JP5245291B2 - Document analysis apparatus, document analysis method, and computer program - Google Patents

Description

  The present invention relates to a document analysis apparatus, a document analysis method, and a computer program. More specifically, the present invention relates to a document analysis apparatus, a document analysis method, and a computer program that execute processing for inputting Japanese document data and correctly analyzing the meaning of each word.

  Specifically, the present invention is applicable to, for example, an automatic translation processing system, and performs document analysis that enables accurate translation by accurately grasping the meaning of character strings constituting Japanese in order to perform correct translation. The present invention relates to a document analysis apparatus, a document analysis method, and a computer program.

  For example, an automatic translation system that translates Japanese into a foreign language analyzes input Japanese data and correctly recognizes the meaning of each character string (word, etc.) and translates it into the corresponding foreign language, such as English. Is required.

  However, Japanese includes ambiguous expressions that can be interpreted as multiple meanings from the character string alone. Some of the nouns that make up Japanese also work as ordinary expressions, resulting in ambiguity in dependency relationships. For example, formal nouns are a group of nouns that can stand on their own by connecting to other words and phrases, for example, “Koto”, “No”, “When”, “Migiri”, “Place”, etc. There is.

As a specific example, a sentence using “place” will be described.
(1) The building was defective when he investigated.
(2) When he investigated, the building was defective.
“Place” in (1) appears as an antecedent of “He surveyed”, but “Place” in (2) is connected to “He surveyed” to summarize the phrases and correspond to adverb phrases It appears as a former that forms a continuous modifying component. This dependency relationship is directly reflected in the difference in meaning. For example, “Place” in (1) means a place, but “Place” in (2) means a situation. Therefore, for example, when interpreting the above Japanese sentence in automatic translation processing, correct translation cannot be performed unless this ambiguity is resolved.

  For example, the following documents include the following as conventional techniques disclosing processing for solving the ambiguity of Japanese meaning and syntax. Patent Document 1 (Japanese Patent Laid-Open No. 5-298349) discloses co-occurrence example statistical information (co-occurrence relationship appearance frequency, appearance frequency of each word, co-occurrence for each word in order to resolve dependency ambiguity. (Number of different words) and weight of co-occurrence cases (strength of co-occurrence relations, degree of freedom of nouns, degree of freedom of predicate) are stored in the co-occurrence relation dictionary, and weights of co-occurrence cases are based on the results In the dependency analysis, a configuration is disclosed in which the likelihood of a dependency candidate matching the co-occurrence case is evaluated based on the weight of the co-occurrence case.

  Further, Patent Document 2 (Japanese Patent Laid-Open No. 6-301716) uses a likelihood model according to a probability model to select an appropriate solution from a plurality of solutions. A configuration is disclosed that enables fine likelihooding and reduces ambiguity by adding the conditional probability of each illuminated solution to the likelihood.

  Furthermore, Patent Document 3 (Japanese Patent Laid-Open No. 2000-330987) inputs a phrase structure analysis result, extracts a reliable dependency relationship between two nouns from the phrase structure analysis result, and modifies each noun based on this. Stores the value of the appearance frequency of the decoration. A configuration that reduces ambiguity by estimating modifiers / modifiers based on the appearance frequency values of other dependency relationships is disclosed.

However, ambiguity arises in the dependency relations such as “Koto”, “No”, “When”, “Migiri”, and “Place” explained earlier. For example, ambiguity such as formal nouns appears. It cannot be solved by frequency or probability model. Because the two sentences mentioned above, namely
(1) The building was defective when he investigated.
(2) When he investigated, the building was defective.
This is because the words appearing in the sentences of (1) and (2) are completely the same, and it is difficult to prioritize and narrow down with the methods such as the appearance frequency and probability model disclosed in the prior art. .
Japanese Patent Laid-Open No. 5-298349 JP-A-6-301716 JP 2000-330987 A

  The present invention has been made in view of the above-described problems, and is a formal noun that causes ambiguity in the dependency relationship such as “to”, “no”, “boom”, “migiri”, “place”, and the like. Provides a document analysis device, document analysis method, and computer program that reduce the ambiguity of syntactic and semantic analysis results by applying predetermined classification rules and processing rules to the noun group called, and grasp the correct meaning The purpose is to do.

The first aspect of the present invention is:
A syntactic and semantic analysis unit for executing syntactic and semantic analysis processing of an input sentence;
A role discriminator for executing a discriminating process of whether the role of the formal noun included in the syntactic and semantic analysis result of the input sentence is an antecedent or an antecedent;
A discrimination result output unit for inputting the discrimination result of the role discrimination unit and outputting a syntax-semantic analysis result according to the discrimination result;
The document analysis apparatus is characterized by comprising:

  Further, in one embodiment of the document analysis apparatus of the present invention, the role discriminating unit is a formal noun dictionary having classification information classified according to whether the formal noun has a strong property as an independent word or a strong property as a functional word. On the basis of the above, it is characterized in that the process of determining whether the role of the formal noun included in the result of the syntactic and semantic analysis is an antecedent or a constructor is executed.

