JP2004287683A - Evaluation expression extraction device, program, storage medium, and evaluation expression extraction method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an evaluation expression extraction device capable of extracting a pair of an evaluation object and an evaluation expression corresponding to the evaluation object from a document. <P>SOLUTION: A text data structure generation means 102 converts a text to a data structure retaining phrase information including at least the notation and modification information of constituting words, based on a language analysis result in a language analysis means 101, which performs a language analysis processing including at least morphological analysis processing and phrase modification analysis processing to the text. A specified phrase pair selection means 104 selects a phrase pair which refers to the evaluation of an article from phrase pairs determined to have a modification relation, and an evaluation object/evaluation expression specifying means 106 specifies the evaluation object and the evaluation expression corresponding to this evaluation object from the selected segment pair. According to this, the pair of the evaluation object and the evaluation expression corresponding to this evaluation object can be extracted from the document. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an evaluation expression extraction device, a program, a storage medium, and an evaluation expression extraction method.
[0002]
[Prior art]
If useful information expressing the contents of the document can be extracted from the document, applications such as document search, document classification, and document analysis based on the information can be realized. In particular, when analyzing the results of a questionnaire, it is considered that the evaluation information written in the free description part sent by the respondent has a point of analysis, and extraction of the writer's evaluation expression is an important issue. Therefore, in recent years, various document information extraction techniques have been proposed in order to extract information for finding some knowledge from a large amount of document data.
[0003]
Patent Literature 1 proposes a technique for automatically extracting useful information in a document by measuring the frequency of words in the document, converting the frequency into “weight” of the words, and automatically identifying the keyword. Have been.
[0004]
Patent Document 2 proposes a technique for automatically extracting useful information in a document by extracting a keyword expression using a relationship between words. This is a method of normalizing the arrangement of a plurality of adjacent words and improving information retrieval accuracy.
[0005]
Further, in Patent Document 3, the dependency analysis of a target document is performed, a resulting syntax tree or linear list is constructed, and a frequent pattern of a word and a positional relationship is used by using a constraint condition or a parameter. Techniques for automatically extracting useful information have been proposed.
[0006]
[Patent Document 1]
JP 08-30627 A
[Patent Document 2]
JP-A-08-129554
[Patent Document 3]
JP 2001-84250 A
[0007]
[Problems to be solved by the invention]
However, according to the technology proposed in Patent Literature 1, for example, when “price is high” and a document search is performed using a keyword “price” & “high” as a search key, “price is high with a camera having a high resolution” However, there is a problem in that a document that is completely different from what is expected, such as "camera" instead of "price", is evaluated as "expensive".
[0008]
According to the technique proposed in Patent Document 2, a keyword expression that does not depend on a surface expression can be obtained, but dependency analysis is not performed, and the concept of "high price" is extracted because adjacent words are targeted. If you want to do that, you can't extract from the text "more expensive cameras".
[0009]
Further, according to the technique proposed in Patent Document 3, the concept of “price → high” can be acquired from the text “price is higher”, but the evaluation expression is actually a long syntax. It is rarely represented by a tree, and construction of a high-cost syntax tree or a linear list is not practical.
[0010]
It is an object of the present invention to provide an evaluation expression extraction device, a program, a storage medium, and an evaluation expression extraction method capable of extracting a pair of an evaluation target and an evaluation expression corresponding to the evaluation target from a document.
[0011]
[Means for Solving the Problems]
According to a first aspect of the present invention, there is provided an evaluation expression extracting apparatus for performing a language analysis process on a text including at least a morphological analysis process and a phrase dependency analysis process, based on a language analysis result of the language analysis device. A text data structure generating means for converting a text into a data structure holding phrase information including at least constituent word notation and dependency information; and a phrase pair determined to be in a dependency relationship by the text data structure generating means, A specific phrase pair selecting means for selecting a phrase pair that matches a predetermined selection criterion, and an evaluation target for specifying an evaluation target and an evaluation expression corresponding to the evaluation target from the phrase pairs selected by the specific phrase pair selecting means・ Evaluation expression specifying means.
[0012]
Therefore, it is possible to extract a pair of an evaluation target and an evaluation expression corresponding to the evaluation target from a document, and to apply the analysis to the evaluation of an object in a questionnaire or the like, or to search a document by the evaluation expression. .
[0013]
According to a second aspect of the present invention, in the evaluation expression extracting apparatus according to the first aspect, the selection criterion in the specific phrase pair selecting means is such that a dependency relationship of the target phrase pair is a predicative relationship, Are adjectives, adjective verbs, and auxiliary adjectives.
[0014]
Therefore, in the evaluation expression extraction process, the inter-clause dependency relationship is a predicative relationship, and by targeting only phrase pairs that contain the terms used in the evaluation, the extraction process can be more efficient than targeting all phrase pairs. There is an effect that it can be reduced.
[0015]
According to a third aspect of the present invention, in the evaluation expression extracting apparatus according to the first aspect, the selection criterion in the specific phrase pair selecting means is that the dependency relation of the target phrase pair is a continuous modification relation, Are adjectives, adjective verbs, and auxiliary adjectives.
[0016]
Therefore, in the evaluation expression extraction process, the inter-clause dependency relationship is an adnominal modification relationship, and by targeting only phrase pairs that contain the terms used in the evaluation, the extraction process can be more efficient than targeting all phrase pairs. There is an effect that it can be reduced.
[0017]
According to a fourth aspect of the present invention, in the evaluation expression extracting device according to the first aspect, the evaluation target / evaluation expression specifying means performs evaluation when specifying an evaluation target from a phrase pair and an evaluation expression corresponding to the evaluation target. Use target / evaluation expression specification rules.
[0018]
Therefore, when specifying the evaluation target and the evaluation expression, it is possible to specify the evaluation target and the evaluation expression with high accuracy by using the specific rule utilizing the grammatical knowledge of the natural language.
[0019]
According to a fifth aspect of the present invention, in the evaluation expression extracting device according to the fourth aspect, when the dependency relation of the target clause pair is a predicate relation, the evaluation target / evaluation expression specifying rule includes Identify the independence words that are included in the evaluation, and add the adjectives, adjective verbs, auxiliary adjectives, independence words preceding the adjectives included in the receiving clause, and the auxiliary verb to cancel if they are included. The rules for specifying the evaluation expression are described.
[0020]
Therefore, since the grammatically limited specific rule is included in the dependency clause pair having the predicative relationship, it is possible to more accurately extract the evaluation target and the evaluation expression.
[0021]
According to a sixth aspect of the present invention, in the evaluation expression extracting device according to the fourth aspect, the evaluation target / evaluation expression specifying rule includes a rule in which, when the dependency relationship of the target phrase pair is a predicative relationship, If the independence word contained is a formal noun, the independence word of the dependency clause that receives the phrase and continually modifies it as a phrase is specified as the evaluation target, and the adjective, adjective verb, auxiliary adjective included in the received phrase and When a self-sufficient word and a cancellation auxiliary verb are included, a rule for specifying an expression to which it is added as an evaluation expression is described.
[0022]
Therefore, if the subject to be evaluated is a formal noun that is not an entity for a dependency clause pair having a predicative relationship, the entity can be obtained from the adnominal modifier clause, and the evaluation object and the evaluation expression with higher accuracy can be obtained. Can be extracted.
