JP2009015394A - Dictionary construction support device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To support the construction of dictionary information with high precision without making it necessary for a user to perform selection operation. <P>SOLUTION: A characteristic expression extraction part 31 of a registration candidate expression generation part 3 extracts characteristic expressions from an analysis object data group stored in an analysis object data storage part 21. A similar expression extraction part 32 extracts similar expressions from the analysis object data group. A expression generation condition setting part 33 sets the generation conditions of expressions to be registered in a dictionary. An expression generation part 34 extracts co-occurrence expressions from the analysis object data group under various expression generation conditions set by the expression generation condition setting part 33, and generates the expressions of the registration candidates in dictionary information. A dictionary registration decision part 51 of a dictionary registration part 5 registers the dictionary information by storing the expressions decided to be expressions to be registered in the dictionary information by the dictionary registration decision part 51 in a dictionary registration candidate generation result obtained by the expression generation part 34 in a dictionary information storage part 22 of a storage part 2 together with a group name associated in the dictionary registration candidate generation result. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a dictionary construction support apparatus that supports construction of dictionary information required for text mining of a document data group, for example.

Conventionally, as an apparatus for supporting the construction of a dictionary for a document analysis system, there has been an apparatus that can register an expression extracted from document data in a dictionary with a simple operation (see, for example, Patent Document 1). In this apparatus, for example, after word data is extracted by morphological analysis and syntax analysis of sentence data, a word list is created by enumerating the extracted words. The dictionary is constructed by allowing the user to select a word to be registered in the dictionary from this word list.
JP 2002-140338 A

  However, the user's visual selection of words is a time-consuming operation due to trial and error. On the other hand, in order to create a dictionary with high classification accuracy, it is not sufficient to simply select and register words from a word list arranged in order of appearance frequency.

  Therefore, an object of the present invention is to provide a dictionary construction support apparatus that can support construction of dictionary information with high accuracy without requiring a selection operation by a user.

  That is, the dictionary construction support apparatus according to the present invention includes a storage unit that stores dictionary information for analyzing document data groups, an input unit that inputs document data groups for construction support of dictionary information, and an input document data group. A first input means for extracting a feature expression in the document and a second extraction means for extracting a similar expression in the input document data group, and the input document based on the extracted feature expression and the similar expression An expression different from the expression included in the data group is generated, and dictionary information including the generated expression is stored in a storage unit.

  According to the present invention, it is possible to support the construction of accurate dictionary information without requiring a selection operation by the user.

Embodiments of the present invention will be described below with reference to the drawings.
(First embodiment)
First, a first embodiment of the present invention will be described.
FIG. 1 is a block diagram showing a configuration example of a dictionary construction support apparatus according to the first embodiment of the present invention.
As shown in FIG. 1, the dictionary construction support apparatus according to the first embodiment of the present invention includes an input device 1 as an input means, a storage device 2 as a storage means, a registration candidate expression generation unit 3, and an expression edit. An expression editing unit 4, which is a means, a dictionary registration unit 5, a display device 6, and a classification processing unit 7 which is a classification processing unit, are connected to each other via a bus 8.

The storage device 2 is a storage medium such as a hard disk drive or a non-volatile memory, and is subject to execution by a registration candidate expression generation unit 3, an expression editing unit 4, a dictionary registration unit 5 that is a storage control unit for dictionary information, and a classification processing unit 7. The control program is stored.
In addition, the storage device 2 includes an analysis target data storage unit 21 that stores a document data group used to construct dictionary information, and a dictionary information storage unit 22 that stores dictionary information. In the present embodiment, the dictionary information is classification dictionary information in which expressions for classification of document data groups are associated with classification item names. The storage device 2 also functions as a work memory for the registration candidate expression generation unit 3, the expression editing unit 4, the dictionary registration unit 5, and the classification processing unit 7.

  The registered candidate expression generation unit 3 includes a feature expression extraction unit 31 that is a first extraction unit that extracts a feature expression from a document data group, and a similar expression extraction unit that is a second extraction unit that extracts a similar expression from the document data group. 32, an expression generation condition setting unit 33 that sets a generation condition of an expression to be registered in dictionary information, and an expression generation unit 34 that is a generation unit that generates an expression to be registered in dictionary information. The input device 1 is, for example, a keyboard or a mouse. The display device 6 is a display device.