  Furthermore, in an embodiment of the document analysis apparatus of the present invention, the role determination unit associates formal nouns with meaning sentences including words of place, space, part, direction, or direction in a national language dictionary. If the formal noun has a case relationship with the destination in the result of syntactic and semantic analysis, it is determined that the formal noun is a former, otherwise it is an antecedent. It is the structure which performs the process which judges that it is.

  Furthermore, in an embodiment of the document analysis apparatus of the present invention, the document analysis apparatus further includes a formal noun having classification information classified according to whether the formal noun is strong as an independent word or strong as a function word. A dictionary creation unit that creates a dictionary, the dictionary creation unit classifies formal nouns based on part-of-speech information and word-sentence information corresponding to formal nouns of a national language dictionary in which formal nouns are registered; It is the structure which performs creation.

  Furthermore, in an embodiment of the document analysis apparatus of the present invention, the document analysis apparatus further includes a translation processing unit that inputs a discrimination result of the role discrimination unit and executes a translation process according to the discrimination result. Features.

Furthermore, the second aspect of the present invention provides
A document analysis method in a document analysis device,
A syntax and semantic analysis step in which the syntax and semantic analysis unit executes a syntax and semantic analysis process of the input sentence;
A role discriminating step in which the role discriminating unit executes a discriminating process of whether the role of the formal noun included in the syntactic and semantic analysis result of the input sentence is an antecedent or a constructor;
A discrimination result output unit that inputs a discrimination result of the role discrimination unit and outputs a syntactic and semantic analysis result according to the discrimination result; and
The document analysis method is characterized by comprising:

  Furthermore, in one embodiment of the document analysis method of the present invention, the role determining step is performed on the formal noun dictionary having classification information classified according to whether the formal noun is strong as an independent word or strong as a function word. On the basis of this, it is a step of executing a process of discriminating whether the role of the formal noun included in the result of the syntactic and semantic analysis is an antecedent or a constructor.

  Furthermore, in one embodiment of the document analysis method of the present invention, the role determination step associates formal nouns with meaning sentences including words of place, space, part, direction, or direction in the national language dictionary. If the formal noun has a case relationship with the destination in the result of syntactic and semantic analysis, it is determined that the formal noun is a former, otherwise it is an antecedent. It is the step which performs the process which judges that it is.

  Furthermore, in one embodiment of the document analysis method of the present invention, the document analysis method is further classified by the dictionary creation unit according to whether the formal noun is strong as an independent word or as a function word. A dictionary creating step for creating a formal noun dictionary having information, wherein the dictionary creating step classifies formal nouns based on part-of-speech information and semantic text information corresponding to formal nouns in a national language dictionary in which formal nouns are registered. This is a step of creating a formal noun dictionary.

  Furthermore, in an embodiment of the document analysis method of the present invention, the document analysis method further includes a translation process in which a translation processing unit inputs a discrimination result of the role discrimination unit and executes a translation process according to the discrimination result It has a step.

Furthermore, the third aspect of the present invention provides
A computer program for executing document analysis processing in a document analysis device;
A syntax and semantic analysis step for causing the syntax and semantic analysis unit to execute a syntax and semantic analysis process of the input sentence;
A role determination step for causing the role determination unit to execute a determination process as to whether the role of the formal noun included in the syntactic and semantic analysis result of the input sentence is an antecedent or a predecessor;
A discrimination result output step for inputting a discrimination result of the role discrimination unit to a discrimination result output unit and outputting a syntax-semantic analysis result according to the discrimination result;
There is a computer program characterized by comprising:

  The computer program of the present invention is a computer program that can be provided by, for example, a storage medium provided in a computer-readable format in a general-purpose computer system that can execute various program codes. By providing such a program in a computer-readable format, processing corresponding to the program is realized on the computer system.

  Other objects, features, and advantages of the present invention will become apparent from a more detailed description based on embodiments of the present invention described later and the accompanying drawings. In this specification, the system is a logical set configuration of a plurality of devices, and is not limited to one in which the devices of each configuration are in the same casing.

  According to the configuration of the present invention, the syntactic / semantic analysis process of the input sentence is executed, the determination process is executed to determine whether the role of the formal noun included in the syntactic / semantic analysis result is an antecedent or a predecessor. Outputs the semantic analysis result according to the result. In the process of discriminating whether the role of a formal noun is an antecedent or a former, a formal noun dictionary having classification information classified according to whether the formal noun is strong as an independent word or strong as a function word Determine based on. Specifically, when a formal noun is registered in the Japanese language dictionary in association with a word meaning sentence including any word of place, space, part, direction, or direction, or If the semantic analysis result alone has a case relationship with the destination, the formal noun is determined to be a former, and otherwise, it is determined to be an antecedent. Based on this determination result, it is possible to obtain a more accurate syntactic and semantic analysis result or a highly accurate translation result based on this syntactic and semantic analysis result.

  Details of a document analysis apparatus, a document analysis method, and a computer program according to embodiments of the present invention will be described below with reference to the drawings.

  FIG. 1 shows the configuration of a document analysis apparatus according to an embodiment of the present invention. As shown in FIG. 1, the document analysis apparatus 100 includes a sentence input unit 101, a syntax and semantic analysis unit 102, a dictionary creation unit 103, a national language dictionary storage unit 104, a formal noun dictionary storage unit 105, a role determination unit 106, and a determination result output. Part 107. The document analysis apparatus 100 of the present invention is an apparatus that outputs a correct syntactic and semantic analysis result even when a sentence including an ambiguous expression that may be interpreted into a plurality of different meanings is input. Hereinafter, details of each component and processing to be executed will be described.