[0023]
According to a seventh aspect of the present invention, in the evaluation expression extracting apparatus according to the fourth aspect, the evaluation target / evaluation expression specifying rule includes a rule in which, when the dependency relation of the target phrase pair is a continuous modification relation, If the included adjective, adjective verb, auxiliary adjective and the independent word preceding it, and if the auxiliary verb for cancellation is included, specify the expression to which it is added as the evaluation expression, and identify the independent word included in the receiving phrase The rules to be identified as evaluation targets are described.
[0024]
Therefore, since the grammatically limited specific rule is included in the dependency clause pair having the adnominal modification relation, it is possible to more accurately extract the evaluation target and the evaluation expression.
[0025]
The program according to claim 8 is a program installed or interpreted and executed on a computer, wherein the computer includes at least a morphological analysis process and a phrase dependency analysis process for text. A language analysis function for performing a language analysis process, and a text data structure generation function for converting a text into a data structure holding phrase information including at least constituent word notation and dependency information based on a result of the language analysis in the language analysis function. A specific phrase pair selection function for selecting a phrase pair that meets a predetermined selection criterion from phrase pairs determined to be dependent by the text data structure generation function, and a specific phrase pair selection function Evaluation target / evaluation expression that specifies an evaluation target and an evaluation expression corresponding to this evaluation target from a phrase pair A constant function is executed.
[0026]
Therefore, it is possible to extract a pair of an evaluation target and an evaluation expression corresponding to the evaluation target from a document, and to apply the analysis to the evaluation of an object in a questionnaire or the like, or to search a document by the evaluation expression. .
[0027]
According to a ninth aspect of the present invention, in the program according to the eighth aspect, the selection criterion in the specific phrase pair selection function is that the dependency relation of the target phrase pair is a predicative relation, and the part of speech of the constituent words of the receiving phrase is They are adjectives, adjective verbs, and auxiliary adjectives.
[0028]
Therefore, in the evaluation expression extraction process, the inter-clause dependency relationship is a predicative relationship, and by targeting only phrase pairs that contain the terms used in the evaluation, the extraction process can be more efficient than targeting all phrase pairs. There is an effect that it can be reduced.
[0029]
According to a tenth aspect of the present invention, in the program according to the eighth aspect, the selection criterion in the specific phrase pair selection function is that the dependency relation of the target phrase pair is a continuous modification relation, and the part of speech of the constituent words of the dependency phrase is They are adjectives, adjective verbs, and auxiliary adjectives.
[0030]
Therefore, in the evaluation expression extraction process, the inter-clause dependency relationship is an adnominal modification relationship, and by targeting only phrase pairs that contain the terms used in the evaluation, the extraction process can be more efficient than targeting all phrase pairs. There is an effect that it can be reduced.
[0031]
According to an eleventh aspect of the present invention, in the program according to the eighth aspect, the evaluation target / evaluation expression specifying function is used to specify an evaluation target / evaluation expression for specifying an evaluation target from a phrase pair and an evaluation expression corresponding to the evaluation target. Use expression specific rules.
[0032]
Therefore, when specifying the evaluation target and the evaluation expression, it is possible to specify the evaluation target and the evaluation expression with high accuracy by using the specific rule utilizing the grammatical knowledge of the natural language.
[0033]
According to a twelfth aspect of the present invention, in the program according to the eleventh aspect, when the dependency relation of the target clause pair is a predicative relation, the evaluation target / evaluation expression specifying rule includes a self-reliance clause included in the dependency clause. Identify the word as the evaluation target, and evaluate the adjective, adjective verb, auxiliary adjective included in the receiving phrase and the independent word preceding it, and if the auxiliary verb for cancellation is included, add the expression to the evaluation expression. The rules to be specified are described.
[0034]
Therefore, since the grammatically limited specific rule is included in the dependency clause pair having the predicative relationship, it is possible to more accurately extract the evaluation target and the evaluation expression.
[0035]
According to a thirteenth aspect of the present invention, in the program according to the eleventh aspect, when the dependency relation of the target clause pair is a predicate relation, the evaluation target / evaluation expression specifying rule is included in the dependency clause. If the word is a formal noun, the independence word of the dependency phrase that accepts the phrase as a noun and qualifies as a phrase is identified as the evaluation target, and the adjective, adjective verb, auxiliary adjective included in the received phrase and the independent word preceding it In addition, when an auxiliary verb for cancellation is included, a rule for specifying an expression to which the auxiliary verb is added as an evaluation expression is described.
[0036]
Therefore, if the subject to be evaluated is a formal noun that is not an entity for a dependency clause pair having a predicative relationship, the entity can be obtained from the adnominal modifier clause, and the evaluation object and the evaluation expression with higher accuracy can be obtained. Can be extracted.
[0037]
The invention according to claim 14 is the program according to claim 11, wherein the evaluation target / evaluation expression specifying rule includes an adjective included in a dependency phrase when a dependency relationship of the target phrase pair is a union modification relationship. , An adjective, an auxiliary adjective, an independent word preceding it, and an auxiliary verb to negate, if specified, an expression to which it is added is specified as an evaluation expression, and the independent word included in the receiving phrase is evaluated. The rules to be specified are described.
[0038]
Therefore, since the grammatically limited specific rule is included in the dependency clause pair having the adnominal modification relation, it is possible to more accurately extract the evaluation target and the evaluation expression.
[0039]
A computer-readable storage medium according to a fifteenth aspect stores the program according to any one of the eighth to fourteenth aspects.
[0040]
Therefore, by causing a computer to read the program stored in the storage medium, it is possible to obtain the same operation as the invention according to any one of claims 8 to 14.
[0041]
17. An evaluation expression extracting method according to claim 16, wherein the evaluation expression extracting method is performed by computer processing control, and is a language for performing a language analysis process including at least a morphological analysis process and a phrase dependency analysis process on text. An analysis step, a text data structure generation step of converting a text into a data structure holding phrase information including at least constituent word notation and dependency information based on a result of the language analysis in the language analysis step, and a text data structure generation A specific phrase pair selection step of selecting a phrase pair that meets a predetermined selection criterion from the phrase pairs determined to be dependent by the process, and an evaluation target from the phrase pairs selected in the specific phrase pair selection step. An evaluation target / evaluation expression specifying step of specifying an evaluation expression corresponding to the evaluation target.
[0042]
Therefore, it is possible to extract a pair of an evaluation target and an evaluation expression corresponding to the evaluation target from a document, and to apply the analysis to the evaluation of an object in a questionnaire or the like, or to search a document by the evaluation expression. .
[0043]
According to a seventeenth aspect of the present invention, in the evaluation expression extracting method according to the sixteenth aspect, the selection criterion in the specific phrase pair selection step is such that a dependency relationship of the target phrase pair is a predicative relationship, and a constituent word of the receiving phrase. Are adjectives, adjective verbs, and auxiliary adjectives.
[0044]
Therefore, in the evaluation expression extraction process, the inter-clause dependency relationship is a predicative relationship, and by targeting only phrase pairs that contain the terms used in the evaluation, the extraction process can be more efficient than targeting all phrase pairs. There is an effect that it can be reduced.
