JPH0844741A - Device and method for analyzing tense of japanese sentence - Google Patents

Abstract

PURPOSE:To analyze the omitted tense of a TAIGEN (the sustentives) termination sentence by analyzing the tense by utilizing the interaction between the tense in a subordinate clause and the tense in a main clause. CONSTITUTION:A TAIGEN termination extraction part 100 checks whether the final word of the sentence ending paragraph of an inputted document excluding a punctuation mark ',' is a noun or not and decides a TAIGEN termination sentence when there is the final word terminating with the noun. When the TAIGEN termination extraction part 110 extracts the TAIGEN termination sentence, a time expression extraction part 120 collates syntax analytic result information on the sentence to be processed with a time expression pattern table 140 and outputs all time expressions together with its classification information. A predicate sentence tense analysis part 130 applies a verb predicate sentence time analysis part 131 for a TAIGEN termination sentence decided as a verb predicate sentence or a noun sentence tense analysis part 132 for a TAIGEN termination sentence decided as a noun predicate sentence to output the result. Further, an analysis is taken by using context information as well.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a Japanese sentence tense analysis device and a Japanese sentence tense analysis method, and in particular, when the sentence ending clause of an input Japanese sentence is a noun sentence ending sentence, it is omitted. The present invention relates to a Japanese sentence tense analysis device and a Japanese sentence tense analysis method for analyzing tense.

[0002]

2. Description of the Related Art Japanese sentence analysis technology in a conventional document processing system includes extraction of wording expressions, tense analysis,
The following techniques are being considered for supplementing omitted expressions. As the first conventional method, in the technique of extracting wording-stop expressions, if there are "ga (wa)" and "wo" cases in the sentence as part of the syntactic analysis, the wording of the sahen noun in the sentence Using the semantic constraint of the case element that modifies the verb that the part of speech of the stop is a verb, it is determined whether or not the case element in front of the observed word (noun) modifies the word. Techniques for extracting word-of-speech and its part-of-speech by using are being investigated (Kudada et al., "Sahen noun + part-of-speech discrimination of reading point", 42nd National Convention of IPSJ, 1
991).

As a second conventional method, in the technique of analyzing tenses, it is premised that expressions (such as "ta" and "teru") representing tenses appear in sentences, mainly in machine translation systems and the like. As a result, a technology for analyzing whether they represent "past" or "currently completed" is being studied (Shirai et al .: "Translation of tense in newspaper article translation", IEICE Autumn National Convention, 1990
Year).

As a third conventional method, as an abbreviation in Japanese sentences, for example, "Please tell me the deadline for application" and "Already closed", "Application" before "deadline" Is omitted. As a technique that complements such abbreviated expressions, targeting case elements of predicates, using contextual information such as the type of subordinate clauses, in-sentence information, connectives and vocabulary cohesion with preceding and following sentences,
Techniques for estimating omitted expressions have been studied (Kudo et al .: "Context Processing Mechanism for Dialogue Translation Systems and Its Performance Evaluation", IPSJ Transactions, Vol. 33, No. 2, 199).
2 years).

[0005]

However, the above-described document processing system according to the prior art has the following problems when processing a Japanese sentence such as a newspaper article in which a false stop sentence frequently appears. Like the first conventional method described above, it is possible to correctly extract the wording stop by the technology for extracting the wording stop expression and improve the accuracy of the syntactic analysis process. However, a document processing application system such as a machine translation system. It is not possible to extract the tense information needed in. That is, when translating from Japanese into a foreign language, it is impossible to translate the tense expression while it is ambiguous. Therefore, it is essential to extract the tense information.

Further, since the technique of tense analysis shown in the second conventional method is based on the premise that an expression representing tense appears, a sentence in which the expression representing tense is omitted, such as an improper sentence, is processed. You cannot analyze the tense because it is not subject to. Furthermore, even with the technique of complementing omitted expressions like the third method in the past, the supplementation of tenses in word-stop sentences is not subject to consideration, so tense expressions omitted using this technique are not considered. Cannot be complemented.
In addition, the tense analysis method of a word stop sentence that uses in-sentence information analyzes the tense of the word stop sentence based on the information of the time expression and the predicate expression in one sentence. Cannot be correctly parsed for sentences whose tense cannot be recognized unless the preceding and following sentences appear.

The present invention has been made in view of the above points, and enables the analysis of the tense in which a word-stop sentence is omitted even when there is no expression representing time in the sentence, and appears in the sentence. Japan for analyzing time expressions and their types, end-sentence type, and relations between subordinate clauses and main clauses in compound sentences, and abbreviated tenses of word-stop sentences that could not be analyzed in the past. An object is to provide a language sentence tense analysis device and a Japanese sentence tense analysis method.

A further object of the present invention is to supplement the analysis method of the tense of the sentence stop using the in-sentence information, refer to the tense of the sentences before and after, and to predicate the sentence stop sentence and the subsequent sentence. We provide a Japanese sentence tense analysis device and a Japanese sentence tense analysis method that can analyze the tense of a body stop sentence that does not include an expression that indicates tense in the sentence by analyzing the parallelism and transitivity of movements. The purpose is to do.

[0009]

FIG. 1 is a block diagram showing the principle of the present invention. The present invention, in a Japanese sentence tense analysis device for analyzing the tense of a Japanese sentence in a Japanese sentence processing system for inputting and processing a Japanese sentence, when the result of parsing the Japanese sentence to be processed is input, In addition to the first analysis unit 100 and the first analysis unit 100, which analyzes the tense in which a sentence-fixed sentence in which a sentence ending sentence of a Japanese sentence ends with a noun is omitted, in addition to the first analyzing unit 100, context information It has the 2nd analysis means 800 analyzed by.

In the first analysis means 100, the extraction means 110 for extracting a word stop sentence from a Japanese sentence and the word stop phrase that is the last sentence clause of the word stop sentence is a verb predicate sentence or a noun predicate sentence. Predicate determination means 115 for determining
And for each verb predicate sentence or noun predicate sentence judged by the predicate judging means 115, analyze the syntactic structure of the Japanese sentence,
And a tense analysis means 130 using a syntactical pattern that outputs a tense analysis result of an injunction sentence.

Further, the tense analysis means 130 utilizing the above-mentioned syntactical pattern extracts the time expression pattern storage means for storing the expression pattern representing time and the time expression appearing in the wording stop sentence extracted by the extraction means 110. And a time expression extraction means for comparing with the time expression pattern storage means and outputting classification information indicating whether the time expression expresses an absolute or relative tense.