[Sentence input part]
The sentence input unit 101 inputs a sentence that is a target of syntax and semantic analysis processing. For example, there are various input forms such as a sentence recorded in a database or an input sentence by a user. Hereinafter, as a specific processing example, it is assumed that the following sentence is input.
(Input sentence) “The building was flawed where he investigated.”

  The above sentence includes ambiguous expressions that can be interpreted from a character string alone into a plurality of meanings. In other words, "Place" is a formal noun that can stand on its own by connecting to other words and phrases, and "Place" may mean a place or a situation. is there.

[Syntax / Semantic Analysis]
Next, the syntax and semantic analysis unit 102 performs syntax and semantic analysis of the input sentence. The syntactic and semantic analysis process will be described. Natural languages written in various languages such as Japanese and English are inherently abstract and have high qualities, but they can be processed computerically by handling sentences mathematically. As a result, various applications / services related to natural language are realized by automated processing such as machine translation, dialogue system, search system, and document analysis device. Such natural language processing is generally divided into processing phases of morphological analysis, syntax analysis, semantic analysis, and context analysis.

  In morpheme analysis, a sentence is segmented into morphemes which are the smallest semantic units, and part-of-speech recognition processing is performed. In syntax analysis, sentence structure such as phrase structure is analyzed based on grammatical rules. Since the grammatical rule is a tree structure, the parsing result generally has a tree structure in which individual morphemes are joined based on a dependency relationship. In semantic analysis, a semantic structure that expresses the meaning conveyed by a sentence is obtained based on the meaning (concept) of the words in the sentence and the semantic relationship between words, and the semantic structure is synthesized. In context analysis, a sentence series (discourse) is regarded as a basic unit of analysis, and a discourse structure is constructed by obtaining a semantic group between sentences.

  In the field of natural language processing, syntactic analysis and semantic analysis are considered to be indispensable techniques for realizing applications such as dialog systems, machine translation, document proofreading, and document summarization.

  In the syntax analysis, a natural language sentence is received, and a dependency relationship between words (sentences) is determined based on grammatical rules. The parsing result can be expressed in the form of a tree structure (dependency tree) called a dependency structure. In the semantic analysis, it is possible to perform a process of determining a case relationship in a sentence based on a dependency relationship between words (sentences). The case relationship here refers to a grammatical role such as a subject (SUBJ) and an object (OBJ) possessed by each element constituting a sentence. In addition, semantic analysis may include processing for determining sentence tense, appearance, speech, and the like.

  For examples of syntactic and semantic analysis systems, see, for example, non-patent literature “Masuichi, Okuma,“ Construction of a practical Japanese analysis system based on Lexical Functional Grammar ”, Natural Language Processing, Vol. 79-109 (2003) "is an LFG system. The syntactic and semantic analysis unit 102 of the document analysis apparatus 100 of the present invention can apply a natural language processing system based on this LFG, for example.

  FIG. 2 shows a configuration of a syntax and semantic analysis system 300 that executes natural language processing based on Lexical Functional Grammar (LFG) LFG. The morpheme analysis unit 302 has a morpheme rule 302A and a morpheme dictionary 302B relating to a specific language such as Japanese, and performs a part-of-speech recognition process by segmenting an input sentence into morphemes that are semantic minimum units. For example, if the sentence “My daughter speaks English” is input, the {up} daughter {Noun} of I {Noun} will {up} English {Noun} as {up} as the morphological analysis result. } Speak {Verb1} {tr} Mas {jp}. {Pt} "is output.

  Such a morphological analysis result is then input to the syntax / semantic analysis unit 303. The syntax / semantic analysis unit 303 has dictionaries such as a grammar rule 303A and a valence dictionary 303B, and is based on analysis of phrase structure based on the grammar rule and the like, meaning of words in a sentence, semantic relation between words, and the like. Analyzing the semantic structure expressing the meaning conveyed by the sentence (The valence dictionary describes the relationship between verbs and other components in the sentence such as the subject, and the semantic relation between the predicate and the related word. Can be extracted). As a result of the parsing, the input sentence is questioned based on “c-structure (constituent structure)” representing a phrase structure of a sentence composed of words, morphemes and the like as a tree structure, and a case structure such as a subject and an object. “F-structure (functional structure)” is output as a result of semantically and functionally analyzing sentences, past tense, polite sentences, and the like.

  In other words, c-structure expresses the structure of a natural language sentence as a tree structure by collecting sentence morphemes into upper phrases, and f-structure is based on the concept of grammatical functions. Semantic information such as structure, tense, aspect, and speech is expressed in a matrix structure of attribute-attribute value.

For example, in this example, the input sentence input in the sentence input unit 101 is
(Input sentence) “The building was flawed where he investigated.”
It is. FIG. 3 and FIG. 4 show the f-structure as a result of syntactic and semantic analysis of this input sentence. The f-structure clearly expresses a grammatical function, and includes a grammatical function name, a semantic form, and a feature symbol. By referring to f-structure, it is possible to obtain an understanding of the meaning of a subject, an object, an complement, a modifier, and so on. The f-structure is a set of features attached to each node of the c-structure shown as a tree structure, and is expressed in the form of an attribute-attribute value matrix as shown in FIGS. That is, the left side in [] is a feature (attribute) name, and the right side is a feature value (attribute value).