The expression editing unit 4 edits the expression extracted by the feature expression extraction unit 31 and the similar expression extraction unit 32 of the registered candidate expression generation unit 3 in accordance with an input operation to the input device 1.
The dictionary registration unit 5 registers the expression generated by the expression generation unit 34 of the registration candidate expression generation unit 3 in the dictionary information. The dictionary registration unit 5 includes a dictionary registration determination unit 51 that determines an expression to be registered in the dictionary information among the expressions generated by the expression generation unit 34.
The classification processing unit 7 classifies document data groups that require classification based on dictionary information stored in the dictionary information storage unit 22 of the storage device 2.

Next, the operation of the dictionary construction support apparatus having the configuration shown in FIG. 1 will be described.
In this embodiment, for the purpose of building dictionary information for questionnaire analysis of a travel agency, questionnaire response document data is assumed as an analysis target data group for building dictionary information. However, the analysis target data is not particularly limited as long as it is text data.
Each analysis target data group includes attribute information of the analysis target text that is the body. In this embodiment, the attribute information includes a distinction between whether the trip to be surveyed is domestic or overseas, a distinction between whether the opinion of the questionnaire is a popular opinion or an unfavorable opinion, and a distinction between use facilities and courses. It is done.

FIG. 2 is a flowchart showing an example of the processing operation of the dictionary construction support apparatus according to the first embodiment of the present invention.
It is assumed that the document data group serving as the analysis target data group is already stored in the storage device 2. This document data group may be newly created by operating the input device 1 and stored in the storage device 2, or may be stored in the storage device 2 from an external device via an input / output interface (not shown).

  First, the registration candidate expression generation unit 3 receives a document data group stored in the storage device 2 when an analysis target data group is specified and dictionary information is constructed by a predetermined input operation to the input device 1. Among these, the specified analysis target data group is read out, and the questionnaire response document data which is each analysis target data in the data group is associated with the information associated with the attribute information included in the response document data as the analysis target data storage unit 21. (Step S1).

FIG. 3 is a diagram showing an example of the analysis target data group stored in the analysis target data storage unit of the dictionary construction support apparatus according to the first embodiment of the present invention in a tabular format.
One row of the analysis target data group shown in FIG. 3 corresponds to one questionnaire response document data. The analysis target data of each row is the analysis target text that is the questionnaire response body, the attribute A that distinguishes whether the survey target travel is domestic or overseas, and whether the opinion of the questionnaire is a positive opinion or a bad opinion The attribute B that distinguishes and the attribute C that distinguishes the use facility and the course are associated with each other.
For example, in the analysis target data corresponding to the first row of the analysis target data group shown in FIG. 3, the attribute A is “domestic”, the attribute B is “favored opinion”, the attribute C is “hotel”, The text to be analyzed is “Hotel meals are good and satisfied”.

Next, the feature expression extraction unit 31 of the registration candidate expression generation unit 3 extracts a feature expression from the analysis target data group stored in the analysis target data storage unit 21 (step S2).
Specifically, the feature expression extraction unit 31 calculates, for example, a word frequency or a product of TF (Term Frequency) and IDF (Inverse Document Frequency) by morphological analysis or syntax analysis, and uses these values (scores) as a reference. A feature expression is extracted from the analysis target data by a known feature word extraction method such as making a characteristic word.
TF is a value obtained by summing up the appearance frequency in each document for all documents. IDF is the logarithm of the total number of documents and the quotient of DF. DF is the number of documents in which an expression appears.

  In addition, when attribute information is given to the analysis target data, the analysis target data is divided into several sets based on this value, and based on values such as mutual information amount, group-specific expressions and characteristics for each group It is also possible to extract expressions. The mutual information amount is an amount that indicates the degree of association between words based on the degree at which two words appearing in a sentence / document appear at the same time. If these two words are a group and a word belonging to the group, the mutual information amount Is also an amount representing the degree to which this word belongs to a group.

  The feature expression extraction unit 31 extracts feature information for each group in accordance with the feature information extraction conditions stored in the storage device 2. Here, the feature expression extraction conditions belonging to the first group are “group division designation”, “expression extraction for each group”, and “attribute B is a popular opinion”. The feature expression extraction conditions belonging to the second group are “group division designation”, “expression extraction for each group”, and “attribute B is unpopular opinion”.

