JPWO2016151838A1 - Prior Research Research System and Prior Research Research Method - Google Patents

Abstract

In the search target term dictionary (21), search target terms that are information that the user wants to know are stored in association with predetermined research fields and categories. When the user designates a research field / category to be searched in the research field / category selection unit (14), the processing apparatus 10 selects a search target term associated with the research field / category from the search target term dictionary (21). Retrieve the text of the document selected from the document DB (31) using the acquired search target term, mark the search target term extracted in the search, Displayed on the display device (50).

Description

  The present invention relates to a prior research research system and a prior research research method suitable for acquiring information that a user wants to know from a literature DB (Database) that accumulates literature.

  When starting a new study, it is essential to first investigate previous studies. In that work, the researcher needs to actually read a large number of documents obtained by keyword search of the document DB, find out the item to be examined, determine its importance, and summarize the results. If all this work is to be done manually, a great deal of labor and time is required.

  For example, Patent Document 1 discloses the accuracy of document search by visualizing the appearance frequency of feature words in the searched document and the distance between the appearance positions of feature words, selecting the visualized feature word, and further searching again. A document search system capable of enhancing the above is disclosed.

Japanese Patent No. 4224131

  However, the document search method disclosed in Patent Document 1 is basically based on a free keyword search, and includes a means of re-search by feature words, but the user knows through the search. The information you want isn't always accurate or directly. Also, depending on the research field, the information that the user wants to know is often the same, but the document search system of Patent Document 1 does not take particular consideration thereof.

  Therefore, in order to extract information that the user wants to know from the retrieved documents, the user needs to set various keywords. Further, for the documents obtained by the search, for example, at least sentences near the keywords. The user's work of obtaining the information they want to know by reading is essential.

  In view of the above-described problems of the prior art, the present invention aims to provide a prior research research system and a prior research research method capable of efficiently acquiring information that a user wants to know from a literature DB of prior research. And

  The prior research research system according to the present invention is communicably connected to a literature DB in which the text of a document and its bibliographic data are stored, and is configured by associating a search target term, which is information that a user wants to know, with a research field and category. A research field / category selection unit for selecting a research field and a category including a search target term to be searched based on a storage device that holds a search target term dictionary, and information operated by a user via an input device, The search target term dictionary is referred to, the search target term associated with the selected research field and category is acquired, and the text of the document selected from the reference DB is searched with the acquired search target term. , Generate a tagged text by attaching a tag to the search term included in the text of the document, and generate the tagged text A tagged text generator that stores the tag in the storage device, and the tagged text is displayed on the display device, and the search target term to which the tag is attached clearly indicates that the tag has been attached. It has the processing apparatus provided with the search object term mark display part which displays a mark, It is characterized by the above-mentioned.

  According to the present invention, it is possible to efficiently acquire information that a user wants to know from a literature DB of prior research.

It is the figure which showed the example of the whole structure of the prior research research system which concerns on embodiment of this invention. It is the figure which showed the example of the structure of the investigation object term dictionary. It is the figure which showed the example of the structure of the search priority table. The figure which showed the example of the search instruction | indication screen displayed on a display apparatus. It is a figure for demonstrating operation when adding a research field and a category to a term object dictionary, (A) is a figure which showed a part required for explanation among search directions screens, and (B) is research. An example of a field addition editing screen, (C) is an example of a category addition editing screen. It is the figure which showed the example of the output instruction | indication screen displayed on a display apparatus. It is the figure which showed the example of the structure of the research result information produced | generated and output by the prior research research system which concerns on embodiment of this invention. The figure which showed the outline | summary of the processing flow of the main process performed in the prior research research system which concerns on embodiment of this invention. The figure which showed the example of the detailed process flow of the research field and category addition process in FIG. It is the figure which showed the example of the method of setting the part to perform a priority search in the text of literature. It is the figure which showed the other example of the method of setting the part to perform a priority search in the text of literature. It is the figure which showed typically the example of the method of extracting the part which is not made into search object from the text of literature.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a diagram showing an example of the overall configuration of a prior research research system 1 according to an embodiment of the present invention. As shown in FIG. 1, the prior research research system 1 is configured by a so-called computer having a processing device 10, a storage device 20, an input device 40, a display device 50, and a communication device (not shown). The prior research research system 1 is communicably connected to a literature DB server 3 that holds a literature DB 31 that stores literatures of prior research via the communication device and the communication network 2.