(Input sentence) “The building was flawed where he investigated.”
The f-structure as the result of the syntactic and semantic analysis for is configured as shown in FIG.

Both FIG. 3 and FIG.
(Input sentence) “The building was flawed where he investigated.”
Although it is f-structure as a syntactic and semantic analysis result corresponding to the same input sentence, it shows the result of performing different analysis on “place” which is a formal noun.
The analysis result of FIG. 3 shows the interpretation that is the object of the verb “investigate” in the embedded phrase because “place” is an antecedent.
The analysis result of FIG. 4 indicates that the object “verb” in the embedded phrase is omitted in this sentence because “place” is a constructor.

[Dictionary creation part]
The dictionary creation unit 103 classifies nouns called formal nouns and stores them in the formal noun dictionary storage unit 105. In general, formal nouns are said to be words that work in the middle of function words such as particles and auxiliary verbs that play a grammatical role and independent words that have a role to indicate something. The dictionary creation unit 103 of the document analysis apparatus 100 of the present invention classifies formal nouns depending on whether they are strong as independent words or strong as functional words.

Here, the formal noun group is a non-patent document “Mizutani,“ The Manuscript Grammar System ”, Tokyo Women's University, Department of Japanese Literature (1991)” p. Assume the following words listed in 104:
“Case”, “Time”, “Claim”, “When”, “Migiri”, “Limit”, “Aida”, “Place”, “House”, “Up”
However, words that are not included above can be added to the formal noun dictionary using this method.

  Taking the “case” as an example, the word classification procedure will be described with reference to the flowchart shown in FIG. 5 and the national language dictionary shown in FIG. First, in S101, an inquiry is made to the national language dictionary storage unit 104 that stores the national language dictionary as shown in FIG. As shown in FIG. 6, the Japanese language dictionary includes various words as “entry words” and describes part-of-speech information, meaning sentences, examples, etc. corresponding to “entry words”. The dictionary creation unit 103 acquires part-of-speech information and word meaning sentences from the national language dictionary. In step S101, for example, part-of-speech information and meaning text corresponding to [Case] are acquired.

  Next, in step S102, it is checked whether or not “participant” or “pronoun” is included in the part of speech. Here, since the part of speech of “case” is only “noun”, the condition of step S102 is not satisfied, and the process proceeds to step S103.

  In step S103, it is checked whether words such as “place”, “space”, “part”, “direction”, and “direction” are included in the meaning part. Here, the meaning of the word “case” is “the situation and the situation when things are being done.” Therefore, the word “part”, “space”, “part”, “direction”, “direction” It is determined that a word such as “is not included. Accordingly, the determination in step S103 is No, and as a result, the “case” is classified into the A group.

  Next, the word classification procedure will be described with reference to the flowchart shown in FIG. 5 and the national language dictionary shown in FIG. First, in step S101, the language dictionary storage unit 104 that stores the language dictionary shown in FIG. In step S <b> 101, the dictionary creation unit 103 acquires part-of-speech information and meaning text corresponding to [AIDA].

  Next, in step S102, it is checked whether or not “participant” or “pronoun” is included in the part of speech. Here, “Aida” is neither a directive nor a pronoun, so the condition is not satisfied. Accordingly, the process proceeds to step S103. In step S103, it is checked whether words such as “place”, “space”, “part”, “direction”, and “direction” are included in the meaning part.

  In step S103, the meaning of “between” is checked. Here, since the meaning sentence of “Aida” is “the part that is sandwiched between two things”, the word “part” is included. Accordingly, the determination in step S103 is Yes, and as a result, “Ama” is classified into the B group.

  Next, taking “other” as an example, a word classification procedure will be described with reference to the flowchart shown in FIG. 5 and the national language dictionary shown in FIG. First, in step S101, the language dictionary storage unit 104 that stores the language dictionary shown in FIG. In step S101, the dictionary creation unit 103 acquires part-of-speech information and meaning text corresponding to [others].

  In step S102, it is checked whether or not “indicator” or “pronoun” is included in the part of speech. Here, since the part of speech of “other” is “noun” and “pronoun”, the condition of step S102 is satisfied, and the determination is Yes, and “other” is classified into the group C. All words are classified by the process as described above to obtain a formal noun dictionary shown in FIG. As shown in FIG. 7, the formal noun dictionary is a dictionary in which formal nouns are classified into three categories (groups A to C). The classification standard is a classification based on whether the formal noun has a strong characteristic as an independent word or a strong characteristic as a functional word. Specifically, according to the flow shown in FIG. 5, Japanese characters corresponding to each formal noun are classified. Classification is based on part-of-speech information, which is registration information for processing, and a word meaning sentence.