  Based on the analysis target data group having the contents shown in FIG. 3, the feature expression extraction unit 31 shows the feature expression, the score of the feature expression, and the analysis target text from which the feature expression is extracted as shown in FIG. In the table of the analysis target data group, the name of the attribute B corresponding to the attribute group and the temporary group name unique to each attribute name are associated with each other, and the processing data is generated by arranging them in order of score.

FIG. 4 is a diagram showing an example of processing data for feature expression extraction by the registration candidate expression generation unit of the dictionary construction support apparatus according to the first embodiment of the present invention in a table format.
In the processing data shown in FIG. 4, the feature expression belonging to the group name “Group A” has the attribute B corresponding to “analyzed opinion” corresponding to the analysis target text in the table of the analysis target data group shown in FIG. 3. The feature expression that is “satisfied”, “excellent”, “good”, “interesting” and belonging to the group name “group B” is the table of the analysis target data group shown in FIG. And the corresponding attribute B is “unsatisfactory opinion” and “dirty”.

  The feature expression extraction unit 31 generates data in which the description of the temporary group name in the group name item is replaced with the attribute information name description of the extraction condition item in the processing data shown in FIG. It is displayed on the display device 6 as a feature expression extraction result.

FIG. 5 is a diagram showing an example of the feature expression extraction result by the registration candidate expression generation unit of the dictionary construction support apparatus according to the first embodiment of the present invention in a table format.
In the feature expression extraction result shown in FIG. 5, “Group A” in the processing data shown in FIG. 4 is replaced with “favored opinion”, and “Group B” is replaced with “unpopular opinion”. As a result, the feature expression related to the group name “popular opinion” becomes “satisfied”, “excellent”, “good”, “interesting”, and the feature expression related to the group name “unpopular opinion” becomes “dirty”.

  When the input operation related to the expression editing to the input device 1 is performed in a state where the feature expression extraction result shown in FIG. 5 is displayed on the display device 6, the expression editing unit 4 displays the contents displayed on the display device 6. The feature expression extraction result is edited, and the edited result is overwritten and stored in the storage device 2.

Next, the similar expression extraction unit 32 of the registration candidate expression generation unit 3 extracts a similar expression from the analysis target data group stored in the analysis target data storage unit 21 (step S3).
Specifically, the similar expression extraction unit 32 uses a known document classification method such as a method of generating several sets (clusters) of documents whose contents are similar to each other, such as clustering. Divide the data group into several such sets.

The similar expression extraction unit 32 calculates, for each cluster, a score indicating the degree of belonging, which is the degree to which the expression belongs to the cluster, for each word. Based on the scores of these words, the similar expression is a word having a similar meaning for each cluster. Extract the expression.
The similar expression extraction unit 32 associates the similar expression, the score of the similar expression, and the temporary cluster name unique to the cluster to which the similar expression belongs based on the analysis target data group having the contents shown in FIG. Then, processing data is generated in which these are collected for each cluster and arranged in order of score.

FIG. 6 is a diagram showing an example of processing data for extraction of similar expressions by the registration candidate expression generation unit of the dictionary construction support apparatus according to the first embodiment of the present invention in a table format.
In the processing data shown in FIG. 6, the similar expressions belonging to the cluster name “cluster A” are “museum”, “museum”, and “museum” included in the analysis target text of the analysis target data group shown in FIG. Similar expressions belonging to the cluster name “cluster B” are “bath” and “bathroom” included in the analysis target text of the analysis target data group shown in FIG.

  The similar expression extraction unit 32 generates data obtained by replacing the description of the item of the cluster name with the expression having the highest score among the similar expressions belonging to the cluster name in the processing data shown in FIG. The extracted result is displayed on the display device 6.

FIG. 7 is a diagram showing an example of a similar expression extraction result by the registration candidate expression generation unit of the dictionary construction support apparatus according to the first embodiment of the present invention in a table format.
The feature expression extraction result shown in FIG. 7 shows that the cluster name “cluster A” in the processing data shown in FIG. 6 is among the similar expressions “museum”, “museum”, and “museum” belonging to the “cluster A”. The “Museum” with the highest score is replaced, and the cluster name “Cluster B” in the processing data is “Bath” with the highest score among the similar expressions “Bath” and “Bathroom” belonging to the “Cluster B”. Has been replaced. As a result, the similar expressions relating to the cluster name “museum” are “museum”, “art museum”, and “museum”, and the characteristic expressions relating to the cluster name “bath” are “bath” and “bathroom”.