  The processing device 10 includes a memory such as a ROM (Read Only Memory) and a RAM (Random Access Memory) in addition to a so-called CPU (Central Processing Unit), and the CPU executes a predetermined program stored in the memory. As a result, various functions are realized. In this embodiment, the processing device 10 includes a keyword search unit 11, a feature word extraction / display unit 12, a document text display unit 13, a research field / category selection unit 14, a research field / category addition unit 15, and tagged text. It has functional blocks such as a generation unit 16, a survey target term mark display unit 17, a learning unit 18, and a survey result information output unit 19. Note that the functions realized by these functional blocks will be sequentially described in the following description of the embodiments.

  The storage device 20 includes a memory such as a RAM, a hard disk device, an SSD (Solid State Disk) device, and the like, and stores information necessary for the processing device 10 to execute a predetermined program. In the present embodiment, the storage device 20 includes the technical term dictionary 21, the search reference list 22, the reference text 23, the search target term dictionary 24, the tagged text 25, the teacher text 26, the search priority table 27, and the search result information 28. Etc. are retained.

  The input device 40 includes a keyboard, a mouse, a touch panel, and the like, and is used by a user to input information to the processing device 10. The display device 50 is configured by an LCD (Liquid Crystal Display) or the like, and is mainly used by the processing device 10 to transmit processing results and the like to the user.

  In the present embodiment, the document DB 31 is provided in the document DB server 3, which is a computer different from the computer of the preceding research research system 1, but is incorporated in the computer of the preceding research research system 1. May be. The prior research research system 1 may be composed of a plurality of computers.

  Next, the configurations of the search target term dictionary 24 and the search priority table 27, which are components specific to the present embodiment, will be described.

  FIG. 2 is a diagram showing an example of the configuration of the search target term dictionary 24. As shown in FIG. 2, the survey target term dictionary 24 is a dictionary in which terms (hereinafter referred to as survey target terms) representing information that the user wants to know for each research field is classified into categories and registered, and a so-called thesaurus dictionary is configured. have. Specifically, there is a category of terms often used in the research field at the lower level of the research field, and individual terms are further subordinate to the category.

  In addition, as in the medical field of the example illustrated in FIG. 2, there may be a term that is one level lower than the term, and there may be a term that is deeper than a plurality of levels. Similarly, the category may have a multi-layer structure.

  Therefore, in this embodiment, the user can search a plurality of terms belonging to the specified research field and category at a time by specifying the research field and category by using the search object term dictionary 24. It becomes. For example, when a user investigates the content of a document and specifies the research field as “medicine” and the category as “disease name”, the text of the document reads “malignant tumor”, “colon cancer”, and “breast cancer”. The terms such as are extracted at once.

  It is assumed that such a search target term dictionary 24 is created in advance and stored in the storage device 20. This is because in many research fields, the information that the user wants to know, that is, the terms to be investigated are often the same. Therefore, in the present embodiment, the efficiency of the literature search can be improved by preparing the search target term dictionary 24 in advance.

  However, if the research fields and categories are only predetermined, the flexibility and expandability of the system will be impaired. Therefore, in the present embodiment, it is assumed that the user can define a new research field or category and add it to the search target term dictionary 24 as appropriate. This will be described in detail with reference to the accompanying drawings.

  FIG. 3 is a diagram showing an example of the configuration of the search priority table 27. As shown in FIG. 3, the search priority table 27 includes a place priority, a correlation priority, and a proximity for searching for each category of each research field defined in the search term dictionary 24 of FIG. 2. It is a table in which priorities and weights of these priorities are set.