The predicate determination means 115 determines whether or not there is a phrase that includes an adverb, a noun, a connecting particle, or a phrase that is a case element of a verb in all the phrases related to the noun of the phrase to stop the sentence. It has a judgment means which judges whether the clause is used as a verb predicate sentence or a noun predicate sentence. Further, the tense analysis means 130 using the above-mentioned syntactic pattern, when the noun storage means that stores nouns representing the past and the predicate determination means 115 determine that the noun phrase is a verb predicate sentence, the noun sentence By analyzing the syntactic structure of the noun phrase, the type of noun phrase, the time expression, and the classification information based on the matching result of the noun of the noun phrase and the existing noun storage means, the tense in which the noun phrase is omitted is analyzed. Verb predicate sentence tense analysis means to analyze,
When the predicate determination unit 115 determines that the nomination phrase is a noun predicate sentence, the tense in which the nomination phrase is omitted is analyzed by analyzing the syntactic structure, time expression, and classification information of the nomination phrase. And a noun predicate sentence tense analyzing means.

Further, the above-mentioned second analysis means 800 is
In addition to the extraction unit 110 of the first analysis unit, the predicate determination unit 115, and the tense analysis unit 130 using a syntactic pattern, a context information storage unit 300 that holds information in sentence units, and a wording sentence After that, a tense analysis rule 500 describing a rule for analyzing the tense of the word stop from the syntax pattern of the sentence, the analysis result of the word stop sentence acquired from the first analysis means 100, the context information storage means 300, and the tense analysis rule. It further includes a tense analysis unit 400 corresponding to a tense analysis unit that collates 500 and analyzes the tense of the noun sentence based on the parallelism of the predicate of the noun sentence and the predicate of the next sentence and the transitivity of the verb.

According to the present invention, in a Japanese sentence tense analysis method for analyzing the tense of a Japanese sentence in a document processing system for inputting and processing a Japanese sentence, the result of parsing the Japanese sentence to be processed is inputted. And, the tense in which the sentence stop sentence in which the sentence end sentence in the Japanese sentence ends with a noun is omitted is analyzed by the property of the sentence end sentence. FIG. 2 is a flow chart for explaining the principle of the present invention.

Further, the Japanese sentence tense analysis device of the present invention is
When the syntax analysis result of a Japanese sentence is input (step 1), a sentence stop sentence in which the sentence ending phrase ends with a noun is extracted from the Japanese sentence (step 2), and the sentence sentence that is the sentence ending sentence clause is extracted. It is determined whether the stop clause is a verb predicate sentence or a noun predicate sentence (step 3), and for each determined verb predicate sentence or noun predicate sentence, the syntactic structure of the verbal stop sentence is analyzed (step 4), The tense analysis result of the improper sentence is output (step 5).

Further, after extracting the wording stop, the time expression appearing in the word stop sentence is extracted and collated with a predetermined pattern to output the classification information indicating whether the time expression expresses an absolute or relative tense. To do. In addition, when determining whether the extracted noun sentence is a verb predicate sentence or a noun predicate sentence, an adverb in all the phrases related to the noun of the noun sentence phrase,
Judgment is made based on the presence or absence of bunsetsu including verbs and connective particles, and bunsetsu that is a case element of verbs.

Further, when it is determined that the noun phrase is a verb predicate sentence, the noun of the noun phrase is compared with the existing noun storage means that stores a noun representing the past, and the same noun is registered. Is analyzed, and by analyzing the syntactic structure of the word stop sentence and the time expression and classification information in the word stop sentence, the tense in which the word stop clause is omitted is analyzed, while the word stop sentence is a noun predicate sentence. When it is determined that there is the sentence stop sentence, the tense in which the sentence stop sentence is omitted is analyzed by analyzing the syntactic structure of the sentence stop sentence, the time expression, and the classification information.

In the above Japanese sentence tense analysis method, when the tense cannot be parsed because the Japanese sentence does not include an expression that indicates the tense (step 6), refer to the tenses before and after the sentence stop sentence. Then, the tense of the word stop sentence is analyzed by analyzing the parallelism of the predicates and the transitivity of the action between the word stop sentence and the subsequent sentence of the word stop sentence (step 8).

That is, when the syntactic analysis result of a Japanese sentence is input, the syntactic analysis result is held as context information in the order of input, and when the Japanese sentence includes a wording sentence, first the time expression and the syntax in the wording sentence are written. If a pattern is used to analyze the tense of a word stop sentence, and if the information in the word stop sentence cannot be analyzed, a rule to analyze the tense using the syntactic information of the sentences before and after the word stop sentence is used. The tense of the sentence stop sentence is determined by reference.

[0020]

According to the invention of claim 1, since the tense of the word stop which is omitted in the sentence is determined based on the property (verb / noun) of the sentence ending clause of the word stop, and the input context is also referred to. It is possible to analyze from both aspects of bunsetsu character and context of sentences. According to the invention of claim 2, when the sentence ending clause is analyzed based on the property of the sentence ending clause, it is determined whether the sentence ending clause of the sentence ending sentence ends with a verb predicate or a noun predicate, so that the subsequent sentence tense is omitted. The analysis of is distributed according to the property of the predicate. This makes it possible to distinguish and analyze expressions unique to verb predicates and noun predicates.

According to the inventions of claims 3 and 5, it is possible to analyze the tense by acquiring the time expression appearing in the injunction sentence even if it is a relative expression and matching it with the time expression pattern. Is. As a result, it is possible to analyze the abbreviated tense of tense in consideration of not only the tense of the main clause but also the tense of the subordinate clause, that is, considering the difference between the tense of the dependent clause and the tense of the main clause. it can.

According to the fourth aspect of the invention, since the wording clause is divided into a verb predicate sentence or a noun predicate sentence and analyzed for each syntactic structure, an expression that refers to an object or a matter and an expression that refers to an action are separated. It is possible to analyze separately. According to the fifth aspect of the invention, when the extracted wording sentence is a noun predicate sentence, it is determined whether the extracted wording sentence is a compound sentence or a simple sentence, and the tense and the time expression of the subordinate clause are the main clauses. We analyze the tense of a word-stop sentence by performing syntactic analysis to determine which subordinate clause is modified and pattern analysis of time expression. Furthermore, when the extracted wording clause is a verb predicate sentence, it is checked whether the same noun as the wording clause is present in the registered noun storage means, and if the corresponding word exists, the The tense of the noun phrase can be analyzed by quoting the tense of the noun.

The analysis device according to any one of claims 1 to 5 uses the information in the sentence to analyze the tense.
According to the invention, when the in-sentence information cannot be analyzed, the sentence is analyzed by referring to the context of the input syntactic analysis result. When the result of the syntactic analysis of the Japanese sentence to be processed is input, the syntactic information input by the analysis control means is registered in the context information storage means, and if the sentence stop sentence is included, the context information storage means By referring to the contents, the tenses of the sentences before and after the sentence stop sentence are acquired, and the tense of the sentence stop sentence is determined.