  When the input operation related to the expression editing to the input device 1 is performed in a state where the similar expression extraction result shown in FIG. 7 is displayed on the display device 6, the expression editing unit 4 displays the contents displayed on the display device 6. The similar expression extraction result is edited, and the edited result is overwritten and stored in the storage device 2.

Next, the expression generation condition setting unit 33 of the registration candidate expression generation unit 3 sets expression generation conditions to be registered in the dictionary, and displays them on the display device 6 (step S4).
Specifically, the expression generation condition setting unit 33 sets the feature expression extraction result generated by the feature expression extraction unit 31 in the process of step S <b> 2 as the first generation condition. In addition, the expression generation condition setting unit 33 uses, as the second generation condition, a headline expression and a plurality of synonyms belonging to the headline expression based on the similar expression extraction result generated by the similar expression extraction unit 32 in the process of step S3. Set the word.

FIG. 8 is a table showing an example of expression generation conditions according to the similar expression extraction result by the registration candidate expression generation unit of the dictionary construction support apparatus according to the first embodiment of the present invention.
The expression generation condition shown in FIG. 8 is a similar expression related to “museum” of the similar expression extraction result shown in FIG. As the first synonym A belonging to “Museum”, “Museum” is set as the second synonym B, and “Museum” is set as the third synonym C, In addition, the first synonym belonging to the expression “bath” with the heading “bath” among “bath” and “bathroom” which are similar expressions related to “bath” in the similar expression extraction result shown in FIG. As a synonym A that is a word, “bathroom” is set to a synonym B that is a second synonym.

  The expression generation condition setting unit 33 displays the content displayed on the display device 6 when an input operation related to condition editing on the input device 1 is performed in a state where the generation conditions shown in FIG. The expression generation condition is edited, and the edited result is overwritten and stored in the storage device 2.

  Next, the expression generation unit 34 of the registration candidate expression generation unit 3 extracts expressions and co-occurrence expressions that exist independently from the analysis target data group based on various expression generation conditions set by the expression generation condition setting unit 33. By extracting, the expression of the registration candidate to dictionary information is produced | generated (step S5).

  The method for extracting the co-occurrence expression by the expression generation unit 34 is a method for extracting an expression that co-occurs with the expression in the expression generation condition set in the process of step S4. This method is defined by various existing standards such that co-occurring expressions may be adjacent or separated in morpheme units. The expression generation unit 34 may extract a co-occurrence expression using syntax analysis.

FIG. 9 is a diagram showing an example of a dictionary registration candidate generation result by the registration candidate expression generation unit of the dictionary construction support apparatus according to the first embodiment of the present invention in a table format.
In the dictionary registration candidate generation result shown in FIG. 9, the expression obtained by the expression generation unit 34 as a registration candidate for dictionary information, the TF and DF of the expression, and the group name to which the expression belongs are associated.

In the dictionary registration candidate generation result, the group name related to the co-occurrence expression is expressed by the expression generation unit 34 as various expressions constituting the co-occurrence expression, expressions in various expression generation conditions, feature expression extraction results, and similar expression extraction results. It is generated based on the group name and cluster name to be associated.
In this dictionary registration candidate generation result, the TF and DF related to the co-occurrence expression are expressed by the expression generation unit 34 based on various expression generation conditions, feature expression extraction results, scores of various expressions in similar expression extraction results, and the like. Calculated.

  The expression generation unit 34 uses the feature expressions “satisfied”, “excellent”, “good”, and “dirty” in the characteristic expression extraction result shown in FIG. 5 set as the first generation condition, As shown in FIG. 9, the expressions for candidate registration in the dictionary information are “satisfied”, “meal satisfaction”, “tour price satisfaction”, “excellent scenery”, “meal satisfaction”, “meal is good”, “room is Get "dirty". The group names of these registered candidate expressions are based on “favorite opinions” and “unpopular opinions” that are the group names of the corresponding expressions in the generation conditions of the acquisition source of the expressions.