[0045]
According to an eighteenth aspect of the present invention, in the evaluation expression extracting method according to the sixteenth aspect, the selection criterion in the specific phrase pair selecting step is such that a dependency relationship of the target phrase pair is a continuous modifier relation, and a constituent word of the dependency phrase. Are adjectives, adjective verbs, and auxiliary adjectives.
[0046]
Therefore, in the evaluation expression extraction process, the inter-clause dependency relationship is an adnominal modification relationship, and by targeting only phrase pairs that contain the terms used in the evaluation, the extraction process can be more efficient than targeting all phrase pairs. There is an effect that it can be reduced.
[0047]
According to a nineteenth aspect of the present invention, in the evaluation expression extracting method according to the sixteenth aspect, the evaluation target / evaluation expression specifying step includes the steps of: Use target / evaluation expression specification rules.
[0048]
Therefore, when specifying the evaluation target and the evaluation expression, it is possible to specify the evaluation target and the evaluation expression with high accuracy by using the specific rule utilizing the grammatical knowledge of the natural language.
[0049]
According to a twentieth aspect of the present invention, in the evaluation expression extracting method according to the nineteenth aspect, the evaluation target / evaluation expression specification rule includes a rule that, when the dependency relationship of the target phrase pair is a predicate relationship, Identify the independence words that are included in the evaluation, and add the adjectives, adjective verbs, auxiliary adjectives, independence words preceding the adjectives included in the receiving clause, and the auxiliary verb to cancel if they are included. The rules for specifying the evaluation expression are described.
[0050]
Therefore, since the grammatically limited specific rule is included in the dependency clause pair having the predicative relationship, it is possible to more accurately extract the evaluation target and the evaluation expression.
[0051]
According to a twenty-first aspect of the present invention, in the evaluation expression extracting method according to the nineteenth aspect, the evaluation target / evaluation expression specification rule includes a rule that, when the dependency relationship of the target phrase pair is a predicative relationship, If the included independent word is a formal noun, the independent word of the dependency clause that accepts the phrase and continually modifies it as a phrase is specified as the evaluation target, and the adjective, adjective verb, auxiliary adjective included in the received phrase and When a self-sufficient word and a cancellation auxiliary verb are included, a rule for specifying an expression to which it is added as an evaluation expression is described.
[0052]
Therefore, if the subject to be evaluated is a formal noun that is not an entity for a dependency clause pair having a predicative relationship, the entity can be obtained from the adnominal modifier clause, and the evaluation object and the evaluation expression with higher accuracy can be obtained. Can be extracted.
[0053]
According to a twenty-second aspect of the present invention, in the evaluation expression extracting method according to the nineteenth aspect, the evaluation target / evaluation expression specifying rule includes a rule in which, when the dependency relationship of the target phrase pair is a continuous modification relationship, If the included adjective, adjective verb, auxiliary adjective and the independent word preceding it, and if the auxiliary verb for cancellation is included, specify the expression to which it is added as the evaluation expression, and identify the independent word included in the receiving phrase The rules to be identified as evaluation targets are described.
[0054]
Therefore, since the grammatically limited specific rule is included in the dependency clause pair having the adnominal modification relation, it is possible to more accurately extract the evaluation target and the evaluation expression.
[0055]
BEST MODE FOR CARRYING OUT THE INVENTION
An embodiment of the present invention will be described with reference to FIGS.
[0056]
FIG. 1 is a block diagram schematically showing a hardware configuration of an evaluation expression extraction device 1 to which the present invention is applied. As shown in FIG. 1, the evaluation expression extraction device 1 is, for example, a personal computer or a workstation, and includes a CPU (Central Processing Unit) 2 which is a main part of the computer and centrally controls each unit. The CPU 2 includes a read only memory (ROM) 3 which is a read-only memory storing a BIOS and the like, and a RAM (Random) which rewritably stores various data.
Access Memory) 4 is connected by a bus 5.
[0057]
Further, on the bus 5, an HDD (Hard Disk Drive) 6 as an external storage device, a CD-ROM drive 8 for reading a CD (Compact Disc) -ROM 7 as a mechanism for reading computer software as a distributed program, A communication control device 10 that controls communication between the evaluation expression extraction device 1 and the network 9, an input device 11 such as a keyboard and a mouse, and a display device 12 such as a CRT (Cathode Ray Tube) and an LCD (Liquid Crystal Display) are provided. They are connected via an I / O (not shown).
[0058]
Since the RAM 4 has a property of storing various data in a rewritable manner, the RAM 4 functions as a work area of the CPU 2 and plays a role of, for example, a document buffer described later.
[0059]
In addition, the HDD 6 stores a language analysis dictionary 13 and evaluation target / evaluation expression specification rules 14 in addition to program files for storing various programs.
[0060]
Although not shown, the linguistic analysis dictionary 13 uses a morpheme string as a registration unit (entry), and has “notation” and “speech [category]” of each constituent morpheme as attributes of the entry. .
[0061]
Next, the evaluation target / evaluation expression specifying rule 14 will be described. The evaluation target / evaluation expression specification rule 14 is for specifying an evaluation target and an evaluation expression corresponding to this evaluation target for a phrase pair. FIG. 2 is an explanatory diagram showing an example of the evaluation target / evaluation expression specifying rule 14. As shown in FIG. 2, the evaluation target / evaluation expression specification rule 14 is such that, for example, dependency relations between phrases (adjunct modification relations, predicate relations, etc.) and evaluation expressions expressed in text are “high” or “expensive”. ", The part-of-speech attribute is an evaluation term such as an adjective or an adjective verb, and the key of the evaluation target and the evaluation expression is an independent word included in the phrase, and there is no expression that needs to be added as an adjunct It indicates grammatical knowledge such as auxiliary verbs such as "" and auxiliary adjectives such as "difficult", and their interrelationships. The evaluation target / evaluation expression specifying rule 14 may be described in a table format, a dictionary format, or the like, in addition to the if-then rule shown in FIG.
[0062]
The CD-ROM 7 shown in FIG. 1 embodies a storage medium of the present invention, and stores an OS (Operating System) and various programs. The CPU 2 reads the program stored on the CD-ROM 7 with the CD-ROM drive 8 and installs the program on the HDD 6.
[0063]
As the storage medium, not only the CD-ROM 7 but also various types of media such as semiconductor memories such as various optical disks such as DVDs, various magnetic disks such as magneto-optical disks, and flexible disks can be used. Alternatively, a program may be downloaded from a network 9 such as the Internet via the communication control device 10 and installed on the HDD 6. In this case, the storage device storing the program in the server on the transmission side is also a storage medium of the present invention. Note that the program may operate on a predetermined OS (Operating System), and in that case, the OS may substitute a part of execution of various processes to be described later, May be included as a part of a group of program files constituting the application software or the OS.
[0064]
The CPU 2 that controls the operation of the entire system executes various processes based on a program loaded on the HDD 6 used as a main memory of the system.
[0065]
Next, the contents of various processes executed by the CPU 2 of the evaluation expression extraction device 1 based on a program will be described. FIG. 3 is a functional block diagram of the evaluation expression extraction device 1. As shown in FIG. 3, in the evaluation expression extraction device 1, the input unit 100, the language analysis unit 101, the text data structure generation unit 102, the text data structure storage unit 103, the specific phrase pair selection unit 104, the specific phrase pair storage The functions of the unit 105, the evaluation target / evaluation expression specifying unit 106, the evaluation target / evaluation expression storage unit 107, and the output unit 108 are realized by the CPU 2 operating according to computer software.
