JP2018198002A - Document processing device, document processing method and program - Google Patents

Abstract

To extract a rule from a rule description document with accuracy.SOLUTION: A dependency analysis unit of a document processing device extracts a dependency structure of clauses constituting a rule description document and a dependency structure of clauses constituting a comparison object document. A likelihood ratio calculation unit calculates, for each of destination clauses constituting the rule description document, a likelihood ratio indicating the degree that the destination clauses appear more in the rule description document than the comparison object document based on the dependency structure of the clauses extracted from the rule description document and the dependency structure of the clauses extracted from the comparison object document by the dependency analysis unit. A destination clause extraction unit extracts, among the destination clauses constituting the rule description document, a predetermined number of the destination clauses of which a value of the likelihood ratio calculated by the likelihood calculation unit is relatively large or the destination clauses of which a value of the likelihood ratio calculated by the likelihood ratio calculation unit is larger than the threshold.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an apparatus, a method, and a program for extracting a phrase that appears biased in one document of a set of documents.

  Conventionally, in the field of natural language processing, a method for extracting a word that appears biased in one document of a set of documents is known. For example, Non-Patent Document 1 is a method for extracting technical terms that are exclusively used for general people among technical terms, and numerically indicates the bias in the use of technical terms in a general corpus and a professional corpus. And a method for extracting technical terms that are biased in the public corpus and extracting them as general technical terms.

Kim Ailan, Rika Kiryu, Asako Kondo, Makuro Tanaka, “Attempts to Extract“ Terminology for the General Public: Medical Terminology ””, Proceedings of the 2008 Spring Conference of the Japanese Society, Japanese Association for Japanese Studies, 2008 5 Moon, pp.199-206

  Conventionally, it has been difficult to quickly understand the contents of a screening viewpoint description document that describes the viewpoint of screening used in screening work. In order to assist the quick understanding of the examination viewpoint document, it may be possible to extract and list only the portion of the document that describes the examination viewpoint, using the conventional technology described above. It was not possible to extract accurately with the method using.

  The present invention has been made in view of such circumstances, and an object thereof is to extract a rule from a rule description document with high accuracy.

  In order to solve the above problems, a document processing apparatus according to the present invention includes a dependency structure of a clause constituting a rule description document, a dependency analysis unit for extracting a dependency structure of a clause constituting a comparison target document, For each of the dependency clauses constituting the rule description document, based on the dependency structure of the clause extracted from the rule description document by the dependency analysis unit and the dependency structure of the clause extracted from the comparison target document A likelihood ratio calculating unit that calculates a likelihood ratio indicating a degree of appearance of the dependency clause in the rule description document rather than the comparison target document, and the dependency clause constituting the rule description document, A predetermined number of dependency clauses having a relatively large likelihood ratio value calculated by the likelihood ratio calculation unit or a dependency clause having a likelihood ratio value calculated by the likelihood ratio calculation unit larger than a threshold value Extract And a dependency destination clause extractor.

  In a preferred aspect, the document processing apparatus is configured to, for each of the dependency clauses extracted by the dependency clause extraction unit, based on the dependency structure of the clause extracted from the rule description document by the dependency analysis unit. Further, a relation source phrase extraction unit for extracting the relation source phrase is further provided.

  In a further preferred aspect, the document processing device described above includes a source phrase extracted by the source phrase extracting unit for each of the destination phrases extracted by the destination phrase extracting unit, and a word included in the source phrase. And a related phrase classification unit for classifying based on the meaning of

  In a more preferred aspect, the dependency source phrase extraction unit extracts a relationship source phrase representing a condition.

  In a further preferred aspect, the document processing apparatus analyzes the structure of the rule description document and the structure of the comparison target document, and configures the comparison target document corresponding to one sentence constituting the rule description document. A document structure analysis unit that identifies other sentences to be processed, and the likelihood ratio calculation unit includes, for each of the dependency phrases constituting the one sentence, the dependency phrase is more than the other sentence. A likelihood ratio indicating the degree of appearance in a sentence is calculated, and the dependency clause extraction unit calculates the likelihood ratio calculated by the likelihood ratio calculation unit among the dependency clauses constituting the one sentence. A predetermined number of dependency clauses having a relatively large value or a dependency clause whose likelihood ratio value calculated by the likelihood ratio calculation unit is larger than a threshold value is extracted.