Among the obtained expressions, “satisfied” in the first row of the table shown in FIG. 9 indicates that the expression generating unit 34 directly uses the feature expression “satisfied” in the first generation condition as a candidate for registration in the dictionary information. The “meal is good” surrounded by the line L1 in FIG. 9 matches the expression of the analysis target text in the analysis target data group. However, “meal satisfaction”, “tour price satisfaction”, “excellent scenery”, “meal satisfaction”, and “room is dirty” do not match the expression of the analysis target text in the analysis target data group. The expression generation unit 34 newly generates a feature expression “satisfied”, “excellent”, “good”, “dirty” in the generation conditions of the text and the expression of the analysis target text in the analysis target data group. .
That is, the expression generation unit 34 can obtain an expression including a subject for a predicate, for example, the feature expression “good” in the first generation condition, by using the first generation condition.

  The storage device 2 stores an expression generation rule such as “the expression generation unit 34 may generate an expression including a particle between“ meal ”and“ satisfaction ”as an expression of a candidate for registration in the dictionary information”. Based on this rule, the expression generation unit 34, for the feature expression “satisfaction” in the first generation condition as shown in FIG. Two or more expressions for “meal” such as “I am satisfied” are generated. In this way, the expression generation unit 34 can generate an appropriate expression variation.

  Here, the expression generation unit 34 extracts an expression having a TF of 10 or more from expressions obtained as candidates for registration in the dictionary information, and uses this as a dictionary registration candidate generation result. That is, the expression generation unit 34 functions as a selection unit that selects an expression that satisfies a predetermined condition from the generated expressions.

  In addition, the expression generation unit 34 sets the “Museum”, “Museum” and “Museum” combinations, and “Bath” and “Bathroom” as synonyms in the second generation condition shown in FIG. By using it, as enclosed by the line L2 in FIG. 9, the expressions for registration candidates in the dictionary information, such as “museum is interesting” and “museum is interesting”, the subject of the same predicate consists of synonyms. A group of expressions related to a bath, such as a group or “bath is wide” and “bathroom is dirty” is generated. The group names of these registered candidate expressions conform to “museum” and “bath”, which are the cluster names of the corresponding expressions in the generation condition of the expression acquisition source.

  This “museum is interesting” is a sentence that does not exist in the analysis target data group. On the other hand, “museum is interesting” is a sentence existing in the analysis target data group. Therefore, the expression generation unit 34 can also obtain an expression that is not in the analysis target data group that is the input data by generating an expression using the synonym condition. By such an expression generation method, there is an effect that it is possible to generate a dictionary registration expression that can be applied to the classification of other document data groups to be classified and has high classification accuracy.

Next, the dictionary registration determination unit 51 of the dictionary registration unit 5 registers the expression in the dictionary based on the dictionary registration criteria stored in the storage device 2 based on the dictionary registration candidate generation result obtained by the expression generation unit 34. It is determined whether or not to perform (step S6).
Specifically, the dictionary registration determination unit 51 uses, as a dictionary registration standard, a product of TF and IDF, which is an index generally indicating the weight of a word, and expresses an expression that takes a certain value or more. It is determined that the expression should be registered in the information. This value may be, for example, a value based on an empirical value such as a product of TF and IDF, or what percentage of a value obtained by summing up the whole word, or a statistical analysis result.

The dictionary registration unit 5 uses, in the dictionary registration candidate generation result, an expression determined by the dictionary registration determination unit 51 as an expression to be registered in the dictionary information among the expressions in the dictionary registration candidate generation result obtained by the expression generation unit 34. The dictionary information is registered by storing it in the dictionary information storage unit 22 of the storage device 2 together with the associated group name (step S7). A serial number is assigned to each expression registered as dictionary information.
FIG. 10 is a table showing an example of the dictionary information registration result by the dictionary registration unit of the dictionary construction support apparatus according to the first embodiment of the present invention.

  Then, the dictionary registration unit 5 performs a dictionary of the storage device 2 in accordance with an input operation related to selection of “dictionary creation support” and “dictionary editing” which are various menus related to verification of registered dictionary information to the input device 1. The dictionary information stored in the information storage unit 22 is displayed on the display device 6 (step S8).