  In this way, the dictionary creation unit 103 categorizes nouns called formal nouns that function in the middle of function words such as particles and auxiliary verbs that play a grammatical role, and independent words that have a role of indicating something. A formal noun dictionary is generated by classifying into one category (groups A to C) and stored in the formal noun dictionary storage unit 105. The dictionary creation unit 103 of the document analysis apparatus 100 of the present invention classifies formal nouns depending on whether they are strong as independent words or strong as functional words.

[Role discrimination part]
Next, the role discriminating unit 106 receives the syntax-semantic analysis result (FIGS. 3 and 4) generated by the syntax-semantic analysis unit 102 based on the input sentence, and the syntax-separation analysis result (FIGS. 3 and 4). It is determined whether the included formal noun “Place” is an antecedent or a “former”.

  A processing sequence in the role determination unit 106 will be described with reference to a flowchart of FIG. The role discriminating unit 106 first checks whether or not the noun of interest in step S201 belongs to the group A with reference to the formal noun dictionary 105 (see FIG. 7). The formal noun “Place” belongs to [Group B] as shown in the formal noun dictionary of FIG.

  Accordingly, the determination in step S201 is No, and the process proceeds to step S202. In step S202, it is checked whether or not the formal noun and verb concerned are in a case relationship. Referring to the f-structure as the result of the syntactic and semantic analysis shown in FIG. 3 and FIG. 4, “Place” is the OBL (essential case) of the verb “A”. The determination in S202 is Yes, it is determined that “Place” is [preceding], the analysis result of FIG. 4 is discarded, and the analysis result of FIG. 3 is finally adopted.

That means
(Input sentence) “The building was flawed where he investigated.”
The analysis results of FIG. 3 and FIG. 4 which are f-structures as the result of syntactic and semantic analysis corresponding to the same input sentence as described above, the analysis result of FIG. FIG. 4 shows the interpretation that is the object of the verb “investigate” in the embedded phrase, and FIG. 4 has “place” as a constructor, and the object of the verb “investigate” in the embedded phrase is omitted in this sentence. However, in the role determination processing sequence shown in FIG. 8 in the role determination unit 106, it is determined that the role of “Place” is “precedent”, the analysis result of FIG. 4 is discarded, and FIG. The analysis result is finally adopted.

Note that the role determination unit 106 executes a determination process as to whether the formal noun included in the syntax-semantic analysis result is an antecedent or a constructor according to the flow shown in FIG. As can be understood from the flow shown in FIG. 8 and FIG. 5, the role determination unit 106 selects one of the words of place, space, part, direction, and direction in the Japanese language dictionary for the formal nouns included in the syntactic and semantic analysis result. When a meaning sentence that contains is registered in association with each other, or a formal noun is not registered as such, and the formal noun is not in a case relationship Performs a process of determining that the formal noun is a constructor and determining that it is an antecedent in other cases.

(Processing example 2)
Next, processing when the following sentence is input in the sentence input unit 101 will be described.
(Input sentence) When he investigated, the building was defective.
In the above input sentence, as in the example of processing described above, “where” is a formal noun that can stand on its own by connecting to another word or phrase, and “where” means place. Or it may mean a situation.

[Syntax / Semantic Analysis]
The syntax and semantic analysis unit 102 performs syntax and semantic analysis of the input sentence. The syntactic and semantic analysis processing is executed as processing to which the syntactic and semantic analysis system 300 that executes natural language processing based on the Lexical Functional Grammar (LFG) LFG described above with reference to FIG. 2 is applied.

(Input sentence) When he investigated, the building was defective.
The f-structure as the result of the syntactic and semantic analysis for is configured as shown in FIG. 9 or FIG.
The analysis result of FIG. 9 shows the interpretation that is the object of the verb “investigate” in the embedded phrase because “place” is an antecedent.
The analysis result of FIG. 10 indicates that the object “verb” in the embedded phrase is omitted in this sentence because “Place” is a constructor.

[Role discrimination part]
Next, the role discriminating unit 105 receives the syntax-semantic analysis result (FIGS. 9 and 10) generated by the syntax-semantic analysis unit 102 based on the input sentence, and the syntax-separation analysis result (FIGS. 9 and 10). It is determined whether the included formal noun “Place” is an antecedent or a “former”.

  A processing sequence in the role determination unit 106 will be described with reference to a flowchart of FIG. The role discriminating unit 106 first checks whether or not the noun of interest in step S201 belongs to the group A with reference to the formal noun dictionary 105 (see FIG. 7). The formal noun “Place” belongs to [Group B] as shown in the formal noun dictionary of FIG.

  Accordingly, the determination in step S201 is No, and the process proceeds to step S202. In step S202, it is checked whether or not the formal noun and verb concerned are in a case relationship. Referring to the f-structure as the syntactic and semantic analysis results shown in FIG. 9 and FIG. 10, “Place” is an ADJUNCT (adverb equivalent modification component) of the verb “A”. Not applicable.

  Accordingly, the determination in step S202 is No, it is determined that “Place” is “former”, the analysis result of FIG. 9 is discarded, and the analysis result of FIG. 10 is finally adopted.