  The document processing method according to the present invention is a document processing method executed by a document processing apparatus, and includes a dependency structure of clauses constituting a rule description document and a dependency structure of clauses constituting a comparison target document. For each of the dependency analysis step to extract and the dependency clauses constituting the rule description document, the dependency structure extracted from the rule description document and the comparison target document extracted by the dependency analysis step A likelihood ratio calculating step for calculating a likelihood ratio indicating a degree of occurrence of the dependency clause in the rule description document rather than the comparison target document based on the dependency structure of the clause; and the rule description document Among the dependency clauses to be configured, a predetermined number of dependency clauses having a relatively large likelihood ratio value calculated by the likelihood ratio calculation step or the likelihood ratio calculation step The value of the calculated likelihood ratio Ri comprises a dependency destination clause extracting a large dependency destination clause than the threshold value.

  Further, the program according to the present invention includes a computer, a dependency structure of clauses constituting a rule description document, a dependency analysis unit for extracting a dependency structure of clauses constituting a comparison target document, and the rule description document. For each of the dependency clauses constituting the dependency clause, based on the dependency structure of the clause extracted from the rule description document by the dependency analysis unit and the dependency structure of the clause extracted from the comparison target document, Is a likelihood ratio calculation unit that calculates a likelihood ratio indicating a degree of appearance of the rule-description document rather than the comparison target document, and the likelihood ratio calculation among the related clauses constituting the rule-description document A predetermined number of dependency clauses having a relatively large likelihood ratio value calculated by the unit or a dependency clause having a likelihood ratio value calculated by the likelihood ratio calculation unit larger than a threshold value To function as a clause extractor.

  According to the present invention, it is possible to accurately extract a rule from a rule description document.

2 is a diagram illustrating an example of a configuration of a document processing apparatus 1. FIG. It is a flowchart which shows an example of an examination viewpoint extraction process. 4 is a diagram illustrating an example of a table stored in a document storage unit 103. FIG. It is a figure which shows an example of the result of the analysis by the dependency analysis part 104. It is a figure which shows an example of the table memorize | stored in the dependency structure storage part. It is a figure which shows an example of the table memorize | stored in the examination viewpoint description document receiving part memory | storage part. 6 is a diagram illustrating an example of a table stored in a comparison target document receiver storage unit 108. FIG. It is a figure which shows an example of the table memorize | stored in the receiver list memory | storage part 111. FIG. It is a figure which shows an example of the table memorize | stored in the related clause list memory | storage part 113. FIG. It is a figure which shows an example of the table memorize | stored in the examination logic list memory | storage part.

1. Embodiment 1-1. Configuration FIG. 1 is a diagram showing an example of a configuration of a document processing apparatus 1 according to an embodiment of the present invention. The document processing device 1 is a computer that includes an arithmetic processing device such as a CPU, a storage device such as an HDD, and a communication device such as a NIC, and extracts a viewpoint of examination from a document described in an examination viewpoint. As shown in FIG. 1, the document processing apparatus 1 includes a document input unit 101, a document structure analysis unit 102, a document storage unit 103, a dependency analysis unit 104, a dependency structure storage unit 105, and a clause extraction. Unit 106, examination viewpoint description document receiver storage unit 107, comparison target document receiver storage unit 108, LLR (Log-Likelihood Ratio) calculation unit 109, receiver list creation unit 110, and receiver list storage unit 111, a clause list creation unit 112, a clause list storage unit 113, a clause classification unit 114, a screening logic list creation unit 115, a screening logic list storage unit 116, and an output unit 117. Among these functions, the document storage unit 103, the dependency structure storage unit 105, the examination viewpoint description document receiver storage unit 107, the comparison target document receiver storage unit 108, and the receiver list storage unit 111, The clause list storage unit 113 and the examination logic list storage unit 116 are realized by a storage device, and other functions are realized by the arithmetic processing device executing a program stored in the storage device. The document processing apparatus 1 may be composed of a plurality of server apparatuses connected to each other via a communication line.