[0066]
The input unit 100 stores the text as the “document to be extracted” input from the input device 11 in the document buffer. The extraction of the evaluation expression in the evaluation expression extraction device 1 is performed on the “document to be extracted”. The text that is the “document to be extracted” input from the input device 11 is stored in the document buffer by the input unit 100 and then sent to the language analysis unit 101.
[0067]
The language analysis unit 101 performs a morphological analysis process and a dependency analysis process on the text that is the “document to be extracted” stored in the document buffer. The language analysis unit 101 first performs a morphological analysis process. The morphological analysis process is a process of dividing a text into words and adding attributes of each word such as a part of speech. The principle is described in detail in "Japanese Information Processing Chapter 4" Morphological Analysis "". Existing methods are used, such as a matching method, a minimum cost method, an example search method, etc. Next, the language analysis unit 101 performs a dependency analysis process, which is a clause that is one unit of the dependency process. Is a process to identify the relation between bunsetsu and bunsetsu. The principle is "Japanese information processing Chapter 5" Syntax analysis "" and "Japan based on the dependency between two bunsetsu Use existing methods as described in "Syntax analysis of words" (Yoshida). Normally, a clause is composed of one independent word and zero or more adjuncts. Depending on the analysis method, there is a definition method that produces a result in which one clause includes a plurality of independent words. In the embodiment, an analysis method for generating a phrase is used so that the phrase always contains only one independent word.
[0068]
The text data structure generating means 102 includes information obtained by the language analysis processing (morphological analysis processing and dependency analysis processing) in the language analysis means 101 at least including notation of constituent words and dependency information as shown in FIG. The data is converted into a data structure holding phrase information, and stored in a text data structure storage unit 103 formed in the HDD 6. Here, FIG. 5 is an example of information managed by each component of the text data structure. As shown in FIG. 5, the phrase information includes, in addition to the word information constituting the phrase, related phrase information relating to the relevant phrase, receiving phrase information related to the relevant phrase, and the like.
[0069]
The specific phrase pair selection unit 104 checks whether there is any dependency phrase pair data in the text data structure storage unit 103, places the data into the buffer from the first phrase pair, and determines whether or not the specific phrase pair matches the selection criteria. Then, the matched phrase pair is stored in the specific phrase pair storage unit 105 formed in the HDD 6.
[0070]
The evaluation target / evaluation expression specifying unit 106 checks whether or not the constituent words of the phrase stored in the specific phrase pair storage unit 105 match the evaluation target / evaluation expression specification rules 14 and determines the specified evaluation. The target and the evaluation expression corresponding to the evaluation target are stored in the evaluation target / evaluation expression storage unit 107 formed in the HDD 6.
[0071]
The output unit 108 outputs the evaluation target / evaluation expression data stored in the evaluation target / evaluation expression storage unit 107 to the display device 12.
[0072]
Next, the overall flow of the evaluation expression extraction process in the evaluation expression extraction device 1 will be described with reference to FIG. As shown in FIG. 6, when an instruction to extract an evaluation expression is issued (Y in step S1), the process proceeds to step S2 to check whether there is a document to be extracted.
[0073]
If there is a document to be extracted (Y in step S2), the document is stored in the document buffer in order from the first document (step S3), and the text in the document buffer is subjected to morphological analysis processing (step S4) and dependency analysis processing (Step S5) is performed.
[0074]
When the morphological analysis processing and the dependency analysis processing of the text in the document buffer are completed (Y in step S6), the analysis result is stored in the text data structure storage unit 103 (step S7), and when the next document exists. (Y in step S8), the process returns to step S3, and the next document is stored in the document buffer.
[0075]
If the linguistic analysis processing (morphological analysis processing and dependency analysis processing) has been completed for all documents (N in step S8), the process proceeds to step S9, and the presence / absence of dependency phrase pair data in the text data structure storage unit 103 is determined. Check
[0076]
If there is a dependency phrase pair data in the text data structure storage unit 103 (Y in step S9), the data is put into the buffer in order from the first phrase pair (step S10), and the process proceeds to step S11 to meet the selection criterion. Judge whether it is a specific phrase pair. The selection criterion is, specifically, that the dependency relation is a predicative relation or an adnominal modification relation, and in the former case, the receiving clause contains an adjective, an adjective verb, or an auxiliary adjective, and in the latter case, Is a criterion that a dependency clause contains an adjective, an adjective verb, or an auxiliary adjective.
[0077]
When the phrase pair matches the selection criterion (Y in step S11), the process proceeds to step S12, where the phrase pair is stored in the specific phrase pair storage unit 105 as a specific phrase pair.
[0078]
The check is completed for all the phrase pairs stored in the text data structure storage unit 103 (N in step S13), and when there is a phrase pair stored in the specific phrase pair storage unit 105 (Y in step S14). ), And puts them in the buffer in order from the first phrase pair in the specific phrase pair storage unit 105 (step S15), and specifies the evaluation target / evaluation expression for the dependency relation name, the dependency phrase, and the constituent words of the respective phrases. It is checked whether or not rule 14 is satisfied (step S16).
[0079]
Next, after storing the evaluation target specified by the inspection as to whether it matches the evaluation target / evaluation expression specification rule and the evaluation expression corresponding to the evaluation target in the evaluation target / evaluation expression storage unit 107 (step S17), Proceeding to step S18, it is checked whether there is a next clause.
[0080]
If there is a next phrase (Y in step S18), the next phrase is extracted, and an evaluation target and an evaluation expression corresponding to the evaluation target are extracted (steps S15 to S17). That is, the process of extracting the evaluation target and the evaluation expression corresponding to the evaluation target (steps S15 to S17) is repeated until the specific phrase pair storage unit 105 has no specific phrase pairs (N in step S18).
[0081]
If there is no specific phrase pair in the specific phrase pair storage unit 105 (N in step S18), the process proceeds to step S19 to check for an output instruction.
[0082]
If there is an output instruction (Y in step S19), the process proceeds to step S20, and it is checked whether the evaluation target / evaluation expression data is stored in the evaluation target / evaluation expression storage unit 107.
[0083]
When the evaluation object / evaluation expression data is stored in the evaluation object / evaluation expression storage unit 107 (Y in step S20), the evaluation object / evaluation expression data stored in the evaluation object / evaluation expression storage unit 107 is deleted. Output to the display device 12 (step S21).
[0084]
Here, based on the linguistic analysis result of the linguistic analysis means 101 for performing linguistic analysis processing including at least morphological analysis processing and phrase dependency analysis processing on the text, the text data structure generation means 102 converts the text into at least constituent word notations. Then, it is converted into a data structure holding phrase information including dependency information. Then, specific phrase pair selecting means 104 selects phrase pairs referring to the evaluation of the thing from the phrase pairs determined to be in a dependency relationship, and evaluates the evaluation target and the evaluation expression corresponding to the evaluation target from the selected phrase pairs. Are specified by the evaluation target / evaluation expression specifying means 106. As a result, a pair of an evaluation target and an evaluation expression corresponding to the evaluation target can be extracted from the document, so that the evaluation expression can be applied to analysis of an object in a questionnaire or a document search using the evaluation expression. Can be.