That means
(Input sentence) “When he investigated, the building was defective.”
The analysis results of FIG. 9 and FIG. 10, which are f-structures as the result of the syntactic and semantic analysis corresponding to the same input sentence, as described above, the analysis result of FIG. FIG. 10 shows an interpretation that is an object of the verb “investigate” in the embedded phrase, and FIG. 10 has “place” as a constructor, and the object of the verb “investigate” in the embedded phrase is omitted in this sentence. However, in the role determination processing sequence shown in FIG. 8 in the role determination unit 106, it is determined that the role of “Place” is “former”, the analysis result of FIG. 9 is discarded, and FIG. The analysis result is finally adopted.

[Example of document analysis device that executes translation processing]
For example, an automatic translation system that translates Japanese into a foreign language analyzes input Japanese data and correctly recognizes the meaning of each character string (word, etc.) and translates it into the corresponding foreign language, such as English. Is required. However, as explained above, Japanese includes ambiguous expressions that can be interpreted into a plurality of meanings only from a character string. For example, it is a formal noun such as “Koto”, “No”, “When”, “Migiri”, and “Place”.

  By performing the analysis processing by the document analysis apparatus described with reference to FIGS. 1 to 10, it is possible to determine whether a formal noun is a constructor or an antecedent, and using this analysis result The correct translation process becomes possible by translating. Hereinafter, the configuration and processing of a document analysis apparatus that executes translation processing based on the analysis result obtained by the document analysis processing described above will be described.

  FIG. 11 shows the configuration of a document analysis apparatus that executes translation processing. As shown in FIG. 11, the document analysis apparatus 400 according to the present exemplary embodiment includes a sentence input unit 401, a syntax and semantic analysis unit 402, a role determination unit 403, a formal noun dictionary storage unit 404, a translation processing unit 405, and a bilingual dictionary storage unit 406. The translation result output unit 407 is provided. The document analysis apparatus 400 according to the present embodiment outputs a correct translation result based on a correct syntactic and semantic analysis result even when a sentence including an ambiguous expression that may be interpreted into a plurality of different meanings is input. It is.

  In the configuration shown in FIG. 11, the sentence input unit 401, the syntax and semantic analysis unit 402, the role determination unit 403, and the formal noun dictionary storage unit 404 have the same configuration as the configuration shown in FIG. 1 and execute the same processing.

That is, the sentence input unit 401 inputs a sentence to be analyzed and translated, and the syntax and semantic analysis unit 402 performs a syntax and semantic analysis of the input sentence. The syntactic and semantic analysis processing is executed as processing to which the syntactic and semantic analysis system 300 that executes natural language processing based on the Lexical Functional Grammar (LFG) LFG described above with reference to FIG. 2 is applied. The role discriminating unit 403 receives the syntactic and semantic analysis result generated by the syntactic and semantic analyzing unit 402 based on the input sentence, and the formal noun included in the syntactic and semantic analysis result is “precedent” or “former”. Determine if it exists. Less than,
The input sentence is
(Input sentence) “The building was flawed where he investigated.”
Processing for the above sentence will be described. The configuration and processing after the translation unit 405 will be described.

[Translation Department]
The translation unit 405
(Input sentence) “The building was flawed where he investigated.”
Along with the above input sentence, the formal noun “Place” included in the syntax semantic analysis result (FIG. 3 and FIG. 4) generated by the syntax semantic analysis unit 402 based on the input sentence from the role determination unit 403 is “preceding”. Input the discrimination result of whether it is a “former”. That is, as explained above,
(Input sentence) “The building was flawed where he investigated.”
The formal noun “Tokoro” in is received the discrimination result that it is “preceding”.

The translation unit 405 refers to the bilingual dictionary stored in the bilingual dictionary storage unit 406 and generates a bilingual for the input sentence. FIG. 12 shows an example of a bilingual dictionary. As shown in FIG. 12, the bilingual dictionary includes a bilingual translation corresponding to a noun, in this example, an English translation.
(A) a parallel translation when the noun is an antecedent;
(B) a parallel translation when the noun is “former”;
It is configured as a dictionary in which parallel translations in the cases (a) and (b) are registered.

According to the dictionary
If "Place" is an antecedent, the translation is "place"
If "Place" is "Former", the parallel translation is "when"
It is registered as.

The translation unit 405 receives the role determination unit 403 from
(Input sentence) “The building was flawed where he investigated.”
The formal noun “Tokoro” has received the discriminant result that it is “preceding”, and according to this discriminating result, “Tokoro” is judged to be “preceding”, "Place" is obtained as a translation of "." The translation result thus executed is output via the translation result output unit 407.

  In the document analysis apparatus 400 of the present embodiment, an ambiguous and plural interpretable formal noun is analyzed, and it is determined whether the formal noun is “precedent” or “former”. Since the translation is performed based on the determination result, correct translation is possible, and a highly accurate and highly accurate translation result can be output.

  Finally, an example of the hardware configuration of the information processing apparatus constituting the document analysis apparatus that executes the above-described processing will be described with reference to FIG. A CPU (Central Processing Unit) 501 performs processing corresponding to an OS (Operating System), syntactic and semantic analysis processing based on the input sentence described in the above-described embodiment, formal noun dictionary creation processing, and whether the formal noun is an antecedent. A role discrimination process, a translation process, and the like are executed to execute a discrimination process as to whether it is a creator. These processes are executed according to a computer program stored in a data storage unit such as a ROM or a hard disk of each information processing apparatus.