  Among the functions of the document processing apparatus 1, the document input unit 101 acquires examination viewpoint description document data and comparison target document data from a storage device or a communication device. Here, the examination viewpoint description document is a document in which the examination viewpoint is described. In other words, it is a procedure manual or a rule book. The examination viewpoint description document is mainly composed of a set of conditional clauses and sentences (for example, “B in case of A”). As a specific example of the examination viewpoint document, notice of notice is given. Here, the notice of notice is a document in which notice of implementation of the notified contents is described. Examples of notices include notices on implementation due to a partial revision of the calculation method of medical fees, notes on material price calculation for specified insurance medical materials, and the amount of expenses required for specific education and childcare. There are implementation considerations accompanying the revision of the calculation standards and implementation considerations accompanying the revision of the calculation standards for the amount of expenses required for designated welfare services for the disabled. Next, the comparison target document is a document to be compared with the examination viewpoint description document. The document to be compared is a document that is referred to in order to extract only the viewpoint of examination out of the contents described in the examination viewpoint description document, so the field of description contents overlaps with the examination viewpoint description document, and the creator is the same It is preferable that Further, it preferably has a document structure corresponding to the examination viewpoint description document. Specifically, it is preferable that the contents related to each other are described in the same order between the comparison target document and the examination viewpoint description document. A specific example of the comparison target document is notification. When the points to be noted in implementation due to the partial revision of the calculation method of medical fees are entered as a document that describes the examination viewpoint, a notification to revise a part of the calculation method of medical fees will be issued as a comparison target document. Entered. When considerations related to the calculation of material prices for specified insurance medical materials are entered as a document for the viewpoint of examination, a notification regarding the revision of a part of the material prices for specified insurance medical materials is entered as a comparison target document . When the notes for implementation accompanying the revision of the calculation standard for the amount of expenses required for specific education / childcare, etc. are entered as a document for the viewpoint of examination, the cost of expenses required for specific education / childcare, etc. will be used as a comparison target document. Notification of the revision of a part of the amount calculation standard is entered. When the notes for implementation accompanying the revision of the calculation standard for the amount of expenses required for the designated disability welfare service, etc. are entered as the document for the examination viewpoint, Notification of the revision of a part of the amount calculation standard is entered.

  The document structure analysis unit 102 analyzes the document structure for each of the examination viewpoint description document represented by the examination viewpoint description document data acquired by the document input unit 101 and the comparison target document represented by the comparison target document data. The sentence constituting the comparison target document corresponding to each sentence constituting the examination viewpoint description document is specified. At that time, the document structure analysis unit 102 specifies the correspondence between sentences by using the numbers assigned to the headings of the sentences and blank lines in the document as clues. Here, the heading is a heading of a division such as a chapter (part), a chapter, a section, or a paragraph. When the document structure analysis unit 102 identifies corresponding sentences in the comparison target document for the sentences constituting the examination viewpoint description document, the document structure analysis unit 102 associates the corresponding sentence sets with the identification information of the classification to which the sentences belong, and stores the document. Store in the unit 103. If the document structure analysis unit 102 cannot specify the correspondence between the examination viewpoint description document and the comparison target document in units of sentences, the document structure analysis unit 102 converts the entire examination viewpoint description document and the entire comparison target document into a set of corresponding sentences. Is stored in the document storage unit 103.

  The dependency analysis unit 104 divides the sentence of the examination viewpoint description document corresponding to each category stored in the document storage unit 103 and the sentence of the comparison target document into sentence units, performs dependency analysis, and composes the sentence. The dependency structure (in other words, dependency tree) of the phrase to be extracted is extracted. At that time, dependency analysis includes, for example, CaboCha (https://taku910.github.io/cabocha/) and KNP (http://nlp.ist.i.kyoto-u.ac.jp/index. You may use a well-known dependency analyzer such as php? KNP). When the dependency analysis unit 104 extracts the dependency structure of the clauses for the sentence of the examination viewpoint description document and the sentence of the comparison target document corresponding to one category, the set of the extracted dependency structure and the identification of the one category are extracted. The information is stored in the dependency structure storage unit 105 in association with the information.

  The receiver extracting unit 106 refers to the dependency structure of the clauses of the examination viewpoint description document and the comparison target document corresponding to each category stored in the dependency structure storage unit 105, and determines the dependency destinations constituting the dependency relationship. (Hereinafter simply referred to as “acceptance clause”). The receiving clauses extracted here are in particular the verbs at the end of the sentence. When the receiving clause extracting unit 106 extracts a receiving clause by referring to the clause dependency structure of the examination viewpoint description document corresponding to one category, the receiving clause and the frequency at which the receiving clause appears in the one category. (In other words, the number of appearances) is stored in the examination viewpoint description document receiver storage unit 107 in association with the identification information of the one category. On the other hand, when a receiving clause is extracted with reference to the dependency structure of the clause of the comparison target document corresponding to one category, the receiving clause and the frequency at which the receiving clause appears in the one category (in other words, the number of appearances) Are stored in the comparison target document receiver storage unit 108 in association with the identification information of the one category.