  Here, the place priority is information representing which place is preferentially searched in the text of a document when searching for terms in the category. Here, when the text is divided into a plurality of parts such as chapters, the place refers to the divided parts. The correlation priority is information representing the priority of a location (a chapter or the like) that is a target of correlation analysis for evaluating the accuracy of a term hit in the search. Further, the proximity priority is information indicating whether a term in the category appears near a term in any other category. Further, the weight is information indicating whether the search is performed with a weight given to a place priority, a correlation priority, a proximity priority, or manual learning when searching for a term.

  Note that values representing these priorities are set in advance for predetermined research fields and categories. However, when the user adds a new research field or category to the research target term dictionary 24, the values indicating these priorities are appropriately set via the display screen of FIG. 5C described later. Can be set.

  FIG. 4 is a diagram showing an example of the search instruction screen 101 displayed on the display device 50 in the prior research research system 1. As shown in FIG. 4, the search instruction screen 101 is provided with a pull-down menu 102 for selecting the document DB 31 to be searched. Therefore, the user selects the document DB 31 to be searched from the list of document DB 31 names displayed in the pull-down menu 102. Next, the user inputs a keyword necessary for narrowing down documents in the text box 103. A plurality of keywords can be input, and sentences may be input.

  Next, when the user clicks the search button 105 on the search instruction screen 101, the processing device 10 (keyword search unit 11) starts a search process using the input keyword for the selected document DB 31. In this search, the user may designate the technical term dictionary 21 used for the search from the pull-down menu 104. Further, the number of documents to be displayed may be set in the text box 126. In this text box 126, a predetermined number may be set in advance.

  When the processing device 10 finishes the above search processing, the processing device 10 displays the titles 109 of the number of documents specified in the text box 126 in the document list display area 107. Further, the processing device 10 (feature word extraction / display unit 12) extracts feature words 111 that frequently appear in the top or all of the retrieved documents, and displays the extracted feature words 111 in the feature word display area 110. To do. Here, the number of feature words 111 displayed in the feature word display area 110 can be designated in advance in the text box 127. Alternatively, the number of feature words may be a predetermined number.

  In the feature word display area 110, the feature word 111 is displayed at a position corresponding to the distance of the relationship between the feature words 111. For example, the feature words 111 that appear in the vicinity of each other in the text of the retrieved document are displayed at positions close to each other and connected by lines. When the user finds the interesting feature word 111 by looking at the arrangement of the feature words 111 and selects the keyword as a keyword, and clicks the search button 112, the user re-searches the document. Can be executed. Furthermore, the user can redisplay the relationship with other feature words 111 when the feature word 111 is centered by selecting the feature word 111 of interest and clicking the related button 113. At that time, the user may change the number in the text box 127 to change the number of feature words to be displayed.

  In the document list display area 107, check boxes 108 are displayed so as to correspond to the respective titles 109 of the displayed documents. Here, a black square indicates that the check has been performed, and a white square indicates that the check has not been performed. The user can freely select a document by appropriately checking the check box 108 corresponding to each document. Therefore, when the user selects a document by checking the check box 108 and clicks the text display button 106, the processing device 10 (the document text display unit 13) puts the text of the selected document in the text display area 114. indicate.

  When the text 122 of the document selected in this way is displayed in the text display area 114, the user selects from the item 116 of the pull-down menu 115 a research field that includes the research field to which the information that the user wants to know, that is, the search target term belongs. For example, when the user selects medicine as a research field, the processing device 10 (research field / category selection unit 14) refers to the search target term dictionary 24 to extract a category name included in the medicine, and a category pull-down menu. As the item 118 of 117, the extracted category name is displayed. At this time, in the item 118, for example, a category name peculiar to medicine such as a disease name and a subject is displayed.