[0085]
[Concrete example]
Here, a specific process in the evaluation expression extraction device 1 as described above will be exemplarily described. Here, there are a plurality of questionnaire free description documents for a certain product, and the user analyzes the product evaluation based on these documents. Note that a keyboard is assumed as the input device 11, and an LCD is assumed as the display device 12. In such a situation, the user activates the evaluation expression extraction device 1 and issues an evaluation expression extraction instruction.
[0086]
The evaluation expression extraction device 1 first checks the presence or absence of a document to be extracted, and if there is a document to be extracted, stores the document in the document buffer in order from the first document, and performs morphological analysis on the text in the document buffer. , Break the text into words. Subsequently, a dependency analysis process is performed to identify a dependency phrase pair between phrases in the text. For example,
A: The price is still too high and I will not buy it.
B “High-priced camera.”
C "Price is hard to fall."
D "It's better to sell."
E "Price is unusual."
If there is such a description sentence, a language analysis processing result as shown in FIG. 7 is obtained.
[0087]
When the dependency analysis processing of the text in the document buffer is completed, the analysis result is stored in the text data structure storage unit 103, and the next document is stored in the document buffer. When these linguistic analyzes have been completed for all documents stored in the document buffer, the presence / absence of dependency phrase pair data in the text data structure storage unit 103 is checked.
[0088]
Since there is phrase pair data as shown in FIG. 7 in the text data structure storage unit 103, the data is buffered in order from the first phrase pair of the data, and is included in the grammatical dependency relationship and the dependency phrase or the reception phrase. Independent words or adjuncts are used as selection criteria to check whether the criteria are met.
[0089]
In this example, the dependency relation is a predicative relation and an adnominal modification relation, and further, a phrase pair having the attributes of the adjective, adjective verb, and auxiliary adjective in the part of speech of the phrase constituent word is selected. From the phrase pair data shown in FIG. 7, five phrase data shown in FIG. 8 are selected.
[0090]
The phrase pairs to be sorted as shown in FIG. 8 are stored in the specific phrase pair storage unit 105. When the selection process is completed for all the dependent phrase pair data stored in the text data structure storage unit 103, the presence / absence of a specific phrase pair in the specific phrase pair storage unit 105 is checked. Since the specific phrase pair storage unit 105 stores the phrase pairs to be sorted as shown in FIG. 8, the phrase pairs are put into the buffer in order from the first phrase pair, and the dependency relation name, the dependency phrase, and the reception phrase are respectively stored. In accordance with the evaluation target / evaluation expression specification rule 14 shown in FIG. 2, an evaluation target and an evaluation expression corresponding to this evaluation target are extracted for the constituent words of the phrase.
[0091]
First, rule 1 is applied to the predicate relation 1 shown in FIG. 8, and since the constituent independent word of the dependency clause is not a formal noun, the evaluation target is “price” which is the independent word of the dependency clause. Next, since the independent word in the receiving phrase is an adjective and does not include an auxiliary verb for cancellation, the evaluation expression is "high", which is the final form of the independent word.
[0092]
The rule 2 is applied to the union modification relation 1 shown in FIG. 8, and the evaluation target is “price” which is an independent word of the receiving clause. Next, since the independent word of the dependency phrase is an adjective and does not include an auxiliary verb for cancellation, the evaluation expression is “high”, which is the final form of the independent word.
[0093]
In the predicate relation 2 shown in FIG. 8, rule 1 is applied, and since the constituent independent word of the dependency clause is not a formal noun, the evaluation target is “price” which is the independent word of the dependency clause. Next, the independent word of the receiving phrase is a verb, but includes the auxiliary adjective "difficult" and does not include the auxiliary verb of defeat, so the evaluation expression is the independent word "fall" + the auxiliary adjective "difficult".
[0094]
In the predicate relation 3 shown in FIG. 8, rule 1 is applied, and since the constituent independent word of the dependency clause is a formal noun, a continuous modifier clause that receives the dependency clause “hoga” as a clause is searched. As shown in FIG. 9, since there is a continuous modifier clause “Sell”, the evaluation target is the independent word “Sell”. Next, since the independent word in the receiving phrase is an adjective and does not include an auxiliary verb for cancellation, the evaluation expression is "good", which is the final form of the independent word.
[0095]
In Rule 4 shown in FIG. 8, rule 1 is applied, and since the constituent independent word of the dependency clause is not a formal noun, the evaluation target is "price" which is the independent word of the dependency clause. Next, since the independent word of the receiving phrase is an adjective verb and includes an auxiliary verb for cancellation, the evaluation expression is "general" which is the final form of the independent word + cancellation.
[0096]
The evaluation object and the evaluation expression thus extracted are stored in the means for storing the evaluation object and the evaluation expression. The specific phrase pair storage unit 105 checks whether there is a next phrase. If the specific phrase pair storage unit 105 has a next phrase, the next phrase is extracted and the specific phrase pair storage unit 105 is deleted until there is no specific phrase pair. The extraction process of the evaluation target and the evaluation expression corresponding to the evaluation target is repeated. Finally, an evaluation target and an evaluation expression as shown in FIG. 10 are extracted.
[0097]
When the extraction process is completed, it is checked whether there is an output instruction. Here, the evaluation target and the corresponding evaluation expression are collectively output based on the evaluation target. Here, FIG. 11 is an example of output to the display device 12. As a method of outputting to the display device 12, as shown in FIG. 11, in addition to a method of collectively outputting evaluation expressions corresponding to evaluation objects, a method of outputting one pair of evaluation objects and evaluation expressions, A method of outputting the evaluation objects corresponding to the evaluation expressions collectively can be considered.
[0098]
【The invention's effect】
According to the evaluation expression extracting apparatus of the first aspect of the present invention, a language analyzing means for performing a language analysis process including at least a morphological analysis process and a phrase dependency analysis process on a text, and a language analysis result in the language analysis device A text data structure generating means for converting the text into a data structure holding phrase information including at least constituent word notation and dependency information based on the text data, and a phrase pair determined to have a dependency relationship by the text data structure generating means. A specific phrase pair selecting means for selecting a phrase pair that meets a predetermined selection criterion, and an evaluation target and an evaluation expression corresponding to the evaluation target are specified from the phrase pairs selected by the specific phrase pair selecting means. By providing an evaluation target / evaluation expression specifying means, a pair of an evaluation target and an evaluation expression corresponding to this evaluation target can be extracted from a document. It is possible to, apply to the analysis of the evaluation of things in such surveys, it is possible to perform such as a document search by evaluation expressions.
[0099]
According to the second aspect of the present invention, in the evaluation expression extracting device according to the first aspect, the selection criterion in the specific phrase pair selecting means is such that the dependency relationship of the target phrase pair is a predicative relationship, Since the constituent parts of speech are adjectives, adjective verbs, and auxiliary adjectives, in the evaluation expression extraction process, the inter-clause dependencies are subject-to-subject relations, and target only phrase pairs that include the terms used in evaluation. Thus, there is an effect that the extraction processing can be reduced as compared with the case where all the phrase pairs are targeted.