According to the inventions of claims 7 to 11, the tense is analyzed using one in-sentence information, and even if there is no keyword expressing the tense in the sentence due to the relation between the tense of the main clause and the subordinate clause. It enables tense analysis. According to the invention of claim 12, unlike the inventions of claims 7 to 11, the tense is not analyzed in the sentence, but the relevance of the context with the syntax before and after the wording is stopped according to the results of a plurality of input syntactic analyzes. Since the tense is analyzed by, it is effective when the tense cannot be analyzed from the information in the sentence.

According to the thirteenth aspect of the present invention, when the tense analysis cannot be performed using the syntactic information of the preceding sentence and the tense analysis rule of the impersonation sentence, the constraint of each memory described in the syntactic information of the next sentence and the tense analysis rule. By analyzing the tense of the word-stop sentence by comparing with the condition. In this way, it becomes possible not only to analyze the tense based on the information in one sentence, but also to analyze the tense of a sentence in which the tense expression is omitted due to the contexts before and after the word stop sentence.

[0026]

Embodiments of the present invention will now be described in detail with reference to the drawings. [First Embodiment] It is assumed that the Japanese sentence input to this device has been subjected to a syntactic analysis process. The syntactic analysis uses the result of morphological analysis, which divides a Japanese sentence written in Kanji and Kana into word units, and analyzes the relations between the clauses and the modifiers that compose the Japanese sentence. This is a process of recognizing the type of relationship. Generally, a Japanese document processing system is provided with processing functions of morphological analysis and syntactic analysis, and the results thereof are input.

FIG. 3 is a system configuration diagram of the first embodiment of the present invention. The system shown in the figure includes an in-sentence information use type tense analysis unit 100, a time expression pattern table 140,
It is composed of a preexisting noun table 150 and an analysis control unit 200. FIG. 4 shows the configuration of the Japanese sentence tense analyzing apparatus according to the first embodiment of the present invention. The Japanese sentence tense analysis device shown in the figure corresponds to the sentence information utilizing type tense analysis unit 100 in FIG.

The Japanese sentence tense analysis device, when a Japanese sentence is input, extracts a sentence stop sentence in which the sentence ending clause ends with a noun from the input sentence. A predicate determination unit 115 that determines whether a sentence or a noun predicate sentence, a time expression extraction unit 120 that outputs the time expression in the input sentence and its type, and a verb predicate sentence or a noun predicate sentence based on the determination result of the predicate determination unit 115. The predicate sentence tense analysis unit 130 that analyzes and outputs the abbreviated tense of each wording stop clause, the time expression pattern table 140 having the expression pattern representing time and its classification information, and the noun representing the past are stored. It is composed of the noun table 150.

The verbal stop extraction unit 110 checks whether or not the final word of the sentence ending clause excluding the punctuation "." From the input sentence is a noun, and if there is a final word ending with a noun, the verbal stop sentence is detected. To determine. When the predicate determination unit 115 determines that the sentence is a sentence stop sentence, the predicate determination unit 115 targets all the clauses that modify the final clause and includes a clause that includes an adverb or a verb, or a case element that includes a clause or a verb. It is determined whether there is a clause that becomes, and if there is, the subject stop sentence is determined to be a verb predicate sentence, and if not, it is determined to be a noun predicate sentence.

The time expression extraction unit 120 collates the syntactic analysis result information of the processing target sentence with the time expression pattern table 140 when the word expression stop extraction unit 110 extracts a word expression stop sentence, and all time expressions are expressed. Is output together with the classification information. The predicate sentence tense analysis unit 130 uses the verb predicate sentence tense analysis unit 131 for the noun sentence that is determined to be a verb predicate sentence by the predicate determination unit 115, and the noun sentence sentence that is determined to be a noun predicate sentence. Applies the noun predicate sentence tense analysis unit 132 and outputs the result.

The verb predicate sentence tense analysis unit 131 uses the noun table 150, the syntactic analysis result information, and the time expression extraction result for the verbal predicate sentence determined to be a verbal predicate sentence. From the appearance pattern of, the tense in which the wording stop clause is omitted is determined. The noun predicate sentence tense analysis unit 132 uses the syntactic analysis result information and the time expression extraction result for the noun predicate sentence determined as a noun predicate sentence, and a sentence in which the noun predicate sentence is described with respect to time. From the analysis of whether or not and the appearance pattern of the time expression,
Judge the tense in which the word-stop clause is omitted.

FIG. 5 shows a flow chart of the operation of the wording stop extraction unit and the predicate determination unit of the first embodiment of the present invention. (Step 110) The word stop extractor 100 determines whether or not the word in the final phrase excluding the punctuation from the input Japanese sentence is a noun, and if the word is a noun, it is extracted as a word stop sentence. On the other hand, if it is other than a noun, the Japanese sentence is not a sentence stop sentence, so the process proceeds to step 150.

(Step 120) The phrase stop extracting unit 11
If 0 determines that the word in the final phrase is a noun, the predicate determining unit 115 determines whether the noun is a sa variant noun, and if it is not a sa variant no, proceeds to step 160. Transition. (Step 130) If the word in the final phrase is a sa variant noun, it is determined whether or not there is an adverb or a verb in the phrase related to the end phrase in the extracted sentence stop sentence. Move to 170.

(Step 140) Step 130
In, when there is no adverb or verb in the phrase related to the end-of-sentence phrase, the end-of-sentence word is a sub-particle or a case particle “from”, “de”, or “to” in the phrase related to the end-of-sentence phrase.
It is determined whether there is a "to", "ni", "yori", "wo", or a connection particle. If none of the auxiliary particle, the case particle, or the connecting particle is present, the process proceeds to step 160, and if any of them exists, the process proceeds to step 170.

(Step 150) Step 11 above
In 0, when the final word is a part of speech other than a noun, the determination result that the sentence is not a sentence stop sentence is output. (Step 160) In the above step 120, the predicate determination unit 115 determines that the final word of the extracted sentence-fixing sentence is not a SA variant noun, and in step 140, the phrase ending word in the phrase related to the sentence ending phrase is a sub particle. If none of the case particles and the connecting particles are present, it is determined that the noun predicate sentence is a sentence stop sentence.