  A ROM (Read Only Memory) 502 stores programs used by the CPU 501, calculation parameters, and the like. A RAM (Random Access Memory) 503 stores programs used in the execution of the CPU 501, parameters that change as appropriate during the execution, and the like. These are connected to each other by a host bus 504 including a CPU bus.

  The host bus 504 is connected to an external bus 506 such as a PCI (Peripheral Component Interconnect / Interface) bus via a bridge 505.

  A keyboard 508 and a pointing device 509 are input devices operated by the user. The display 510 includes a liquid crystal display device, a CRT (Cathode Ray Tube), or the like, and displays various types of information as text and images.

  An HDD (Hard Disk Drive) 511 includes a hard disk, drives the hard disk, and records or reproduces a program executed by the CPU 501 and information. The hard disk is used, for example, for storing various dictionary data such as a national language dictionary, formal noun dictionary, bilingual dictionary and the like, and further stores various computer programs such as a data processing program.

  The drive 512 reads data or a program recorded on a removable recording medium 521 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, and the data or program is read out from the interface 507 and the external bus 506. , And supplied to the RAM 503 connected via the bridge 505 and the host bus 504.

  The connection port 514 is a port for connecting the external connection device 522 and has a connection unit such as USB or IEEE1394. The connection port 514 is connected to the CPU 501 and the like via the interface 507, the external bus 506, the bridge 505, the host bus 504, and the like. The communication unit 515 is connected to a network and executes communication with a client and a network connection server.

  Note that the hardware configuration example of the information processing apparatus applied as the document analysis apparatus illustrated in FIG. 13 is an example of an apparatus configured by applying a PC, and the document analysis apparatus of the present invention has the configuration illustrated in FIG. The configuration is not limited as long as the processing described in the above-described embodiments can be executed.

  The present invention has been described in detail above with reference to specific embodiments. However, it is obvious that those skilled in the art can make modifications and substitutions of the embodiments without departing from the gist of the present invention. In other words, the present invention has been disclosed in the form of exemplification, and should not be interpreted in a limited manner. In order to determine the gist of the present invention, the claims should be taken into consideration.

  The series of processing described in the specification can be executed by hardware, software, or a combined configuration of both. When executing processing by software, the program recording the processing sequence is installed in a memory in a computer incorporated in dedicated hardware and executed, or the program is executed on a general-purpose computer capable of executing various processing. It can be installed and run. For example, the program can be recorded in advance on a recording medium. In addition to being installed on a computer from a recording medium, the program can be received via a network such as a LAN (Local Area Network) or the Internet and can be installed on a recording medium such as a built-in hard disk.

  Note that the various processes described in the specification are not only executed in time series according to the description, but may be executed in parallel or individually according to the processing capability of the apparatus that executes the processes or as necessary. Further, in this specification, the system is a logical set configuration of a plurality of devices, and the devices of each configuration are not limited to being in the same casing.

  As described above, according to the configuration of the present invention, the syntax / semantic analysis processing of the input sentence is executed, and it is determined whether the role of the formal noun included in the syntax / semantic analysis result is an antecedent or a predecessor. Execute the process and output the syntax and semantic analysis result according to the discrimination result. In the process of discriminating whether the role of a formal noun is an antecedent or a former, a formal noun dictionary having classification information classified according to whether the formal noun is strong as an independent word or strong as a function word Determine based on. Specifically, when a formal noun is registered in the Japanese language dictionary in association with a word meaning sentence including any word of place, space, part, direction, or direction, or If the semantic analysis result alone has a case relationship with the destination, the formal noun is determined to be a former, and otherwise, it is determined to be an antecedent. Based on this determination result, it is possible to obtain a more accurate syntactic and semantic analysis result or a highly accurate translation result based on this syntactic and semantic analysis result.

It is a figure explaining the structural example of the document analysis apparatus which concerns on one Embodiment of this invention. It is a figure explaining the process example and structural example of a syntax semantic analysis part in the document analysis apparatus which concerns on one Embodiment of this invention. It is a figure explaining the example of f-structure as a syntax semantic analysis result produced | generated in the document analysis apparatus concerning one Embodiment of this invention. It is a figure explaining the example of f-structure as a syntax semantic analysis result produced | generated in the document analysis apparatus concerning one Embodiment of this invention. It is a figure which shows the flowchart explaining the process sequence of the dictionary creation part in the document analyzer which concerns on one Embodiment of this invention. It is a figure explaining the structural example of the national language dictionary applied to the process of the dictionary creation part in the document analyzer which concerns on one Embodiment of this invention. It is a figure explaining the structural example of the formal noun dictionary produced | generated by the process of the dictionary creation part in the document analyzer which concerns on one Embodiment of this invention. It is a figure which shows the flowchart explaining the processing sequence of the role discrimination | determination part in the document analysis apparatus concerning one Embodiment of this invention. It is a figure explaining the example of f-structure as a syntax semantic analysis result produced | generated in the document analysis apparatus concerning one Embodiment of this invention. It is a figure explaining the example of f-structure as a syntax semantic analysis result produced | generated in the document analysis apparatus concerning one Embodiment of this invention. It is a figure explaining the structural example of the document analysis apparatus which concerns on one Embodiment of this invention. It is a figure explaining the structural example of the parallel translation dictionary applied to the process of the translation process part in the document analyzer which concerns on one Embodiment of this invention. It is a figure explaining the hardware structural example of the document analysis apparatus which concerns on one Embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Document analysis apparatus 101 Sentence input part 102 Syntax semantic analysis part 103 Dictionary preparation part 104 Japanese language dictionary storage part 105 Formal noun dictionary storage part 106 Role discrimination part 107 Discrimination result output part 300 Syntax semantic analysis system 302 Morphological analysis part 302A Morphological rule 302B Morphological dictionary 303 Syntax and semantic analysis unit 303A Grammar rule 303B Value valence dictionary 400 Document analysis device 401 Sentence input unit 402 Syntax and semantic analysis unit 403 Role discrimination unit 404 Formal noun dictionary storage unit 405 Translation processing unit 406 Bilingual dictionary storage unit 407 Translation result output 501 CPU (Central Processing Unit)
502 ROM (Read-Only-Memory)
503 RAM (Random Access Memory)
504 Host bus 505 Bridge 506 External bus 507 Interface 508 Keyboard 509 Pointing device 510 Display 511 HDD (Hard Disk Drive)
512 drive 514 connection port 515 communication unit 521 removable recording medium 522 external connection device