[0100]
According to the third aspect of the present invention, in the evaluation expression extracting device according to the first aspect, the selection criterion in the specific phrase pair selecting means is such that a dependency relationship of the target phrase pair is a continuous modification relationship, Since the constituent parts of speech are adjectives, adjective verbs, and auxiliary adjectives, in the evaluation expression extraction process, inter-phrase dependency relations are adnominal modification relations, and target only phrase pairs containing the terms used in evaluation. Thus, there is an effect that the extraction processing can be reduced as compared with the case where all the phrase pairs are targeted.
[0101]
According to the fourth aspect of the present invention, in the evaluation expression extracting device according to the first aspect, the evaluation target / evaluation expression specifying unit is configured to specify an evaluation target from a phrase pair and an evaluation expression corresponding to the evaluation target. When specifying the evaluation target and the evaluation expression by using the rules for specifying the evaluation target / evaluation expression, use the specific rules that utilize natural language grammar knowledge to specify the evaluation target and the evaluation expression with high accuracy can do.
[0102]
According to the fifth aspect of the present invention, in the evaluation expression extracting device according to the fourth aspect, the evaluation target / evaluation expression specifying rule may include a rule that, when the dependency relationship of the target phrase pair is a predicate relationship, Independent words included in the phrase were identified as targets for evaluation, and adjectives, adjective verbs, auxiliary adjectives, independent words preceding the adjective included in the receiving phrase, and additional auxiliary verbs were added if they were included. Since the rules that specify expressions as evaluation expressions are described, the grammatically limited specific rules are included for the dependent clause pairs that have a predicative relationship, so the evaluation target and evaluation expression with higher accuracy Can be extracted.
[0103]
According to the sixth aspect of the present invention, in the evaluation expression extracting apparatus according to the fourth aspect, the evaluation target / evaluation expression specifying rule includes a rule that, when the dependency relationship of the target phrase pair is a predicate relationship, If the independence word included in the phrase is a formal noun, the independence word of the dependency phrase that accepts the phrase and continually modifies it as a phrase is specified as the evaluation target, and the adjective, adjective verb, auxiliary adjective and the If the independence word in front and the auxiliary verb for cancellation are included, the rule that specifies the expression to which it is added as an evaluation expression is described, so that the expression On the other hand, if the evaluation target is a formal noun that is not an entity, the entity can be further obtained from the union-modifying clause, and more accurate evaluation targets and evaluation expressions can be extracted.
[0104]
According to the seventh aspect of the present invention, in the evaluation expression extracting device according to the fourth aspect, the evaluation target / evaluation expression specifying rule may include a rule that, when the dependency relation of the target phrase pair is a continuous modification relation, If an adjective, adjective verb, auxiliary adjective and an independent word in front of it are included in the phrase, and if an auxiliary verb for canceling is included, the expression to which it is added is specified as an evaluation expression, and the independence included in the receiving phrase Since the rules that specify the word as the evaluation target are described, the grammatically limited specific rules are included for the dependent clause pairs having the adnominal modification relation, so that a more accurate evaluation target and evaluation expression can be used. Can be extracted.
[0105]
According to the program of the invention of claim 8, the program is installed on a computer or interpreted and executed, the computer comprising at least a morphological analysis process and a phrase dependency analysis process for text. A language analysis function that performs language analysis processing that includes text, and a text data structure that converts text into a data structure that holds at least the notation of constituent words and phrase information that includes dependency information based on the language analysis result of the language analysis function A specific phrase pair selection function for selecting a phrase pair that meets a predetermined selection criterion from the phrase pairs determined to be dependent by the text data structure generation function, and a selection using the specific phrase pair selection function Evaluation target / evaluation that identifies the evaluation target and the evaluation expression corresponding to this evaluation target from the phrase pairs By executing the expression specifying function, the pair of the evaluation target and the evaluation expression corresponding to this evaluation target can be extracted from the document. Document search by expression can be performed.
[0106]
According to the ninth aspect of the present invention, in the program according to the eighth aspect, the selection criterion in the specific phrase pair selection function is such that the dependency relationship of the target phrase pair is a predicative relationship, and Since the parts of speech are adjectives, adjective verbs, and auxiliary adjectives, in the evaluation expression extraction process, the inter-clause dependency relationship is a predicative relationship, and by targeting only phrase pairs containing the terms used for evaluation, There is an effect that the extraction processing can be reduced as compared with the case of targeting all phrase pairs.
[0107]
According to the tenth aspect of the present invention, in the program according to the eighth aspect, the selection criterion in the specific phrase pair selection function is such that the dependency relation of the target phrase pair is a continuous modification relation, Because the parts of speech are adjectives, adjective verbs, and auxiliary adjectives, in the evaluation expression extraction process, the inter-clause dependency relation is a noun modification relation, and by targeting only phrase pairs containing the terms used for evaluation, There is an effect that the extraction processing can be reduced as compared with the case of targeting all phrase pairs.
[0108]
According to the invention described in claim 11, in the program according to claim 8, the evaluation target / evaluation expression specifying function is used to specify an evaluation target from a phrase pair and an evaluation expression corresponding to the evaluation target. -When specifying the evaluation target and the evaluation expression by using the evaluation expression specification rule, it is possible to specify the evaluation target and the evaluation expression with high accuracy by using the specific rule utilizing the grammatical knowledge of natural language. it can.
[0109]
According to the twelfth aspect of the present invention, in the program according to the eleventh aspect, the evaluation target / evaluation expression specification rule includes the dependency clause in a case where the dependency relation of the target clause pair is a predicative relation. Independent words to be evaluated are identified as evaluation targets, and expressions containing adjectives, adjective verbs, auxiliary adjectives, independent words in front of them, and auxiliary verbs for cancellation are included if they are included in the receiving phrase Extracting higher-precision evaluation targets and evaluation expressions, because grammatically limited specific rules are included for dependency clause pairs that have a predicative relationship by describing rules that specify expressions. Can be.
[0110]
According to a thirteenth aspect of the present invention, in the program according to the eleventh aspect, when the dependency relation of the target clause pair is a predicative relation, the rule for specifying the evaluation target / evaluation expression is included in the dependency clause. If the independence word is a formal noun, the independence word of the continuation clause that modifies the adjunct as a continuation clause is identified as the subject to be evaluated, and the adjective, adjective verb, auxiliary adjective included in the receiving phrase and before it If a rule that specifies an independent word and an ancillary auxiliary verb, if any, as an evaluation expression is described, the phrase is evaluated against a pair of dependent phrases having a predicative relationship. If the target is a formal noun that is not an entity, the entity can be further obtained from the union-qualifying clause, and more accurate evaluation targets and evaluation expressions can be extracted.
[0111]
According to the fourteenth aspect of the present invention, in the program according to the eleventh aspect, when the dependency relation of the target phrase pair is a union modification relation, the evaluation target / evaluation expression specifying rule is included in the dependency phrase. If adjectives, adjective verbs, auxiliary adjectives and independence words preceding the adjectives are included, and if an auxiliary verb for cancellation is included, the expression with the adjective added is specified as the evaluation expression, and the independence word included in the receiving phrase is evaluated Extracting more accurate evaluation targets and evaluation expressions, because specific rules that are grammatically limited are included for dependency clause pairs that have an adnominal modification relationship by describing the rules that identify them as targets. Can be.