(Step 170) The above step 130
In the case where there is an adverb or a verb in the phrase related to the end-of-sentence phrase, and in step 140, the end-of-sentence word in the phrase related to the end-of-sentence phrase is any of a sub-particle, a case particle, and a connective particle. If it exists, it is determined to be a verbal predicate sentence stop sentence. Next, when the input Japanese sentence is a sentence stop sentence, the time expression extraction unit 120
Refers to the time expression pattern table 140 and extracts, as information used to analyze the omitted tense of the wording sentence, an expression representing time and its classification pattern from the wording sentence.

FIG. 6 is a conceptual diagram of the time expression pattern table 140 according to the first embodiment of this invention. The time expression pattern table 140 is a table that stores typical expressions representing time in the form of a word chain pattern, and each time expression pattern, whether the expression pattern represents the past or the future, and “last year”. It is composed of classification information such as whether it represents a relative time based on the present such as "tomorrow" or an absolute time such as the Christian era.

For example, the classification information corresponding to the combination of the time expression pattern "" Yesterday "+" Year | Month | Day | Season | Morning | Daytime | Evening | Evening "" is "Relative Past", "Right Now" + "Year | Month | Day | Season | Morning | Day | Evening | Evening" "is the relative information, and" Number + "Year | Month | Day | Hour" "is the classification information. Relative future ”.

Next, the operation of the predicate sentence tense analysis unit 130 for analyzing the tense in which the infix word is omitted when the above descriptive word determination unit 115 determines that the injunction sentence is a verb predicate sentence will be described. First, the existing noun table 150 used by the verb predicate sentence tense analysis unit 131 will be described.

FIG. 7 is a conceptual diagram of the noun table in advance according to the first embodiment of the present invention. The noun table 150 already exists
In the case of being used as a word stop such as "point out" or "replacement", it is a table that lists nouns that always represent the past (past). In the verb predicate sentence tense analysis unit 131,
By comparing it with the noun of the end sentence of the word-stop sentence, it is possible to analyze the tense in which word-stop is omitted.

FIG. 8 is a flow chart showing the operation of the verb predicate sentence tense analysis unit of the first embodiment of the present invention. (Step 311) When the predicate determination unit 115 determines that the verbal predicate sentence is a verb predicate sentence, the verb predicate sentence tense analysis unit 131 includes the temporal expressions in the verbal predicate sentence extracted by the time expression extraction unit 120. To modify the end clause,
It is determined whether or not there is something that represents the past, and if it exists, the process proceeds to step 316.

(Step 312) In step 311, the verb predicate sentence tense analysis unit 131 modifies the sentence end phrase in the time expression in the sentence stop sentence, and when there is no thing indicating the past, the verb predicate sentence tense analysis unit 131 It is determined whether the same noun as the noun is already registered in the noun table 150, and if it is registered, the process proceeds to step 316. (Step 313) By step 312, the same noun as the noun at the end of the sentence stop sentence is already in the noun table 1.
If it is not registered in 50, the verb predicate sentence tense analysis unit 131 determines whether the wording stop sentence is a compound sentence or a simple sentence, and if it is a simple sentence, the process proceeds to step 317.

(Step 314) Above Step 313
In the case where the verb predicate sentence tense analysis unit 131 determines that the incomprehension sentence is a compound sentence, the predicate of the subordinate clause of the incomprehension sentence is the present tense and the time extracted by the time expression extraction unit 120. It is judged whether or not there is an expression that modifies the predicate of the subordinate clause and that represents the absolute past, and if there is, the process proceeds to step 316.

(Step 315) The above step 314
As a result, the verb predicate sentence tense analysis unit 131 determines that the predicate of the subordinate clause of the noun sentence is the past tense, and the time expression extraction unit 12
If the predicate of the subordinate clause is modified in the time expressions extracted in 0 and there is no representation of the absolute past, then the predicate of the subordinate clause of the said word stop sentence is the past tense, and the time The expression extracting unit 120 determines whether or not the time expressions extracted are those that modify the predicate of the subordinate clause and represent the relative future. If they do not exist, step 317.
If it exists, the process proceeds to step 316.

(Step 316) The above step 311
In the case where a sentence end phrase is modified in the extracted time expressions to represent the past, or when the same noun as the sentence end noun is already registered in the noun table 150 in step 312, Alternatively, in step 314, if the predicate of the subordinate clause of the improper sentence is the present tense, and there is an expression that modifies the predicate of the subordinate clause and represents the absolute past in the extracted time expression, or In step 315, when the predicate of the subordinate clause of the impersonation sentence is the past tense, and the extracted temporal expressions modify the predicate of the subordinate clause and represent the relative future,
The verb predicate sentence tense analysis unit 131 determines that the omitted tense of the wording sentence is past.

(Step 317) The above step 313
In step 315, the predicate of the subordinate clause of the substantive sentence is a past tense, and the predicate of the subordinate clause is modified in the extracted time expression, and When there is no one that represents the relative future, the verb predicate sentence tense analysis unit 131 determines that the omitted tense of the instant word stop sentence is the present.

FIG. 9 is a flow chart showing the operation of the noun predicate sentence tense analysis unit of the first embodiment of the present invention. (Step 321) The noun predicate sentence tense analysis unit 132
When the language determination unit 115 determines that the wording stop sentence is a noun predicate sentence, whether the time expressions in the wording stop sentence extracted by the time expression extraction unit 120 include a past expression. If it is determined that there is nothing that represents the past, the process proceeds to step 324.

(Step 322) The noun predicate sentence tense analysis unit 132, if there is a word representing the past in the time expressions in the word stop sentence, the word stop sentence describes the time. It is determined whether there is a statement, and if the statement is a statement of time, the process proceeds to step 324.
Here, the statement that states time is X is a time expression, and when A, B, and * are arbitrary expressions, "A is / ga * X.""* X is / ga B." It is a sentence of a pattern.

(Step 323) The above step 322
In, if the sentence-fixing sentence is not a statement that states time, then whether the sentence-fixing sentence is a compound sentence with the complement sentence indicator “of” or “thing” and whether the predicate of the subordinate clause is the past tense judge. Here, if the sentence stop sentence is a compound sentence having the complement sentence indicator "no" or "koto" and the predicate of the subordinate clause is the past tense, the process proceeds to step 324.

(Step 324) The noun predicate sentence tense analysis unit 132 determines in the above step 321 that the time expression in the sentence stop sentence does not indicate the past, and in the step 322, the sentence stop sentence. Is a statement stating time, step 323 above
In, in the case where the injunction sentence is a compound sentence having the complement sentence indicator “of” or “koto” and the predicate of the subordinate clause is the past tense, the omitted tense of the injunction sentence is “present”. judge.