The LLR calculation unit 109 compares, for each receiver stored in the examination viewpoint description document receiver storage unit 107, the classification of the examination viewpoint description document from which the receiver is extracted, with the same classification of the comparison target document. Then, a log likelihood ratio indicating the degree of occurrence of bias is calculated by using the formula 1 with reference to the examination viewpoint description document receiver storage unit 107 and the comparison target document receiver storage unit 108.

  In the formula (1), a represents the frequency at which the receiving node for which the log likelihood ratio is calculated appears in the classification of the examination viewpoint description document from which the receiving node is extracted. b represents the frequency at which the receiving node appears in the same section of the comparison target document. c represents the frequency at which receiving clauses other than the receiving clause appear in the same category of the examination viewpoint description document. “d” represents the frequency at which a receiving node other than the receiving node appears in the same section of the comparison target document. N represents the number of documents (in this embodiment, “2”). Here, log is a common logarithm. After calculating the log likelihood ratio for the receiving node, the LLR calculating unit 109 stores the calculated log likelihood ratio in the examination viewpoint description document receiving node storage unit 107 in association with the receiving node.

  The receiving clause list creation unit 110 extracts a predetermined number of receiving clauses having a relatively large log likelihood ratio value from the receiving clauses stored in the examination viewpoint description document receiving clause storage unit 107. When a predetermined number of receiving clauses are extracted, the receiving clauses are associated with the extracted receiving clauses, the identification information of the categories from which the receiving clauses are extracted, and the log likelihood ratios of the receiving clauses in the order of log likelihood ratios. Store in the list storage unit 111. The receiving clause list creating unit 110 is an example of a “relevant phrase extracting unit” according to the present invention.

  The dependency clause list creation unit 112, for each reception clause stored in the reception clause list storage unit 111, among the dependency structure of the clauses of the examination viewpoint description document stored in the dependency structure storage unit 105, With reference to the dependency structure of the clause corresponding to the category from which the clause is extracted, the dependency clause that forms the dependency relationship with the reception clause (hereinafter simply referred to as “dependency clause”) is extracted. The related clause extracted here is a related clause included in the same sentence as the receiving clause, and is a related clause including an object directly related to the receiving clause. When the related clause list creation unit 112 extracts the related clause, the extracted related clause, the receiving clause related to the related clause, the identification information of the category from which the related clause is extracted, and the related clause appear in the relevant category. Are stored in the clause list storage unit 113 in association with each other (in other words, the number of appearances). The related clause list creation unit 112 is an example of the “related clause extraction unit” according to the present invention.

  The clause classification unit 114 is a clause extracted in the same category as the clause for each of the clauses stored in the clause list storage unit 113, and is a clause having a common receiver clause. Classify based on the meaning of the words included in the clause. At that time, the clause classification unit 114 refers to a synonym dictionary (not shown) and assigns the same classification code to the clause group including the synonyms. Alternatively, the distance between the related clauses is calculated using a well-known similarity calculation method, and the same classification code is assigned to the related clause group whose calculated distance is within the threshold. The classification code assigned to the relevant clause is stored in the relevant clause list storage unit 113 in association with the relevant clause.

  The examination logic list creation unit 115, for each reception stored in the dependency list storage unit 113, of the dependency structure of the clause of the examination viewpoint description document stored in the dependency structure storage unit 105 receives the reception structure. With reference to the dependency structure of the clause corresponding to the category from which the clause is extracted, an attempt is made to extract a dependency clause that constitutes the dependency relationship with the reception clause and represents a condition. In particular, the extracted clause is a clause included in the same sentence as the receiving clause, and is a related clause directly related to the receiving clause. The examination logic list creation unit 115 uses a keyword representing a condition such as “when”, “when”, and “when” as a clue when extracting a clause representing the condition. When the examination logic list creation unit 115 extracts the clauses representing the conditions, the examination logic list creation unit 115 further extracts the clauses related to the extracted clauses and specifies them as the condition portion. In addition, a related clause included in the same sentence as the related clause representing the extracted condition, which is directly related to the received clause stored in the related clause list storage unit 113 and includes an object, is extracted. Then, it is specified as a processing unit together with the receiving node. Note that the extraction may be omitted when there is no such clause. When the condition part and the processing part are specified, they are stored in the examination logic list storage part 116 in association with the extracted classification identification information. The examination logic list creation unit 115 is an example of a “relationship phrase extraction unit” according to the present invention.