FIG. 11 is a diagram showing a display example of dictionary information by the dictionary construction support apparatus according to the first embodiment of the present invention.
When an icon displayed as “Save As” on this screen is selected by operating the input device 1, the dictionary registration unit 5 displays an input screen for a dictionary name that is an identification name of the displayed dictionary information. Is displayed on the display device 6. The dictionary information of the dictionary name input according to this screen is stored in the dictionary information storage unit 22. This completes the construction of the dictionary information.

  Further, the contents of the dictionary information can be changed by an input operation to the input device 1 on the screen shown in FIG. After the change is completed, when an icon displayed as “overwrite” is selected by operating the input device 1, the dictionary information is overwritten and stored in the dictionary information storage unit 22.

  After constructing the dictionary information, the classification processing unit 7 newly reads out a document data group to be classified, which is different from the analysis target data group used for constructing the dictionary information, from the storage device 2. The document data group to be classified may be newly created by operating the input device 1 and stored in the storage device 2, or may be stored in the storage device 2 from an external device via an input / output interface (not shown). Good.

  Then, the classification processing unit 7 collates one of the read document data groups with the expression in the dictionary information, and if there is an expression that matches the expression in the dictionary information among the expressions in the document data, The group name associated in the dictionary information is acquired, and the collated document data is classified as the document data related to the group name (step S9). The classification processing unit 7 performs this classification processing for each document data group.

  As described above, in the dictionary construction support device according to the first embodiment of the present invention, the registration candidate expression generation unit 3 and the dictionary registration unit 5 allow the user to manually register the dictionary information considering the expression from the beginning. Even without this, this dictionary information can be automatically constructed, and the dictionary can be easily constructed.

  In addition, this dictionary construction support device does not use only the expression described in the sentence of the analysis target data group as it is as the expression of the dictionary information, but the expression scattered in each sentence data of the analysis target data group. Originally, expressions that are not directly described in the text of the analysis target data group can be used as the dictionary information expression. Therefore, after the dictionary information is constructed, another document data group to be classified can be classified with high accuracy. it can.

  The present invention is also applicable to the construction of a dictionary for other document analysis systems such as a document proofreading system. Furthermore, it is effective in reducing the time required for the operation of the document analysis system, such as a dictionary information update operation.

(Second Embodiment)
Next, a second embodiment of the present invention will be described. In addition, the description of the same part as the thing shown in FIG. 1 among the structures of the dictionary construction assistance apparatus concerning this embodiment is abbreviate | omitted.

FIG. 12 is a block diagram showing a configuration example of the dictionary construction support apparatus according to the second embodiment of the present invention.
As shown in FIG. 12, the dictionary construction support apparatus according to the second embodiment of the present invention is compared with the dictionary construction support apparatus according to the first embodiment. A dictionary editing unit 42 as editing means is further provided, and each is connected to the bus 8.

The dictionary verifying unit 41 causes the display device 6 to display a correspondence relationship between the analysis target data group and dictionary information constructed based on the data group.
The dictionary editing unit 42 edits the dictionary information stored in the dictionary information storage unit 22 in accordance with an input operation to the input device 1.

FIG. 13 is a flowchart showing an example of the processing operation of the dictionary construction support apparatus according to the second embodiment of the present invention.
In the second embodiment of the present invention, after the processing from steps S1 to S8 described in the first embodiment is performed, the registered dictionary information for verification of the registered dictionary information to the input device 1 is updated. When various menus related to verification, such as “Dictionary creation support”, “Dictionary verification”, and “Display dictionary verification result display” are selected and an operation of inputting a dictionary name to be verified is performed, the dictionary verification unit 41 Of the dictionary information stored in the dictionary information storage unit 22, the dictionary information corresponding to the dictionary name to be verified is read out, and the analysis target data group used for the construction of the dictionary information is stored as the analysis target data in the storage device 2. Read from the storage unit 21, and based on these, display a dictionary information verification screen including a text display field and a dictionary display field on the display device 6 (step S11).

FIG. 14 is a diagram showing a display example of the dictionary information verification result by the dictionary verification unit of the dictionary construction support apparatus according to the second embodiment of the present invention.
As shown in FIG. 14, in the text display field of the dictionary information verification result screen, the text number of each document data in the analysis target data group and the text, that is, the text, are displayed in association with each other. Sentences displayed in the sentence display field at a time are part of the analysis target data group, but sentences corresponding to sentence numbers that are not displayed should be displayed by operating the scroll bar on the screen by the input device 1. Can do.