  Next, when the user selects one item from the item 118, for example, a disease name, the processing device 10 (survey target term mark display unit 17) selects a term (colorectal cancer or the like) belonging to the disease name category from the text 122. ) And a mark 123 such as changing the background color of the extracted term.

  As described above, in this embodiment, it is possible to quickly find a search target term that is information that the user wants to know from the displayed text 122 simply by selecting a research field and a category. For example, if the user selects a medical field and a disease name category, a search target term (such as colorectal cancer) corresponding to the disease name is automatically marked in the text 122 of the displayed document. Become. Therefore, the user can easily determine what kind of disease the document is mainly intended for.

  Subsequently, when there is no appropriate research field in the item 116 of the pull-down menu 115 of the research field, or when there is no appropriate category in the item 118 of the pull-down menu 117 of the category, the user operation is as follows. 5 will be described.

  FIG. 5 is a diagram for explaining an operation when adding a research field and a category to the search term dictionary 24, and FIG. 5A is a diagram showing a part necessary for explanation in the search instruction screen 101. (B) is an example of the research field addition edit screen 101b, and (C) is an example of the category addition edit screen 101c.

  When the user sets a research field other than the default research field, the user selects “Add” from the item 116 of the pull-down menu 115 of the search instruction screen 101. Then, the processing apparatus 10 (research field / category addition unit 15) displays a research field addition editing screen 101b (see FIG. 5B) for inputting a user-defined research field name.

  Therefore, when the user inputs a new research field name “user definition 1” to be added to the text box 128 and clicks the add button 129, “user definition 1” is added to the item 116 of the pull-down menu (not shown). ). In addition, a new research field name “user definition 1” is added to the search target term dictionary 24 (see FIG. 2) linked to the item 116 of the pull-down menu 115.

  For example, when the user selects medicine in the item 116 of the pull-down menu 115, if there is no category to which the information (investigation target term) that the user wants to know belongs to the item 118 of the category pull-down menu 117, the same applies. Categories can be added. That is, when the user selects “add” of the item 118 of the pull-down menu 117, the processing apparatus 10 (research field / category addition unit 15) adds a category addition editing screen 101c (FIG. 5 (FIG. 5)). C)) is displayed.

  Therefore, when the user inputs a new category name “user definition 2” to be added to the text box 130 and clicks the add button 131, “user definition 2” is added to the item 118 of the pull-down menu 117 (not shown). ). In addition, a new category name “user definition 2” is added to the search target term dictionary 24 (see FIG. 2) linked to the item 118 of the pull-down menu 117.

  Furthermore, in the category addition edit screen 101c of FIG. 5C, the priority value of the search location in the search priority table 27 for the added category “user definition 2” can be set. That is, when the user selects a chapter name or the like indicating the search location from the item 133 of the pull-down menu 132 and inputs a priority value in the text box 134, the search priority table 27 (see FIG. 3) includes a category “user”. A place priority value for “Definition 2” is set (not shown).

  Similarly, the priority value of the place (chapter etc.) for the term belonging to the added category “user definition 2” can be set. That is, when the user selects a chapter name indicating a location from the item 136 of the pull-down menu 135 and inputs a priority value in the text box 137, the search priority table 27 (see FIG. 3) includes a category “user definition”. The correlation priority value for “2” is set (not shown).

  Similarly, it is possible to set a priority value of proximity to a term belonging to another category of the term belonging to the added category “user definition 2”. That is, when the user selects a category for analyzing proximity from the item 139 of the pull-down menu 138 and inputs a priority value in the text box 140, the category “user” is displayed in the search priority table 27 (see FIG. 3). A proximity priority value for “Definition 2” is set (not shown).

  The correlation analysis and the proximity analysis are performed in order to evaluate the accuracy of the terms when a plurality of terms (search target terms, that is, information that the user wants to know) are extracted by searching the text 122. . For example, by analyzing the correlation between the plurality of extracted terms and the place where the term appears, and selecting a term having a high correlation with the designated location, a more appropriate term can be selected. In addition, proximity here means that it appears in the same paragraph, for example. However, the range of the proximity may be widened or narrowed a little.