(Step 325) The above step 323.
In the above, the injunction sentence is the supplementary sign "no", "koto"
If the sentence is not a compound sentence or the predicate of the subordinate clause is not the past tense, the abbreviated tense of the sentence stop sentence is determined to be “past”. The following is a specific example of tense analysis for four examples of improper sentences. Sentence 1 "The end of World War II was 1945." Sentence 2 "It was his decision to retire Lake of the North." Sentence 3 "The prime minister emphasizes that it will not lead to inflation." Sentence 4 "The military is 1890. Construction of a military factory on the territory returned by year. ”[First example] (Sentence 1) When“ Sentence 1 ”is used as the Japanese sentence to be input, the wording stop extraction unit 110 displays the sentence shown in FIG. The result information of the syntax analysis shown in () is input. For example, (("The end of World War II" is "[noun-no-noun])" 1945
Year. "[Noun-wa-noun]).

First, the verbal stop extraction unit 110 checks whether the final word "1945" excluding the punctuation in the sentence ending clause is a noun (step 110). Since it is a noun, it is a verbal stop sentence and the final word " A sentence including “1945” is extracted as a wording stop sentence. Next, the predicate determination unit 115
It is checked whether or not the noun is a satypical noun (step 120). However, since it is not a satypical noun, it is judged as a noun predicate sentence (step 160).

Next, the time expression extraction unit 120 causes the "sentence 1".
The time expression and the classification information therein are compared with the syntactic analysis result information and the time expression pattern table 140, and FIG.
The time expression "1945" of the pattern "numerical +" year "" shown in (A) is discovered, and "absolute past" is acquired as the classification information. Since "sentence 1" is determined to be a noun predicate sentence by the predicate determination unit 115, the tense is analyzed by the noun predicate sentence tense analysis unit 132.

The noun predicate sentence tense analysis unit 132 determines that "sentence 1"
On the other hand, the time expression extraction unit 120 obtains a determination result that "1945" is a time expression representing the past, and then checks whether or not the statement is a statement of time (step 322). Since it is determined that the sentence is a sentence that has been written, as shown in FIG. 12A, the tense in which the wording stop is omitted is determined to be "present" (step 324).

[Second Example] (Sentence 2) When "Sentence 2" is used as the Japanese sentence to be input, the syntactic analysis result information shown in FIG. It For example, ((("North Lake is"
"Retired" [ga case]) "is entered in the form of" [supplementary sentence]) ("the person's""decision." [Noun-noun-noun]) [noun-is-noun]).

First, the wording stop extracting unit 110 determines that the final word "decision" excluding the punctuation in the sentence end clause is a noun. Next, the predicate determination unit 115 determines that it is a sa variant noun in step 120. Next, it is checked whether or not there is an adverb or a verb in the phrase that modifies the end phrase (step 130),
Since it does not exist in "Sentence 2", it is checked whether or not the end-of-sentence word is a sub-particle, a case particle, or a connecting particle among the clauses related to the end-of-sentence clause (step 140). However, since they do not exist, it is determined as a noun predicate sentence (step 160).

Next, the time expression extraction unit 120 causes the "sentence 2"
The syntactic analysis result information is collated with the time expression pattern table 140 to obtain the time expression and the classification information therein, but nothing is extracted because the "sentence 2" has no time expression. Since the “sentence 2” is determined to be a noun predicate sentence by the impersonation stopping extraction unit 110, the processing is transferred to the noun predicate sentence tense analysis unit 132 of the predicate sentence tense analysis unit 130.

The noun predicate sentence tense analysis unit 132 determines that "sentence 2".
Since there is no time expression in (step 321), since "has retired" is a subordinate clause of the past tense accompanied by "no" (step 323), the word tense is judged to be "present" (step 324). . [Third example] (Sentence 3) When "Sentence 3" is used as the Japanese sentence to be input, the wording stop extraction unit 110 receives the result information of the syntax analysis shown in FIG.

For example, input in the form of "(prime minister)"(("" not linked to inflation "" [case relation (d)]] "emphasized." [Quote] [case relation (ga)]) To do.

First, the wording stop extracting unit 110 determines that the final word "emphasis" excluding the punctuation in the sentence ending clause is a noun in step 110. The predicate determination unit 115 determines that the wording stop sentence is a sa variant noun. Next, the predicate determination unit 115
Although it is determined in the above step 130 that there is no verb or adverb, it is determined in the above step 140 that there is a clause "when it is not connected" including the connective particle, and it is determined to be a verb predicate sentence (step 170).

Next, the time expression extraction unit 120 causes the "sentence 3".
The syntactic analysis result information is collated with the time expression pattern table 140 in order to obtain the time expression and the classification information therein, but nothing is extracted because "sentence 3" has no time expression. Since “statement 3” is determined to be a verb predicate sentence by the verbal stop extraction unit 110, the processing is transferred to the verb predicate sentence tense analysis unit 131 of the predicate sentence tense analysis unit 130.

The verb predicate sentence tense analysis unit 131 modifies the sentence end clause in the time expressions extracted by the time expression extraction unit 120 and checks whether there is any expression indicating the past (step 311). Since there is no time expression in ",
It is checked whether the same noun as the end-of-sentence noun is already registered in the noun table 150 (step 312). Since the sentence end noun "emphasized" of "sentence 3" is already registered in the noun table 150, the tense in which the wording is omitted is determined to be "present" as shown in FIG. 12C (step 31).
6).

[Fourth example] (Sentence 4) When "Sentence 4" is used as the Japanese sentence to be input, the wording stop extracting unit 110 receives the result information of the syntax analysis shown in FIG. It For example, ("Military is"
((("By 1890""Return" [Category (D)]) "On the territory" [Dictionary-Synopsis]) ("Military factory""Construction." [Category (Wo)]) [Case relationship (d)])).

First, the wording stop extracting unit 110 determines that the final word "construction" excluding the punctuation in the sentence end clause is a noun. The predicate determination unit 115 determines that the wording stop sentence is a sa variant noun. Next, the predicate determination unit 115 determines that there is no time and adverb in the above step 130, but the above step 1
It is determined that there is a case clause "military factory" in 40, and it is determined to be a verb predicate sentence (step 170).

Next, the time expression extraction unit 120 causes the "sentence 4".
By comparing the time expression and its classification information in the syntax analysis result information with the time expression pattern table 140,
The time expression "1890" of the pattern "numerical +" year "" shown in FIG. 11B is discovered, and "absolute past" is acquired as the classification information. “Sentence 4” is the word stop extractor 110
Since it is determined to be a verb predicate sentence, the processing is transferred to the verb predicate sentence tense analysis unit 131 of the predicate sentence tense analysis unit 130.