Claims (7)

A syntactic and semantic analysis unit for executing syntactic and semantic analysis processing of an input sentence;
A role discriminator for executing a discriminating process of whether the role of the formal noun included in the syntactic and semantic analysis result of the input sentence is an antecedent or an antecedent;
A discrimination result output unit for inputting the discrimination result of the role discrimination unit and outputting a syntax-semantic analysis result according to the discrimination result;
I have a,
The role discriminating unit is a formal noun dictionary in which a group of formal nouns that have strong properties as function words and are to be discriminated as they are are registered as distinctive nouns of the first group so that they can be distinguished from other formal nouns. The group of formal nouns is identified as a generator, and the formal nouns that are not the group of formal nouns are identified as a constructor when the formal noun is not in a case-related relationship. A document analyzing apparatus characterized by discriminating an antecedent when a character is in a case relationship with a destination . The document analysis device further includes:
Has a dictionary creation unit that creates the form noun dictionary,
The dictionary creation unit
A formal noun is registered in the national language dictionary without a synonym sentence that does not include a pronoun or indicator as part of speech classification, and the national language dictionary contains a word in place, space, part, direction, or direction. 2. The document analysis apparatus according to claim 1, wherein the formal noun is registered in the formed noun dictionary as the formal noun of the first group . The document analysis device further includes:
The document analysis apparatus according to claim 1, further comprising a translation processing unit that inputs a discrimination result of the role discrimination unit and executes a translation process according to the discrimination result. A document analysis method in a document analysis device,
A syntax and semantic analysis step in which the syntax and semantic analysis unit executes a syntax and semantic analysis process of the input sentence;
A role discriminating step in which the role discriminating unit executes a discriminating process of whether the role of the formal noun included in the syntactic and semantic analysis result of the input sentence is an antecedent or a constructor;
A discrimination result output unit that inputs a discrimination result of the role discrimination unit and outputs a syntactic and semantic analysis result according to the discrimination result; and
I have a,
The role discrimination step includes a formal noun dictionary in which a group of formal nouns that have strong properties as function words and are to be discriminated as they are are registered as distinctive nouns of the first group so that they can be distinguished from other formal nouns. The group of formal nouns is identified as a generator, and the formal nouns that are not the group of formal nouns are identified as a constructor when the formal noun is not in a case-related relationship. A document analysis method characterized by discriminating an antecedent when the character is in a case relationship with the relationship . The document analysis method further includes:
Dictionary creation unit has a dictionary generating step of generating the form noun dictionary,
The dictionary creation unit
A formal noun is registered in the national language dictionary without a synonym sentence that does not include a pronoun or indicator as part of speech classification, and the national language dictionary contains a word in place, space, part, direction, or direction. 5. The document analysis method according to claim 4 , wherein the formal noun is registered in the formed noun dictionary as the formal noun of the first group . The document analysis method further includes:
5. The document analysis method according to claim 4 , further comprising: a translation processing step in which a translation processing unit inputs a discrimination result of the role discrimination unit and executes a translation process according to the discrimination result. A computer program for executing document analysis processing in a document analysis device;
A syntax and semantic analysis step for causing the syntax and semantic analysis unit to execute a syntax and semantic analysis process of the input sentence;
A role determination step for causing the role determination unit to execute a determination process as to whether the role of the formal noun included in the syntactic and semantic analysis result of the input sentence is an antecedent or a predecessor;
A discrimination result output step for inputting a discrimination result of the role discrimination unit to a discrimination result output unit and outputting a syntax-semantic analysis result according to the discrimination result;
I have a,
The role discrimination step includes a formal noun dictionary in which a group of formal nouns that have strong properties as function words and are to be discriminated as they are are registered as distinctive nouns of the first group so that they can be distinguished from other formal nouns. The group of formal nouns is identified as a generator, and the formal nouns that are not the group of formal nouns are identified as a constructor when the formal noun is not in a case-related relationship. A computer program characterized by discriminating it as an antecedent when it is in a case relationship with a destination .

Images