[0112]
According to the computer-readable storage medium of the present invention, since the program according to any one of claims 8 to 14 is stored, the program stored in the storage medium is read by the computer. By doing so, the same operation and effect as the invention according to any one of claims 8 to 14 can be obtained.
[0113]
According to the evaluation expression extracting method of the invention according to claim 16, an evaluation expression extracting method performed by computer processing control, wherein a language analysis process including at least a morphological analysis process and a phrase dependency analysis process for text is performed. A language analysis step to be performed, a text data structure generation step of converting a text into a data structure holding phrase information including at least constituent word notation and dependency information based on a result of the language analysis in the language analysis step, and the text data A specific phrase pair selection step of selecting a phrase pair that meets a predetermined selection criterion from the phrase pairs determined to be dependent by the structure generation step, and evaluating the phrase pairs selected in the specific phrase pair selection step. An evaluation object / evaluation expression specifying step of specifying an object and an evaluation expression corresponding to the evaluation object. Since a pair of an evaluation object and an evaluation expression corresponding to the evaluation object can be extracted from a document, it can be applied to an analysis relating to evaluation of an object in a questionnaire or the like, and a document search by the evaluation expression can be performed. .
[0114]
According to the seventeenth aspect of the present invention, in the evaluation expression extracting method according to the sixteenth aspect, the selection criterion in the specific phrase pair selecting step is that a dependency relationship of the target phrase pair is a predicative relationship, Since the constituent parts of speech are adjectives, adjective verbs, and auxiliary adjectives, in the evaluation expression extraction process, the inter-clause dependencies are subject-to-subject relations, and target only phrase pairs that include the terms used in evaluation. Thus, there is an effect that the extraction processing can be reduced as compared with the case where all the phrase pairs are targeted.
[0115]
According to the eighteenth aspect of the present invention, in the evaluation expression extracting method according to the sixteenth aspect, the selection criterion in the specific phrase pair selection step is such that a dependency relationship of the target phrase pair is a continuous modifier relation, Since the constituent parts of speech are adjectives, adjective verbs, and auxiliary adjectives, in the evaluation expression extraction process, inter-phrase dependency relations are adnominal modification relations, and target only phrase pairs containing the terms used in evaluation. Thus, there is an effect that the extraction processing can be reduced as compared with the case where all the phrase pairs are targeted.
[0116]
According to the nineteenth aspect of the present invention, in the evaluation expression extracting method according to the sixteenth aspect, the evaluation target / evaluation expression specifying step includes specifying an evaluation target from a phrase pair and an evaluation expression corresponding to the evaluation target. When specifying the evaluation target and the evaluation expression by using the rules for specifying the evaluation target / evaluation expression, use the specific rules that utilize natural language grammar knowledge to specify the evaluation target and the evaluation expression with high accuracy can do.
[0117]
According to the invention described in claim 20, in the evaluation expression extracting method according to claim 19, the evaluation target / evaluation expression specification rule includes a case where the dependency relation of the target phrase pair is a predicative relation. Independent words included in the phrase were identified as targets for evaluation, and adjectives, adjective verbs, auxiliary adjectives, independent words preceding the adjective included in the receiving phrase, and additional auxiliary verbs were added if they were included. Since the rules that specify the expression as the evaluation expression are described, the grammatically limited specific rules are included for the dependency clause pair having the predicative relationship, so that the evaluation target and the evaluation expression with higher accuracy can be included. Can be extracted.
[0118]
According to the twenty-first aspect of the present invention, in the evaluation expression extracting method according to the nineteenth aspect, the rule for specifying the evaluation target / evaluation expression includes a rule that includes a case where the dependency relation of the target phrase pair is a predicative relation. If the independence word included in the phrase is a formal noun, the independence word of the dependency phrase that accepts the phrase and continually modifies it as a phrase is specified as the evaluation target, and the adjective, adjective verb, auxiliary adjective and the If the independence word in front and the auxiliary verb for cancellation are included, the rule that specifies the expression to which it is added as an evaluation expression is described, so that the expression On the other hand, if the evaluation target is a formal noun that is not an entity, the entity can be further obtained from the union-modifying clause, and more accurate evaluation targets and evaluation expressions can be extracted.
[0119]
According to the invention described in claim 22, in the evaluation expression extraction method according to claim 19, the evaluation target / evaluation expression specification rule includes a rule that, when the dependency relation of the target phrase pair is a continuous modification relation, When an adjective, adjective verb, auxiliary adjective and an independent word preceding it are included in the phrase, and if the auxiliary verb for canceling is included, the expression added with it is specified as an evaluation expression, and the independent word included in the receiving phrase Since the rules that specify the word as the evaluation target are described, the grammatically limited specific rules are included for the dependent phrase pairs that have an adnominal modification relationship, so that more accurate evaluation targets and evaluation expressions can be used. Can be extracted.
[Brief description of the drawings]
FIG. 1 is a block diagram schematically illustrating a hardware configuration of an evaluation expression extraction device according to an embodiment of the present invention.
FIG. 2 is an explanatory diagram showing an example of an evaluation target / evaluation expression specifying rule.
FIG. 3 is a functional block diagram of the evaluation expression extraction device.
FIG. 4 is an explanatory diagram showing a text data structure.
FIG. 5 is an explanatory diagram showing an example of information managed by each component of a text data structure.
FIG. 6 is a flowchart illustrating an overall flow of an evaluation expression extraction process.
FIG. 7 is an explanatory diagram showing an example of a language analysis processing result.
FIG. 8 is an explanatory diagram showing a phrase pair to be sorted;
FIG. 9 is an explanatory view showing a phrase pair to be sorted.
FIG. 10 is an explanatory diagram showing an example of extracted evaluation targets and evaluation expressions.
FIG. 11 is a front view showing an example of output to a display device.
[Explanation of symbols]
1 Evaluation expression extraction device
7 Storage media
14 Rules for specifying evaluation targets and evaluation expressions
101 Language analysis means
102 Text data structure generation means
104 Specific phrase pair selection means
106 Evaluation target / evaluation expression specifying means

Claims (22)

A language analysis unit that performs a language analysis process including at least a morphological analysis process and a phrase dependency analysis process on the text;
Text data structure generation means for converting a text into a data structure holding phrase information including at least constituent word notation and dependency information based on the language analysis result in the language analysis means,
A specific phrase pair selecting unit that selects a phrase pair that matches a predetermined selection criterion from the phrase pairs determined to be in a dependency relationship by the text data structure generating unit;
An evaluation target / evaluation expression specifying unit that specifies an evaluation target and an evaluation expression corresponding to the evaluation target from the phrase pairs selected by the specific phrase pair selection unit;
An evaluation expression extraction device comprising: The selection criterion in the specific phrase pair selection means is that the dependency relationship of the target phrase pair is a predicative relationship, and the parts of speech of the constituent words of the target phrase are adjectives, adjective verbs, and auxiliary adjectives. 2. The evaluation expression extraction device according to 1. The selection criterion in the specific phrase pair selection means, wherein the dependency relationship of the target phrase pair is a continuous modification relationship, and the parts of speech of the constituent words of the dependency phrase are adjectives, adjective verbs, and auxiliary adjectives. 2. The evaluation expression extraction device according to 1. 2. The evaluation method according to claim 1, wherein the evaluation target / evaluation expression specifying unit uses an evaluation target / evaluation expression specification rule when specifying an evaluation target from a phrase pair and an evaluation expression corresponding to the evaluation target. Expression extraction device. The evaluation target / evaluation expression specifying rules include:
If the dependency relationship of the target phrase pair is a predicative relationship, the independent word contained in the dependency phrase is specified as the evaluation target, and the adjective, adjective verb, auxiliary adjective contained in the receiving phrase and the independent word preceding it , And further includes rules for specifying an expression to which the auxiliary verb to be canceled is included as an evaluation expression when the auxiliary verb is included.