  When the user selects one of the sentences on the screen by operating the input device 1, the dictionary verification unit 41 matches the expression in the selected sentence among the expressions of the dictionary information read out in the process of step S11. The expression is displayed in the dictionary display field together with the group name and serial number associated with the dictionary information.

  For example, in FIG. 10, when the sentence “sentence number 1” is selected with a cursor or the like, the dictionary display field has an expression registered in the dictionary information extracted in this sentence, that is, a selection in the sentence display field. The expressions “meal is good” and “satisfied” are displayed underlined in the completed text. Thereby, the user can easily confirm the expression reflected in the constructed dictionary information among the sentences of the analysis target data group.

This dictionary construction support apparatus has a function for a user to add a representation of dictionary information related to a representation in a sentence selected on this screen with reference to the dictionary information verification result screen.
For example, the user confirms the expression “Uncorresponding” that is not underlined in the sentence of “Sentence No. 1”, and registers this in the dictionary information as the expression “Uncorresponding” related to the group name “Unpopular Opinion”. Suppose that a line is added to the dictionary display field on the screen of FIG. 10 and the group name “unpopular opinion” and the expression “corresponding poor” are input by an input operation to the input device 1.

  When the icon indicated as “overwrite” on the screen is selected by operating the input device 1, the dictionary editing unit 42 overwrites and stores the added expression in the dictionary information storage unit 22. Information is edited (step S12).

  And the dictionary verification part 41 reads the dictionary information corresponding to the dictionary name of verification object among the dictionary information memorize | stored in the dictionary information storage part 22 of the memory | storage device 2 similarly to the process of step S11, and the said The analysis target data group used for constructing the dictionary information is read from the analysis target data storage unit 21 of the storage device 2, and based on these, dictionary information verification including a sentence display field and a dictionary display field after editing the dictionary information A screen is displayed on the display device 6 (step S13).

FIG. 15 is a diagram showing a display example of the dictionary information verification result after dictionary information editing by the dictionary editing unit of the dictionary construction support apparatus according to the second embodiment of the present invention.
In the screen shown in FIG. 15, compared with the screen shown in FIG. 14, “Uncorresponding” is newly added in the sentence of the sentence number “1” selected in the sentence field. In response to this, the group name “unpopular opinion”, the number “9”, and the expression “corresponding poor” are added to the dictionary information field. Thereafter, the process proceeds to steps S9 and S10 described in the first embodiment.

  As described above, the dictionary construction support apparatus according to the second embodiment of the present invention provides the user with the dictionary information already constructed and the analysis used for the construction in addition to the effects of the first embodiment. The dictionary can be edited after verifying the correspondence with the text in the target data group, and the edited dictionary information can also be verified.

  The present invention is not limited to the above-described embodiment as it is, and can be embodied by modifying the constituent elements without departing from the scope of the invention in the implementation stage. Various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the embodiment. For example, some components may be omitted from all the components shown in the embodiment. Furthermore, you may combine suitably the component covering different embodiment.