The verb predicate sentence tense analysis unit 131 has the time expression "1890" representing the past in the step 311, but does not modify the sentence ending clause, and the "construction" also exists in the noun table 150 in the step 312. It is determined that it is not registered in. Furthermore, "statement 4" is determined to be a compound sentence, the predicate of the subordinate clause is the present tense, and it is checked whether or not the extracted time expression modifies the predicate of the subordinate clause and represents the absolute past (step 314). In this example,
The subordinate clause "Return by 1890" is in the "present" form,
Since the time expression "1890" is an absolute past and modifies the predicate of the subordinate clause, the tense in which the wording is omitted is determined to be "past" as shown in FIG. 12 (D) (step 316). .

Further, in this embodiment, the verb predicate sentence tense analysis unit 131 and the noun predicate sentence tense analysis unit 132 are treated as the predicate sentence tense analysis unit 130, but they may be treated separately. Verb predicate sentences and noun predicate sentences often show different linguistic phenomena, and in language analysis and research, they are often treated in different frameworks. [Second Embodiment] In the first embodiment, the tense is analyzed using the in-sentence information, but in the second embodiment described below, an example of analyzing the tense based on the context is shown.

FIG. 13 shows the system configuration of the second embodiment of the present invention. The Japanese sentence tense analysis device shown in the figure is a tense analysis unit 10 using in-sentence information explained in the first embodiment.
0, context information stack 300, tense analysis unit 400, tense analysis rule table 500, and analysis control unit 200. The context information stack 300 is a table that holds the input sentence-by-sentence syntax analysis result information.

The tense analysis rule table 500 is a table in which a rule for analyzing the tense of a word stop sentence is described with the syntax pattern of the word stop sentence and the syntax pattern of the subsequent sentence as constraints. FIG. 14 is a conceptual diagram of the tense analysis rule table according to the second embodiment of this invention. As shown in the figure, the tense analysis rule table 500 has a rule ID 51
0, an incomprehension sentence constraint 520, a postscript constraint 530, and a tense 540.

The in-sentence information utilization type tense analysis unit 100 determines whether or not the sentence is a sentence stop sentence on a sentence-by-sentence basis. In the case of a sentence stop sentence, the time information in the sentence or the noun used for the sentence stop is determined. It uses literary and semantic characteristics to analyze the tense. The analysis control unit 200 is assumed to be input with a syntax-analyzed Japanese sentence. The syntactic analysis uses the result of morphological analysis, which divides a Japanese sentence written in Kanji and Kana into word units, as an input and analyzes the relation between the modifiers and the modifiers of the clauses that compose the Japanese sentence. This is the process of recognizing whether such a modification relationship exists. Generally, a Japanese document processing system has processing functions of morphological analysis and syntactic analysis.

The tense analysis unit 400 uses the context information stack 3
Whether the tense information before and after the sentence stop sentence acquired from 00 and the parallelism between the sentence stop sentence acquired from the tense analysis rule table 500 and the next sentence are analyzed, and whether the sentence stop sentence represents the “present”, Analyze whether it represents the "past". Next, the operation of the outline of this embodiment will be described. FIG. 15 is a flow chart showing the outline operation of the second embodiment of the present invention.

Document-based contexts are stacked on the context information stack 300 (step 501). Analysis control unit 20
In the case of 0, the result of the syntax analysis is input, and the information stacked in the context information stack 300 is transferred one sentence at a time to the sentence information use type tense analysis unit 100 (step 50).
2). The in-sentence information utilization type tense analysis unit 100 analyzes the grammatical and semantic meanings of the nouns used in the body stop sentences. However, if there is no key expression in the sentence to analyze the tense, the tense is indefinite. Therefore, the in-sentence information utilization type tense analysis unit 100, if the result of the analysis indicates that the wording is stopped and the tense is indefinite (step 50).
(3, Yes), the analysis result is transferred to the tense analysis unit 400 (step 504). The tense analysis unit 400 updates the contents of the context information stack 300 (step 505).

Next, the operation of the tense analysis unit 400 will be described. In the following description, the sentence information use type tense analysis unit 100 will be described.
The description will be made assuming that the analysis result is transferred. FIG. 16 is a flow chart showing the operation of the tense analysis unit of the second exemplary embodiment of the present invention. The tense analysis unit 400 determines whether the input sentence is the final sentence (step 511),
If it is the final sentence, the process proceeds to step 517. If it is not the final sentence, the parsing result and the analysis results of the preceding and following sentences are acquired from the context information stack 300 (step 51).
2). The tense analysis unit 400 determines whether the input sentence is the first sentence, and if it is the first sentence, skips the reading and performs the process of step 515. If it is not the first sentence (step 513, No), the tense of the preceding sentence and the tense of the latter sentence are compared. If both tenses are the same, the process proceeds to step 516 (Yes in step 514), and if they are different. (Step 5
No. 14, No.), the tense analysis rule table 500 is searched by using the analysis results of the injunction sentence and the after sentence as keys, and the tense analysis result is acquired (step 515). If the tense of the preceding sentence and the tense of the latter sentence are the same, the tense of the sentence stop sentence is the tense of the preceding sentence and the latter sentence (step 516).

As a result, the analysis control unit 200 updates the context information stack 300 based on the acquired information "comprehension tense = past". [First Example] Next, an example in which the above operation is applied to the following example will be described. The Japanese sentence to be processed is "The government has announced an emergency import of rice. Rice crops are record-breaking crops nationwide. The Food Agency estimates that at least 2 million tons of imports will be needed to meet domestic demand within the year. First, the result of syntax analysis of the above sentence is input to the analysis control unit 200, and the context information stack 300 is displayed in FIG.
The information shown in 7 is stacked. In the example shown in the figure, in the first sentence, “The government has announced an emergency import of rice”, the subject is “government” and the tense is “past”.

The second sentence, "American crops are record-breaking crops nationwide. ”, The subject is stacked as“ rice production ”and the tense as“ indefinite ”. Regarding the third sentence, "The Food Agency predicts that at least 2 million tons of imports will be required to cover domestic receipts within the year," the subject is "at the Food Agency," and the tense is "present."
Is stacked as.

The analysis control unit 200 transfers the information stacked in the context information stack 300 to the in-sentence information use type tense analysis unit 100 one by one. The in-sentence information use type tense analysis unit 100 analyzes each transferred information and outputs an analysis result shown in FIG. In the example of the figure, the in-sentence information use type tense analysis unit 100 sets the subject of the first sentence as “government” and the tense as “past”.
Regarding the sentence, the subject is "rice production", the tense is indefinite, the sentence is a noun sentence, and it is a noun predicate sentence. For the third sentence, the subject is "at the Food Agency" and the tense is "present". I am trying.