  The output unit 117 refers to the receiver list storage unit 111 and outputs a receiver list. Further, a related clause list is output with reference to the related clause list storage unit 113. When outputting the related clause list, the output unit 117 groups and outputs related clause groups having the same classification code, which will be described later. Further, the output unit 117 refers to the screening logic list storage unit 116 and outputs a screening logic list. Here, output refers to screen output of information or print output or transmission of data representing information.

1-2. Operation A document processing method executed by the document processing apparatus 1 will be described. Specifically, the examination viewpoint extraction process for extracting the examination viewpoint from the examination viewpoint description document will be described. FIG. 2 is a flowchart showing an example of examination viewpoint extraction processing. In the explanation of this operation example, precautionary points related to the partial revision of the calculation method for medical fees are entered as a document describing the examination viewpoint, and a part of the calculation method for medical fees is revised as a comparative document. Assume that a notification of the matter to be entered is input.

  When the document input unit 101 acquires the document data for examination viewpoint description and the document data to be compared, the document structure analysis unit 102 compares the document for examination viewpoint description represented by the obtained document data for examination viewpoint description with the acquired comparison. For each of the comparison target documents represented by the target document data, the document structure is analyzed, and the sentences constituting the comparison target document corresponding to the sentences constituting the examination viewpoint description document are specified (S1). When a corresponding sentence in the comparison target document is specified for each sentence constituting the examination viewpoint description document, the corresponding sentence pair and the identification information of the category to which the sentence belongs are associated and stored in the document storage unit 103. FIG. 3 is a diagram illustrating an example of a table stored in the document storage unit 103. In the table shown in the figure, for example, a set of sentences of the document described in the examination viewpoint and the sentence of the comparison target document is associated with the classification identification information “Chapter 1”, “Part 1”, and “General Rules”. , A set of sentences corresponding to each other is stored.

  When the correspondence relationship between the examination viewpoint description document and the comparison target document is specified, the dependency analysis unit 104 compares the sentence with the examination viewpoint description document corresponding to each category stored in the document storage unit 103. The sentence of the target document is divided into sentence units, dependency analysis is performed, and the dependency structure of the clauses constituting the sentence is extracted (S2). As an example, FIG. 4 shows a dependency structure that is output when a dependency analysis unit 104 analyzes the dependency of the sentence “Calculate the initial fee when there is a medical practice called medically as the first diagnosis”. It is a figure which shows an example. In the dependency structure shown in the figure, the first, fourth, seventh, tenth, fourteenth, seventeenth, twenty-first and twenty-fourth lines starting with an asterisk indicate the phrase number and the phrase number of the relation destination (“-1” when there is no relation destination). The other lines are composed of the surface form of the word, the part of speech, the part of speech subcategory 1, the part of speech subcategory 2, the part of speech subcategory 3, the inflection type, the inflection form, the original form, the reading and the pronunciation. An asterisk in the other lines indicates that the information is not registered in the dictionary. The dependency structure shown in the figure indicates that the phrase “medically” is related to the phrase “speak”. When the dependency analysis unit 104 extracts the dependency structure of the clause for the sentence of the examination viewpoint description document and the sentence of the comparison target document corresponding to each category, the combination of the extracted dependency structure and the identification information of the category are obtained. The associated structure is stored in the dependency structure storage unit 105. FIG. 5 is a diagram illustrating an example of a table stored in the dependency structure storage unit 105. In the table shown in the figure, for example, the dependency structure of the clause of the document described in the examination viewpoint and the clause of the comparison target document are associated with the identification information “Chapter 1”, “Part 1”, and “General” of the classification. A set of dependency structures is stored.

  When the dependency structure is extracted for the sentence of the examination viewpoint description document corresponding to each category and the sentence of the comparison target document, the receiving node extraction unit 106 corresponds to each category stored in the dependency structure storage unit 105. Referring to the dependency structure of the clauses of the examination viewpoint description document and the comparison target document, the reception clauses constituting the dependency relationship are extracted (S3). When a receiving clause is extracted with reference to the dependency structure of the clause of the examination viewpoint document corresponding to one category, the receiving clause and the frequency at which the receiving clause appears in the one category are determined. And stored in the examination viewpoint description document receiver storage unit 107 in association with the identification information. FIG. 6 is a diagram illustrating an example of a table stored in the examination viewpoint description document receiver storage unit 107. In the table shown in the figure, for example, the receiving section “describe”, the application frequency “48”, and the LLR “80. 78 "is stored. The LLR “80.78” is stored as a result of execution of step S4 described later. On the other hand, when the receiving clause extracting unit 106 extracts a receiving clause by referring to the dependency structure of the clause of the comparison target document corresponding to one division, the receiving clause and the receiving clause appear in the one division. The frequency is stored in the comparison target document receiver storage unit 108 in association with the identification information of the one category. FIG. 7 is a diagram illustrating an example of a table stored in the comparison target document receiver storage unit 108. In the table shown in the figure, for example, the receiving section “describe” and the application frequency “8” are stored in association with the classification identification information “Chapter 1”, “Part 1”, and “General Rules”. .