  The parameters for determining the search priority are not limited to the search location, correlation, and proximity described above, and other items may be set.

  In addition, the search location, the correlation, the proximity, and the manual learning to be described later are weighted to prioritize the extraction of the terms on the category addition editing screen 101c in FIG. Can do. In that case, the user inputs a weighting value indicating the priority of each item 141 of manual learning, search location, correlation, and proximity into the text box 142.

  As described above, the priority data set on the category addition edit screen 101c is added to the search priority table 27 in FIG. 3 when the user clicks the add button 131.

  Next, manual learning for extracting terms of categories added by the user will be described. First, the user selects a category added by the item 118 of the pull-down menu 117, and marks terms belonging to the category in the text 122 of the text display area 114 using a mouse or the like. Subsequently, when the user presses the save button 119, the marked term is added to the search target term dictionary 24 as a term belonging to the category. At this time, the processing device 10 (learning unit 18) analyzes the contexts before and after, the position in the sentence, the part of speech, and the like, and learns the characteristics of the added category such as the chapter or the place where it appears.

  When learning about the text 122 displayed in the text display area 114 is completed, the processing device 10 (learning unit 18) causes the user to select another document, and the text 122 of the document selected by the user is displayed in the text display area 114. indicate. Therefore, when the user manually marks a term that should belong to the category in the text 122 and presses the save button 119, the processing device 10 (learning unit 18) changes the marked term to the term belonging to the category. To the search term dictionary 24.

  After the user's operation and learning by the processing device 10 are repeated several times, the user calls the text 122 of the new document in the text display area 114 and selects the category added in the item 118 of the pull-down menu 117. Then, the apply button 120 is clicked. Then, the processing device 10 marks the terms belonging to the category in the displayed text 122 using the search priority table 27 and the category characteristics obtained by the learning. At this time, if an appropriate term is not marked, the user can cause the processing device 10 to further learn by marking the appropriate term and clicking the save button 119. Then, the category addition editing screen 101c in FIG. 1C may be displayed, and the priority such as weighting may be reset based on the result obtained by learning as described above.

  Also, for example, when a disease name category is selected, and there is a disease name that is not automatically marked in the displayed text 122, the user manually marks the unmarked term in the text 122. Can do. In this case, when the user further presses the save button 119, the manually-marked disease term is added to the disease name category of the search target term dictionary 24. The search priority of the disease name category is applied to this term.

  When the user clicks the all application button 121 after the category has been selected, the marking of the category selected so far is applied to all the documents listed in the document list display area 107.

  Note that if the survey target term (information you want to know) is not a term such as a disease name but a sentence of conclusion, terms related to statistical processing, for example, numbers such as “significant difference” and “p value” are included in the sentence. You may extract the sentence which is. At that time, the accuracy may be evaluated based on the presence or absence of a context in which the conclusion is affirmed, such as “clarified” or “confirmed”. Moreover, since the figure number figure that appears in the vicinity of the sentence extracted as a conclusion is likely to represent the conclusion, the figure and caption may be extracted.

  The text of the document in the document DB 31 is stored in a markup language format such as xml, for example, and the journal name, author name, publication year, abstract, etc. of the document are marked with tags. Therefore, the processing apparatus 10 can easily extract information marked with the tag from the text of the document, and can store it in a table or the like to create a list of bibliographic information. In addition to these bibliographic information, information that the user wants to know, that is, information specified in a category (survey target term), for example, target disease, number of subjects, measurement method, etc., can be tagged and extracted. That is, when a category is selected in the item 118 of the pull-down menu 117, terms belonging to the category are marked with tags, and marks 123, 124, and 125 are displayed for the corresponding terms in the text 122 in the text display area 114.