In regard to the second sentence, it is analyzed that it is a noun sentence sentence and a noun predicate sentence, but there is no key expression for analyzing tense in the sentence of the second sentence. , The tense is not decided. Then, the analysis control unit 200 transfers the information of the wording stop sentence to the tense analysis unit 400. Since the transferred sentence is the second sentence, the tense analysis unit 400 acquires the first sentence, which is the preceding sentence, and the third sentence, which is the latter sentence, from the context information stack 300, and compares the tenses. In this case, the tense of the first sentence is "past", and the tense of the third sentence is "present". Therefore, the information of the sentence stop sentence (second sentence) and the latter sentence (third sentence) is used as a key. , The tense analysis table 500 is searched.

In this case, since the verbal stop sentence is a noun predicate sentence and the syntax is a verb predicate sentence, a result that “the tense of the verbal stop sentence = the tense of the preceding sentence” is obtained by matching the rule M2. Since the tense of the preceding sentence is "past", the tense of the wording sentence is "past". [Second Example] The following second example will be described. “The World Bank resumed lending to Israel for the first time in 17 years. Israel uses this for public works such as road facilities and dam construction. Aiming to break free from the civil war that lasted for about 20 years. First, the result of syntax analysis of the above sentence is input to the analysis control unit 200, and the context information stack 300 is displayed in FIG.
The information shown in 9 is stacked. In the example shown in the figure, the first sentence "The World Bank resumed its ownership in Israel for the first time in 17 years. ”, The subject is“ World Bank ”and the tense is“ past ”
Is stacked as.

The second sentence, "Israel uses this as road equipment,
Used for public works such as dam construction. , The subject is stacked as “Israel is” and the tense is “indefinite”. The third sentence, "Aiming to escape from the civil war that lasted for about 20 years. ”, The tense is stacked as“ present ”. The analysis control unit 200 transfers the information stacked in the context information stack 300 one sentence at a time to the in-sentence information utilization type tense analysis unit 100.

The sentence information use type tense analysis unit 100
Each transferred information is analyzed and the analysis result shown in FIG. 20 is output. In the example of the figure, the in-sentence information utilization type tense analysis unit 10
0 is the subject of the first sentence is “World Bank”, the tense is “past”, and the second sentence is the subject of “Israel” and the tense is “indefinite”.
Assume that it is a verb predicate sentence, and the tense is “present” for the third sentence.
And

In regard to the second sentence, it is analyzed that it is a verbal predicate sentence and a verb predicate sentence.
Since there is no key expression for analyzing the tense in the second sentence, the tense is not determined, and the wording stop information is transferred to the tense analysis unit 400. The tense analysis unit 400 obtains the information of the preceding sentence (first sentence) and the succeeding sentence (third sentence) from the context information stack 300, because the improper sentence stopping sentence is neither the first sentence nor the final sentence, and compares the tenses. To do. Preamble (first sentence) is "past" and post sentence (third sentence)
Is “present”, the tense analysis rule table 500 is searched using the information of the wording stop sentence and the after sentence as a key.

In this case, the injunction sentence is a verb predicate sentence and has a subject expression of “Israel”, and
(Sentence) is also a verb predicate sentence, it matches rule D4 and obtains the result that "tense tense of postponement = tense of postscript", and the tense of postscript is "present" The sentence tense = present ". The present invention is not limited to the above-mentioned first and second embodiments, and various modifications and applications are possible within the scope of the claims.

[0083]

As described above, according to the present invention, since the tense can be analyzed by utilizing the interaction between the tense in the subordinate clause and the tense in the main clause (end of sentence), the time is expressed in the sentence. Even when there is no expression, it is possible to analyze the tense in which the wording sentence is omitted. In addition, since the tense analysis is performed after distinguishing between the expressions that represent absolute time and the expressions that represent relative time in the subordinate clause, it is possible to analyze in consideration of the difference in the tense standard between the subordinate clause and the main clause. , It becomes possible to analyze the tense in which the wording sentence was omitted, which could not be analyzed in the past.

Furthermore, since the analysis is performed using not only the sentence stop sentence but also the information of the sentences before and after, the tense can be analyzed even when there is no expression suggesting the tense in the sentence stop sentence. In addition, since the information of the sentences before and after is taken into consideration not only the tense but also the parallelism of the action with the word stop sentence and the temporal transitivity, when the tense of the sentence before and after the word stop sentence is different. Also, it becomes possible to grasp the context and analyze it appropriately.

[Brief description of drawings]

FIG. 1 is a principle configuration diagram of the present invention.

FIG. 2 is a diagram for explaining the principle of the present invention.

FIG. 3 is a system configuration diagram of a first embodiment of the present invention.

FIG. 4 is a configuration diagram of a Japanese sentence tense analysis device according to the first embodiment of the present invention.

FIG. 5 is a flowchart showing the operation of the wording stop extraction unit and the predicate determination unit according to the first embodiment of this invention.

FIG. 6 is a conceptual diagram of a time expression pattern table according to the first embodiment of this invention.

FIG. 7 is a conceptual diagram of an existing noun table according to the first embodiment of this invention.

FIG. 8 is a flowchart showing an operation of a verb predicate sentence tense analysis unit of the first exemplary embodiment of the present invention.

FIG. 9 is a flowchart showing an operation of a noun predicate sentence tense analysis unit according to the first embodiment of this invention.

FIG. 10 is a structural diagram of an input Japanese sentence according to the first embodiment of this invention.

FIG. 11 is a conceptual diagram of a time expression extraction result according to the first embodiment of this invention.

FIG. 12 is a diagram showing processing of a verb predicate sentence tense analysis unit and a noun predicate sentence tense analysis unit according to the first embodiment of this invention.

FIG. 13 is a system configuration diagram of a second embodiment of the present invention.

FIG. 14 is a diagram showing an example of a tense analysis rule table according to the second embodiment of the present invention.

FIG. 15 is a flowchart showing a schematic operation of the second embodiment of the present invention.

FIG. 16 is a flowchart of the operation of the tense analysis unit of the second example of the present invention.

FIG. 17 is an analysis result (No. 1) by the analysis control unit according to the second embodiment of the present invention.

FIG. 18 is an analysis result (No. 1) of the sentence information utilizing type tense analysis unit according to the second embodiment of the present invention.

FIG. 19 is an analysis result (No. 2) by the analysis control unit according to the second embodiment of the present invention.

FIG. 20 is an analysis result (No. 2) of the sentence information utilizing type tense analysis unit according to the second embodiment of the present invention.