  When the receiving clause is extracted from the examination viewpoint description document and the comparison target document, the LLR calculation unit 109 extracts the receiving clause for each receiving clause stored in the examination viewpoint description document receiving clause storage unit 107. The log likelihood ratio indicating the degree of appearance of the viewpoint description document in comparison with the same section of the comparison target document is set to the examination viewpoint description document receiver storage unit 107 and the comparison target document receiver storage unit 108. (S4). When the log likelihood ratio is calculated for each receiver node, the calculated log likelihood ratio is stored in the examination viewpoint description document receiver storage unit 107 in association with the receiver node.

  When the log likelihood ratio is calculated for each receiver, the receiver list creation unit 110 compares the log likelihood ratio value among the receivers stored in the examination viewpoint description document receiver storage unit 107. A predetermined number of receiving nodes that are large is extracted (S5). When a predetermined number of receiving clauses are extracted, the receiving clauses are associated with the extracted receiving clauses, the identification information of the categories from which the receiving clauses are extracted, and the log likelihood ratios of the receiving clauses in the order of log likelihood ratios. Store in the list storage unit 111. FIG. 8 is a diagram illustrating an example of a table stored in the receiver list storage unit 111. In the table shown in the figure, for example, the receiving clause “Implement” and LLR “326.54” are stored in association with the identification information “Chapter 1”, “Part 1”, and “General” of the classification. Yes.

  When the acceptance list is created, the dependency list creation unit 112, for each reception clause stored in the reception list storage unit 111, the clauses of the examination viewpoint description document stored in the dependency structure storage unit 105. Among the dependency structures, the dependency structure of the dependency relationship with the reception clause is extracted with reference to the dependency structure of the clause corresponding to the category from which the reception clause is extracted (S6). When extracting a related clause, the extracted related clause, the receiving clause related to the relevant clause, the identification information of the category from which the relevant clause is extracted, and the frequency of occurrence of the relevant clause in the relevant category are associated with each other. Store in the clause list storage unit 113. FIG. 9 is a diagram illustrating an example of a table stored in the related clause list storage unit 113. The table stored in the storage unit is created for each receiver stored in the receiver list storage unit 111. In the table shown in the figure, for example, classification identification information “Chapter 1”, “Part 1” ”And“ general rule ”are stored in association with the clause“ reason ”, the appearance frequency“ 27 ”, and the classification code“ 1 ”. The classification code “1” is stored as a result of execution of step S7 described later.

  When the clause list is created, the clause classification unit 114 is a clause extracted in the same category as the clause for each clause stored in the clause list storage unit 113. Are classified based on the meanings of the words included in the related clauses (S7). When a classification code is assigned to each relevant clause, it is stored in the relevant clause list storage unit 113 in association with the relevant clause.

When each dependency clause is classified, the examination logic list creation unit 115, for each reception clause stored in the dependency list storage unit 113, the clause of the examination viewpoint description document stored in the dependency structure storage unit 105. The dependency structure of the dependency structure of the dependency relationship with the reception clause, with reference to the dependency structure of the clause corresponding to the category from which the dependency clause is extracted. Attempt to extract. When a clause representing a condition is extracted, a clause related to the extracted clause is further extracted and specified as a condition part. In addition, a related clause included in the same sentence as the related clause representing the extracted condition, which is directly related to the received clause stored in the related clause list storage unit 113 and includes an object, is extracted. Then, it is specified as a processing unit together with the receiver (S8). When the condition part and the processing part are specified, they are stored in the examination logic list storage part 116 in association with the extracted classification identification information. FIG. 10 is a diagram illustrating an example of a table stored in the examination logic list storage unit 116. In the table shown in the figure, for example, the condition part “in case of emergency” and the processing part “describe the reason” are associated with the classification identification information “Chapter 1”, “Part 1”, and “General Rules”. Is stored.
The above is the description of the examination viewpoint extraction process.