The block diagram which shows the structural example of the dictionary construction assistance apparatus according to the 1st Embodiment of this invention. The flowchart which shows an example of the processing operation of the dictionary construction assistance apparatus according to the first embodiment of the present invention. The figure which shows an example of the analysis object data group memorize | stored in the analysis object data storage part of the dictionary construction assistance apparatus according to the 1st Embodiment of this invention in tabular form. The figure which shows an example of the process data for the feature expression extraction by the registration candidate expression production | generation part of the dictionary construction assistance apparatus according to the 1st Embodiment of this invention in a table format. The figure which shows an example of the feature expression extraction result by the registration candidate expression production | generation part of the dictionary construction assistance apparatus according to the 1st Embodiment of this invention in a table format. The figure which shows an example of the process data for the similar expression extraction by the registration candidate expression production | generation part of the dictionary construction assistance apparatus according to the 1st Embodiment of this invention in a table format. The figure which shows an example of the similar expression extraction result by the registration candidate expression production | generation part of the dictionary construction assistance apparatus according to the 1st Embodiment of this invention in a table format. The figure which shows an example of the expression production | generation conditions according to the similar expression extraction result by the registration candidate expression production | generation part of the dictionary construction assistance apparatus according to the 1st Embodiment of this invention in a table format. The figure which shows an example of the dictionary registration candidate production | generation result by the registration candidate expression production | generation part of the dictionary construction assistance apparatus according to the 1st Embodiment of this invention in a table format. The figure which shows an example of the dictionary information registration result by the dictionary registration part of the dictionary construction assistance apparatus according to the 1st Embodiment of this invention in a table format. The figure which shows the example of a display of the dictionary information by the dictionary construction assistance apparatus according to the 1st Embodiment of this invention. The block diagram which shows the structural example of the dictionary construction assistance apparatus according to the 2nd Embodiment of this invention. The flowchart which shows an example of the processing operation of the dictionary construction assistance apparatus according to the 2nd Embodiment of this invention. The figure which shows the example of a display of the dictionary information verification result by the dictionary verification part of the dictionary construction assistance apparatus according to the 2nd Embodiment of this invention. The figure which shows the example of a display of the dictionary information verification result after dictionary information editing by the dictionary editing part of the dictionary construction assistance apparatus according to the 2nd Embodiment of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Input device, 2 ... Memory | storage device, 3 ... Registration candidate expression production | generation part, 4 ... Expression edit part, 5 ... Dictionary registration part, 6 ... Display apparatus, 7 ... Classification processing part, 8 ... Bus, 21 ... Analysis object data Storage unit 22 ... Dictionary information storage unit 31 ... Feature expression extraction unit 32 ... Similar expression extraction unit 33 ... Expression generation condition setting unit 34 ... Expression generation unit 41 ... Dictionary verification unit 42 ... Dictionary editing unit 51... Dictionary registration determination unit.

Claims (6)

Storage means for storing dictionary information for analysis of a document data group;
Input means for inputting a document data group for constructing the dictionary information;
First extraction means for extracting feature expressions in the input document data group;
Second extraction means for extracting similar expressions in the input document data group;
Generating means for generating a different expression from the expression included in the input document data group based on the feature expression extracted by the first extraction means and the similar expression extracted by the second extraction means;
A dictionary construction support apparatus comprising: storage control means for storing dictionary information including expressions generated by the generation means in the storage means. The dictionary information is information in which a classification item name and an expression in a document data group related to the classification item name are associated with each other,
Each expression of the input document data group is associated with a classification item name related to the expression,
The first extraction unit extracts the feature expression in association with a classification item name associated with the document data group from which the feature expression is extracted,
The generation means associates the generated expression with a classification item name associated with a feature expression that is the basis for generating the expression,
The storage control means stores dictionary information including the different expression and a classification item name associated with the expression in the storage means,
A second input means for inputting a document data group to be classified;
Classification processing for classifying each document data group to be classified as document data related to the classification item name based on the document data group input by the second input unit and the dictionary information stored in the storage unit The dictionary construction support apparatus according to claim 1, further comprising means. Editing input means for receiving editing input of dictionary information stored in the storage means;
2. The dictionary construction support apparatus according to claim 1, further comprising editing means for editing the dictionary information based on contents input by the editing input means. Selecting means for selecting an expression that satisfies a predetermined condition among the expressions generated by the generating means;
The dictionary construction support apparatus according to claim 1, wherein the storage control unit stores dictionary information including the selected expression in the storage unit. Display control means for displaying on the display device the feature expression extracted by the first extraction means and the similar expression extracted by the second extraction means;
Edit input means for accepting edit input of an arbitrary expression among the displayed characteristic expression and similar expression;
2. The dictionary construction support apparatus according to claim 1, further comprising editing means for editing the corresponding expression in accordance with an input by the editing input means. Editing input means for receiving editing input of dictionary information stored in the storage means;
Editing means for editing the dictionary information based on the content input by the editing input means;
Selection input means for receiving selection input of one document data among the document data group;
Selection means for selecting document data from a document data group stored in the storage means in accordance with an input by the selection input means;
Of the expressions in the document data selected by the selecting means, the expression that matches the expression of the dictionary information edited by the editing means is highlighted and displayed on the display device, and the matching expression of the dictionary information is The dictionary construction support apparatus according to claim 1, further comprising display control means for displaying on the display device.

Images