[Explanation of symbols]

100 first analysis means, in-sentence information utilization type tense analysis section 110 extraction means, word stop extraction section 115 predicate determination means, predicate determination section 120 temporal expression extraction section 130 tense analysis means using syntactic patterns, predicate sentence tense analysis Part 131 Verb predicate sentence tense analysis means, Verb predicate sentence tense analysis part 132 Noun predicate sentence tense analysis means, Noun predicate sentence tense analysis part 140 Temporal expression pattern table 150 Preexisting noun table 200 Analysis control part 300 Context information storage means, context information Stack 400 Temporal analysis unit 500 Temporal analysis rule, tense analysis rule table 510 Rule ID 520 Indicative sentence constraint condition 530 Postscript constraint condition 540 Temporal 800 Second analysis means

Claims (13)

[Claims] 1. A Japanese sentence tense analysis device for analyzing the tense of a Japanese sentence in a Japanese sentence processing system for inputting and processing a Japanese sentence, wherein the sentence ending clause of the Japanese sentence ends with a noun. The present invention is characterized by having a first analyzing means for analyzing a tense in which a stop sentence is omitted based on the property of the sentence ending clause, and a second analyzing means for analyzing by context information in addition to the first analyzing means. Japanese sentence tense analyzer. 2. The first analyzing means extracts the sentence stop sentence from the Japanese sentence, and the sentence stop sentence that is the last sentence clause of the sentence stop sentence is a verb predicate sentence or a noun predicate sentence. A predicate determining means for determining whether or not, and for each of the verb predicate sentence or the noun predicate sentence determined by the predicate determining means, analyzes the syntactic structure of the Japanese sentence,
A Japanese sentence tense analysis device, comprising: a tense analysis means utilizing a syntactic pattern for outputting a tense analysis result of the intelligence sentence. 3. A tense analysis unit using the syntactic pattern extracts a time expression pattern storage unit storing an expression pattern representing time, and a time expression appearing in the word stop sentence extracted by the extraction unit. 3. The time expression extraction means for collating with the time expression pattern storage means and outputting classification information as to whether the time expression expresses an absolute or relative tense. Japanese sentence tense analyzer. 4. The predescriptive word determination means is an adverb in all clauses related to the noun of the wording stop clause,
A judgment means for deciding whether the phrase stop phrase is used as a verb predicate sentence or a noun predicate sentence depending on the presence or absence of a phrase including a verb, a connecting particle or a verb case element. The Japanese sentence tense analyzer described in item 2. 5. The tense analysis means using the syntactic pattern, wherein the noun storage means stores a noun representing the past, and the predescription word determination means determines that the noun phrase is a verb predicate sentence. In, by analyzing the syntactic structure of the noun phrase, the type of nom phrase, the time expression and the classification information based on the matching result of the noun of the noun phrase and the existing noun storage means, Verb predicate sentence tense analysis means for analyzing the tense in which the word stop clause is omitted, and the syntactic structure of the word stop sentence when the word stop clause is determined to be a noun predicate sentence by the predescription word determination means The noun predicate sentence tense analyzing means for analyzing the tense in which the wording clause is omitted by analyzing the time expression and the classification information. Language sentence tense analyzer. 6. The second analysis means is a context information storage means for holding information in sentence units, and a rule for analyzing the tense of the word stop from the syntactic pattern of the word stop sentence and the subsequent sentence. And the tense analysis rule described above, the analysis result of the sentence stop sentence acquired by the first analysis unit is collated with the context information storage unit and the tense analysis rule, and the predicate of the sentence stop sentence and the next sentence The Japanese sentence tense analyzing apparatus according to claim 1, further comprising: tense analyzing means for analyzing the tense of the sentence stop sentence based on parallelism and transitivity of movement. 7. A Japanese sentence tense analysis method for analyzing the tense of a Japanese sentence in a document processing system for inputting and processing a Japanese sentence, wherein the sentence ending sentence in the Japanese sentence ends with a noun. A Japanese sentence tense analysis method characterized by analyzing a tense in which a stop sentence is omitted according to the property of the last sentence clause. 8. A nomination sentence in which the last sentence clause ends with a noun is extracted from the Japanese sentence, and whether the nomination sentence clause, which is the last sentence clause of the last sentence sentence, is a verb predicate sentence or a noun predicate sentence. 8. The Japanese sentence tense analysis method according to claim 7, further comprising: analyzing the syntactic structure of the verbal stop sentence for each of the verb predicate sentence and the noun predicate sentence, and outputting the result of tense analysis of the verbal predicate sentence. . 9. A classification of whether the time expression expresses an absolute or relative tense by extracting a time expression appearing in the word stop sentence after the extraction of the word stop and comparing it with a predetermined pattern. The Japanese sentence tense analysis method according to claim 8, which outputs information. 10. A phrase which is a case element of a verb or a verb including adverbs, verbs, connecting particles, among all the phrases related to the noun of the grammatical stop phrase when determining the type of the grammatical stop phrase. 9. The Japanese sentence tense analysis method according to claim 8, wherein it is determined whether the incomplete phrase is used as a verb predicate sentence or a noun predicate sentence depending on the presence or absence of. 11. A pre-existing noun memory that stores a noun of the noun phrase and a noun representing the past when it is determined that the noun phrase is a verb predicate sentence when the tense of the predescription word sentence is analyzed. Based on the syntactical structure of the body language stop sentence, the time expression in the body language stop sentence and the classification information, the body language stop clause is omitted by collating the means and analyzing whether the same noun is registered. The tense is analyzed, and when it is determined that the wording sentence is a noun predicate sentence, the wording clause is omitted based on the syntactic structure of the word sentence stopping sentence, the time expression, and the classification information. The Japanese sentence tense analysis method according to claim 8, wherein the tense is analyzed. 12. The Japanese sentence tense analysis method according to any one of claims 7 to 11, wherein a tense that cannot be analyzed because the Japanese sentence does not include an expression indicating a tense in the Japanese sentence is used as a tense before and after the sentence stopping sentence. The Japanese sentence tense analysis characterized by analyzing the tense of the above word stop sentence by analyzing the parallelism of the predicates and the transitivity of the actions of the above word stop sentence and the after sentence of the above word stop sentence Method. 13. The syntactic analysis result of the Japanese sentence is input, the syntactic analysis result is held as context information in the order of input, and when the Japanese sentence includes a wording sentence sentence, sentences before and after the word sentence sentence sentence are included. The tense of is analyzed by referring to the context information, and when it is not parseable, the syntactic information of the sentence stop sentence and the sentence after the sentence stop and the rule for analyzing the tense are matched and matched. 13. The Japanese sentence tense analysis method according to claim 12, wherein the tense is the tense of the sentence stop sentence.