis device of FIG. 2;

FIG. 4 is a diagram showing a configuration of a syntactic and semantic analysis rule used in the natural language analysis device of FIG. 2;

FIG. 5 is a diagram showing a configuration of an omitted element complementing rule used in the natural language analysis device of FIG. 1;

FIG. 6 is a diagram illustrating an example of a bilingual sentence input to the natural language analysis device of FIG. 1;

FIG. 7 is a diagram showing a conceptual structure of an input sentence and an ideal translated sentence in the natural language analysis device of FIG. 1;

8 is a diagram showing an input sentence ideal translation sentence word expression pair in the natural language analysis device of FIG. 1;

FIG. 9 is a flowchart showing processing of an input sentence omitting element extraction unit in the natural language analysis device of FIG. 1;

FIG. 10 is a diagram showing input sentence omitting elements in the natural language analysis device of FIG. 1;

FIG. 11 is a flowchart illustrating a process of a target language supplementary element extraction unit in the natural language analysis device of FIG. 1;

12 is a diagram showing an input sentence omitting element and a target language complementing element in the natural language analysis device of FIG. 1;

13 is a diagram showing an input sentence omitting element, a target language complementing element, and a source language complementing element in the natural language analyzer of FIG. 1;

FIG. 14 is a diagram illustrating an example of an omitted element complementing rule obtained as a result of performing a process of adding and changing an omitted element complementing rule in the natural language analysis device of FIG. 1;

FIG. 15 is a flowchart illustrating a learning process according to another embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Bilingual sentence reading part 2 Natural language analysis system part 3 Rule 4 Ideal translation sentence analysis part 5 Omission complement element pair extraction part 6 Omission element complement rule generation change part 7 First natural language input part 8 First natural language analysis structure output part 21 Input sentence syntactic and semantic analysis unit 22 Input sentence elimination element complement unit 31 Syntax and semantic analysis rule 32 Omission element complement rule 41 Ideal translation sentence syntactic and semantic analysis unit 42 Ideal translation sentence elimination element complement unit 43 Ideal translation sentence analytic analysis unit 51 Bilingual expression pair extraction unit 52 Input sentence elimination element extraction unit 53 Target language complement element extraction unit 54 Source language complement element extraction unit 311 Syntax / meaning analysis dictionary 312 Syntax / meaning analysis rule

Claims (15)

[Claims] 1. An input sentence of a first natural language that is a source language is analyzed, a source language analysis structure that is an analysis structure of the input sentence is generated, and an omitted element complementing rule for complementing an omitted element is generated. An input means for inputting an ideal translation of a target language which is a second natural language having a bilingual relationship with the input sentence, and an input for extracting an omitted element omitted from the input sentence. Sentence omitted element extracting means, target language complementary element extracting means for extracting a target language complementary element to be complemented to the omitted element from the ideal translation, and a target language complementary element extracted by the target language complementary element extracting means. Source-language complementing element extraction means for extracting a source-language complementing element, and an omission-element complementing rule for additionally changing the abbreviation-element complementing rule based on the pair of abbreviation-complementary elements, which is a pair of the target language complement and the source language complement Generate A natural language analysis device comprising: a change unit. 2. An input sentence of a first natural language which is a source language is analyzed, a source language analysis structure which is an analysis structure of the input sentence is generated, and an omitted element complementing rule for complementing an omitted element is generated. A natural language analysis apparatus for changing, wherein reading means for reading the input sentence and an ideal translation of a second natural language which are bilingually related to each other, and an abbreviation element omitted from the input sentence from the input sentence and the ideal translation Is extracted from the ideal translation, and a target language complement element to be complemented by the omitted element is extracted. Based on the omitted element and the target language complement element, the omitted element and the source language complement to be complemented by the omitted element are extracted. An omitted complementing element pair extracting means for extracting an omitted complementary element pair which is a pair of elements, and an omitted element for generating and changing the omitted element complementing rule based on the omitted complementary element pair extracted by the omitted complementary element pair extracting means. Supplement A natural language analysis device comprising: perfect rule generation changing means. 3. An abbreviation complementary element pair extracting means, comprising: an input sentence abbreviation element extracting means for extracting an abbreviated element omitted from the input sentence; and a target language complement element complemented from the ideal translation by the abbreviation element. 3. The natural language analysis device according to claim 2, further comprising a target language complementary element extraction unit for extracting. 4. An abbreviation complementary element pair extracting means, comprising: a pair of an input word expression in the input sentence and an ideal translation word expression in the ideal translation having a bilingual relationship with the input word expression from the input sentence and the ideal translation. 3. A bilingual expression pair extracting means for extracting an input sentence ideal translation word expression pair as follows.
Or the natural language analyzer according to 3. 5. The source language complement element extraction unit for extracting the source language complement element based on the target language complement element extracted by the target language complement element extraction unit, wherein the omitted complement element pair extraction unit comprises: 5. The method according to claim 3, wherein
Natural language analyzer as described. 6. An input sentence analyzing means for analyzing the input sentence read by the bilingual sentence reading means to generate an input sentence analysis structure, and utilizing the input sentence analysis structure, The natural language analysis device according to claim 2, wherein a target language complement element and / or the source language complement element are extracted. 7. The natural language analysis device according to claim 6, wherein the input sentence analyzing means includes an input sentence omitting element complementing means for complementing an omitting element in the input sentence. 8. An ideal translated sentence analyzing means for analyzing the ideal translated sentence read by the bilingual reading means to generate an ideal translated sentence analyzing structure, and utilizing the ideal translated sentence analyzing structure, the omitted element and the object. 3. A language completion element and / or the source language completion element is extracted.
8. The natural language analyzer according to any one of claims 7 to 7. 9. The natural language analysis device according to claim 8, wherein said ideal translation analyzing means includes ideal translation omitting element complementing means for complementing omitting elements in said ideal translation. 10. The natural language analysis device according to claim 8, wherein the ideal translation analysis unit has an ideal translation anaphor analysis unit that determines a target of an anaphor expression such as a pronoun in the ideal translation. . 11. The omission element complementing rule generation and change means generates and changes the omitted element complementation rule using the input sentence analysis structure obtained by the input sentence analysis means. The natural language analysis device according to any one of the above. 12. The omission element complementing rule generation / change unit includes an omission complement element pair storage unit that records the omission complement element pair extracted by the omission complement element pair extraction unit, and stores the omission complement element pair storage unit. The omitted element complementing rule is generated and changed using a set of the omitted complementary element pairs stored in the omitted complementary element pair storage means based on a set of a pair of the input sentence and the ideal translation sentence. The natural language analysis device according to any one of claims 2 to 11, wherein 13. The input sentence omitting element complementing means complements an omitting element in the input sentence by using the omitting element complementing rule, and repeatedly executes an omitting element complementing rule learning method. The natural language analysis device according to any one of the above. 14. The natural language analysis device according to claim 13, wherein said omitted element complementing rule generation changing means has a learning end determining means for determining the end of learning. 15. A source language analysis structure history storage unit that records a source language analysis structure history that is a history of the source language analysis structure that is a result of analyzing the input sentence using the omitted element completion rule, 15. The natural language analysis device according to claim 14, wherein the learning end determination unit ends the learning when the recorded content of the source language analysis structure history storage unit is in a steady state.