  In the document processing apparatus 1 described above, the receiver list creation unit 110 creates a list of receivers that appear specifically in the examination viewpoint description document. By referring to this receiving list, readers who are going to read the examination viewpoint document can know which paragraphs in this document can quickly understand the examination viewpoint. In addition, in this document processing apparatus 1, since the log likelihood ratio is calculated by comparing sentences related to each other between the examination viewpoint description document and the comparison target document, compared with a case where the log likelihood ratio is not so, The extraction accuracy is improved. Further, since extraction is performed not in units of words but in terms of phrases, for example, it can be distinguished whether it is “described” or “not described”. In addition, since extraction is performed only for verbs at the end of a sentence and not for all phrases, extraction of noise is suppressed as compared with the case where it is not.

  In the document processing apparatus 1, the related clause list creating unit 112 creates a list of related clauses related to the receiving clauses that appear specifically in the examination viewpoint description document. By referring to this section list, it becomes easier for readers who will read the examination viewpoint description document to understand the examination viewpoint. In addition, each of the related clauses is classified by the related clause classification unit 114, and the related clause groups having the same classification code are grouped and output, so that it is easy to grasp the outline of the examination viewpoint.

  In the document processing apparatus 1, the examination logic list creation unit 115 extracts a list of examination logic including a condition part and a processing part. By referring to this list of examination logic, an outline of the examination viewpoint can be grasped without reading the examination viewpoint description document.

2. Modifications The above embodiment may be modified as described below. Note that one or more modified examples described below may be combined with each other.

2-1. Modification 1
The document structure analysis unit 102 may be omitted. In that case, the document storage unit 103 corresponds to a set of the entire examination viewpoint description document represented by the examination viewpoint description document data acquired by the document input unit 101 and the entire comparison target document represented by the comparison target document data. Is stored in the document storage unit 103 as a set of sentences to be processed.

2-2. Modification 2
The document processing apparatus 1 is assumed to process a Japanese document, but may process a language other than Japanese. For example, when processing an English document, the dependency analysis unit 104 uses, for example, Maltparse (http://www.maltparser.org/) to create a dependency structure of clauses constituting an English document. It may be extracted.

2-3. Modification 3
The above-mentioned receiving clause extracting unit 106 extracts receiving clauses that are verbs at the end of a sentence. However, the extracted receiving clauses may not be limited to receiving clauses at the end of a sentence or receiving clauses of verbs.

2-4. Modification 4
Although the common logarithm is taken in the above formula 1 for calculating the log likelihood ratio, a natural logarithm may be taken. Alternatively, the logarithm may not be taken in the first place.

The formula for calculating the log likelihood ratio is not limited to the formula 1 above. The LLR calculation unit 109 may calculate the log likelihood ratio using, for example, the formula 2 instead of the formula 1.

  In Expression 2, a represents the frequency at which the receiving node for which the log likelihood ratio is calculated appears in the classification of the examination viewpoint description document from which the receiving node is extracted. b represents the frequency at which the receiving node appears in the same section of the comparison target document. c represents the frequency of occurrence of all receiving clauses in the same category of the examination viewpoint description document (in other words, the total number of receiving clauses). d represents the frequency at which all receiving clauses appear in the same section of the comparison target document (in other words, the total number of receiving clauses).

2-5. Modification 5
Instead of extracting a predetermined number of receiving nodes having a relatively large log likelihood ratio value, the receiving node list creation unit 110 extracts a receiving node having a log likelihood ratio value that is greater than a threshold value. Also good.

2-6. Modification 6
The above-mentioned clause list creation unit 112 extracts a clause including an object directly related to the receiver clause. The extracted clause is a clause or object directly related to the receiver clause. It does not need to be limited to the hanging clause including.

2-7. Modification 7
The above-mentioned clause classifying unit 114 classifies among the clauses extracted in the same category and having the common clauses, but classifies among the clauses extracted in different categories. It is also possible to classify in a clause that does not share the receiving clause.

2-8. Modification 8
A program for realizing each function included in the document processing apparatus 1 may be provided via a recording medium readable by the computer apparatus. Here, the recording medium is, for example, a magnetic recording medium such as a magnetic tape or a magnetic disk, an optical recording medium such as an optical disk, a magneto-optical recording medium, or a semiconductor memory. In addition, this program may be provided via a network such as the Internet.

2-9. Modification 9
The examination viewpoint description document is an example of a rule description document. Here, the rule description document is a document in which the rule is described. More specifically, for example, it is a document in which matters that become the standard of human actions and office procedures are described.