5. The evaluation expression extraction device according to claim 4, wherein: The evaluation target / evaluation expression specifying rules include:
If the dependency relation of the target clause pair is a predicate relation, and the independent word contained in the dependency clause is a formal noun, the independent word of the dependency clause that modifies and accepts the phrase as an adjunct is evaluated. A rule that specifies an adjective, adjective verb, auxiliary adjective, an independent word preceding it, and a supplementary adjective, if any, included in the receiving phrase as an evaluation expression Is described,
5. The evaluation expression extraction device according to claim 4, wherein: The evaluation target / evaluation expression specifying rules include:
If the dependency relation of the target phrase pair is an adnominal modification relation, and the adjective, adjective verb, auxiliary adjective included in the dependency phrase and an independent word preceding it, and an auxiliary verb for cancellation, A rule that specifies an expression to which it is added as an evaluation expression and specifies an independent word included in the receiving phrase as an evaluation target is described.
5. The evaluation expression extraction device according to claim 4, wherein: A program installed or interpreted and executed on a computer, wherein the computer has
A language analysis function for performing a language analysis process on the text including at least a morphological analysis process and a phrase dependency analysis process;
A text data structure generation function of converting a text into a data structure holding phrase information including at least constituent word notation and dependency information based on a result of the language analysis in the language analysis function;
A specific phrase pair selection function of selecting a phrase pair that matches a predetermined selection criterion from the phrase pairs determined to have a dependency relationship by the text data structure generation function;
An evaluation target / evaluation expression identification function for identifying an evaluation target and an evaluation expression corresponding to the evaluation target from the phrase pairs selected by the specific phrase pair selection function;
A program characterized by executing The selection criterion in the specific phrase pair selection function is that the dependency relationship of the target phrase pair is a predicative relationship, and the parts of speech of the constituent words of the target phrase are adjectives, adjective verbs, and auxiliary adjectives. 8. The program according to 8. The selection criterion in the specific phrase pair selection function, wherein the dependency relationship of the target phrase pair is a continuous modification relationship, and the parts of speech of the constituent words of the dependency phrase are adjectives, adjective verbs, and auxiliary adjectives. 8. The program according to 8. 9. The program according to claim 8, wherein the evaluation target / evaluation expression specifying function uses an evaluation target / evaluation expression specification rule when specifying an evaluation target from a phrase pair and an evaluation expression corresponding to the evaluation target. . The evaluation target / evaluation expression specifying rules include:
If the dependency relationship of the target phrase pair is a predicative relationship, the independent word contained in the dependency phrase is specified as the evaluation target, and the adjective, adjective verb, auxiliary adjective contained in the receiving phrase and the independent word preceding it , And further includes rules for specifying an expression to which the auxiliary verb to be canceled is included as an evaluation expression when the auxiliary verb is included.
The program according to claim 11, characterized in that: The evaluation target / evaluation expression specifying rules include:
If the dependency relation of the target clause pair is a predicate relation, and the independent word contained in the dependency clause is a formal noun, the independent word of the dependency clause that modifies and accepts the phrase as an adjunct is evaluated. A rule that specifies an adjective, adjective verb, auxiliary adjective, an independent word preceding it, and a supplementary adjective, if any, included in the receiving phrase as an evaluation expression Is described,
The program according to claim 11, characterized in that: The evaluation target / evaluation expression specifying rules include:
If the dependency relation of the target phrase pair is an adnominal modification relation, and the adjective, adjective verb, auxiliary adjective included in the dependency phrase and an independent word preceding it, and an auxiliary verb for cancellation, A rule that specifies an expression to which it is added as an evaluation expression and specifies an independent word included in the receiving phrase as an evaluation target is described.
The program according to claim 11, characterized in that: A computer-readable storage medium storing the program according to any one of claims 8 to 14. An evaluation expression extraction method performed by processing control of a computer,
A language analysis step of performing a language analysis process on the text including at least a morphological analysis process and a phrase dependency analysis process;
A text data structure generating step of converting a text into a data structure holding phrase information including at least constituent word notation and dependency information based on a result of the language analysis in the language analysis step;
A specific phrase pair selecting step of selecting a phrase pair that matches a predetermined selection criterion from the phrase pairs determined to have a dependency relationship by the text data structure generating step;
An evaluation target / evaluation expression specifying step of specifying an evaluation target and an evaluation expression corresponding to the evaluation target from the phrase pairs selected in the specific phrase pair selection process;
An evaluation expression extraction method, comprising: The selection criterion in the specific phrase pair selection step is that the dependency relationship of the target phrase pair is a predicative relationship, and the parts of speech of the constituent words of the target phrase are adjectives, adjective verbs, and auxiliary adjectives. 16. The method for extracting an evaluation expression according to item 16. The selection criterion in the specific phrase pair selection step, wherein the dependency relationship of the target phrase pair is a continuous modification relationship, and the parts of speech of the constituent words of the dependency phrase are adjectives, adjective verbs, and auxiliary adjectives. 16. The method for extracting an evaluation expression according to item 16. 17. The evaluation according to claim 16, wherein the evaluation target / evaluation expression specifying step uses an evaluation target / evaluation expression specification rule when specifying an evaluation target from a phrase pair and an evaluation expression corresponding to the evaluation target. Expression extraction method. The evaluation target / evaluation expression specifying rules include:
If the dependency relationship of the target phrase pair is a predicative relationship, the independent word contained in the dependency phrase is specified as the evaluation target, and the adjective, adjective verb, auxiliary adjective contained in the receiving phrase and the independent word preceding it , And further includes rules for specifying an expression to which the auxiliary verb to be canceled is included as an evaluation expression when the auxiliary verb is included.
20. The evaluation expression extracting method according to claim 19, wherein: The evaluation target / evaluation expression specifying rules include:
If the dependency relation of the target clause pair is a predicate relation, and the independent word contained in the dependency clause is a formal noun, the independent word of the dependency clause that modifies and accepts the phrase as an adjunct is evaluated. A rule that specifies an adjective, adjective verb, auxiliary adjective, an independent word preceding it, and a supplementary adjective, if any, included in the receiving phrase as an evaluation expression Is described,
20. The evaluation expression extracting method according to claim 19, wherein: The evaluation target / evaluation expression specifying rules include:
If the dependency relation of the target phrase pair is an adnominal modification relation, and the adjective, adjective verb, auxiliary adjective included in the dependency phrase and an independent word preceding it, and an auxiliary verb for cancellation, A rule that specifies an expression to which it is added as an evaluation expression and specifies an independent word included in the receiving phrase as an evaluation target is described.
20. The evaluation expression extracting method according to claim 19, wherein:

Images