DESCRIPTION OF SYMBOLS 1 ... Document processing apparatus, 101 ... Document input part, 102 ... Document structure analysis part, 103 ... Document storage part, 104 ... Dependency analysis part, 105 ... Dependency structure storage part, 106 ... Receiving part extraction part, 107 ... Examination Perspective description document receiving section storage section 108... Comparison target document receiving section storage section 109... LLR calculating section 110... Receiving section list creating section 111... Receiving section list storing section 112. Related Section List Storage Unit 114. Related Section Classification Unit 115 115 Examination Logic List Creation Unit 116 116 Examination Logic List Storage Unit 117 Output Unit

Claims (7)

A dependency structure of clauses constituting the rule description document, a dependency analysis unit for extracting the dependency structure of clauses constituting the comparison target document,
For each of the dependency clauses constituting the rule description document, based on the dependency structure of the clause extracted from the rule description document by the dependency analysis unit and the dependency structure of the clause extracted from the comparison target document A likelihood ratio calculation unit that calculates a likelihood ratio indicating a degree of occurrence of the dependency clause in the rule description document rather than the comparison target document;
Of the dependency clauses constituting the rule description document, the likelihood ratio value calculated by the likelihood ratio calculation unit is a relatively large number of dependency clauses or calculated by the likelihood ratio calculation unit A document processing apparatus comprising: a dependency clause extracting unit that extracts a dependency clause having a likelihood ratio value larger than a threshold value.   For each of the dependency clauses extracted by the dependency clause extraction unit, a dependency source phrase extraction for extracting the dependency source clause based on the dependency structure of the clause extracted from the rule description document by the dependency analysis unit The document processing apparatus according to claim 1, further comprising a unit.   A relational phrase classification unit that classifies the relational phrase extracted by the relational phrase extraction part for each of the relational phrases extracted by the relational phrase extraction unit based on the meaning of the word included in the relational phrase. The document processing apparatus according to claim 2, further comprising:   The document processing apparatus according to claim 2, wherein the dependency source phrase extracting unit extracts a relationship source phrase representing a condition. A document structure analysis unit that analyzes the structure of the rule description document and the structure of the comparison target document, and identifies another sentence constituting the comparison target document corresponding to the one sentence constituting the rule description document; In addition,
The likelihood ratio calculation unit calculates a likelihood ratio indicating the degree of appearance of the dependency clause more biased to the one sentence than the other sentences for each of the dependency clauses constituting the one sentence. ,
The dependency clause extraction unit includes a predetermined number of dependency clauses having a relatively large likelihood ratio value calculated by the likelihood ratio calculation unit among the dependency clauses constituting the one sentence, or the likelihood. 5. The document processing apparatus according to claim 1, wherein a dependency clause having a likelihood ratio value calculated by the degree ratio calculation unit larger than a threshold value is extracted. A document processing method executed by a document processing apparatus,
A dependency structure of clauses constituting the rule description document, a dependency analysis step of extracting a dependency structure of clauses constituting the comparison target document,
For each dependency clause constituting the rule description document, based on the dependency structure of the clause extracted from the rule description document and the dependency structure of the clause extracted from the comparison target document in the dependency analysis step. A likelihood ratio calculating step of calculating a likelihood ratio indicating a degree of occurrence of the dependency clause in the rule description document rather than the comparison target document;
Among the dependency clauses constituting the rule description document, a predetermined number of dependency clauses having a relatively large likelihood ratio value calculated by the likelihood ratio calculation step or the likelihood ratio calculation step A dependency clause extraction step of extracting a dependency clause whose likelihood ratio value is larger than a threshold value. Computer
A dependency structure of clauses constituting the rule description document, a dependency analysis unit for extracting the dependency structure of clauses constituting the comparison target document,
For each of the dependency clauses constituting the rule description document, based on the dependency structure of the clause extracted from the rule description document by the dependency analysis unit and the dependency structure of the clause extracted from the comparison target document A likelihood ratio calculation unit that calculates a likelihood ratio indicating a degree of occurrence of the dependency clause in the rule description document rather than the comparison target document;
Of the dependency clauses constituting the rule description document, the likelihood ratio value calculated by the likelihood ratio calculation unit is a relatively large number of dependency clauses or calculated by the likelihood ratio calculation unit A program for functioning as a dependency clause extraction unit that extracts a dependency clause whose likelihood ratio is larger than a threshold value.

Images