  As described above, in the present embodiment, the user can freely add a research item and a category via the search instruction screen 101, and can further set a research object associated with the research item and the category. . Furthermore, the user can freely set priorities for determining priorities such as search places for each research item and category via the category addition editing screen 101c. Therefore, certain expandability and flexibility of the prior research research system 1 according to the present embodiment are ensured.

  FIG. 6 is a diagram showing an example of the output instruction screen 201 displayed on the display device 50 in the prior research research system 1. As shown in FIG. 6, the output instruction screen 201 is provided with a document list display area 206, an output item selection area 211, and an output file format selection area 222. Note that the user can freely switch between the search instruction screen 101 (see FIG. 4) and the output instruction screen 201 by clicking a tab portion of each screen.

  In the document list display area 206, a document title 208 of a document searched by a search instruction on the search instruction screen 101 and a check box 207 corresponding to each document title 208 are displayed.

  Here, when the user clicks the all selection button 202, all of the check boxes 207 are checked, and all the documents are output as search target terms (information that the user wants to know). When the user clicks the upper button 203 and inputs the number of documents in the text box 204, the check box 207 of the document title 208 of the upper document of the input number of documents is checked. When the user clicks the individual selection button 205, the check box 207 can be freely checked. At this time, the user can freely select the document title 208 by clicking the check box 207 freely. It should be noted that after the user selects all items or selects a higher level, the individual document check box 207 may be unchecked or entered using the individual selection button 205.

  In the output item selection area 211, items of information that the user wants to know, such as bibliographic information, are displayed. For example, if a bibliographic information item 217 is displayed and there is an item that the user wants to output, a check box 216 corresponding to that item is checked. At this time, when the user clicks the all selection button 209, all the check boxes 216 are checked, and all the bibliographic information items 217 are output. When the user clicks the individual selection button 210 and clicks the individual check box 216, the bibliographic information item 217 corresponding to the clicked check box 216 becomes the output target.

  Next, the user selects, from among the items 213 in the pull-down menu 212, the research field previously selected in the item 116 of the pull-down menu 115 of the search instruction screen 101 (see FIG. 5). Then, the category to which the term (for example, marks 123, 124, and 125: see FIG. 5) selected by the category pull-down menu 117 on the search instruction screen 101 and marked with the text 122 belongs is displayed in the output item selection area 211. It is displayed as a category item 219. Therefore, the user checks the check box 218 corresponding to the category item 219 to which the investigation target term (information to be known) to be output belongs.

  The user-defined research field is selected from the item 213 in the pull-down menu 212 that is the same as the default research field. Alternatively, the user-defined research field may be separated into another pull-down menu 214, and the user-defined research field item 215 may be selected using the pull-down menu 214. In this case, the user checks the check box 220 corresponding to the user-defined category item 221 and selects the category item 221 to which the investigation target term to be output belongs.

  Next, the user clicks the check box 223 corresponding to the file format name 224 displayed in the output file format selection area 222 to select the file format of the survey result information 28 to be output. Further, the user designates a save destination path in the save destination path designation box 225 and enters a file name in the text box 226. When the user clicks the preview button 227, the survey result information 28 is displayed on the screen, and when the user clicks the output button 228, the survey result information 28 is stored in the storage device 20.

  Note that most of the processing performed by the processing device 10 in accordance with information input from the input device 40 by the user via the output instruction screen 201 described above is included in the processing of the survey result information output unit 19 (see FIG. 1). .

  FIG. 7 is a diagram showing an example of the configuration of the survey result information 28 generated and output by the prior research survey system 1. As shown in FIG. 7, the research field name 301 of the research result information 28 stores the name of the research field that was investigated. In addition, the title column 302 stores the title of the document that is the subject of the survey, and the bibliographic information column 303 stores the magazine name, author, and publication year of the document. The survey result information column 304 stores information such as disease names and subjects obtained from the document.

  Here, the survey result information column 304 is divided into items checked by the check box 218 of the category item 219 in the output item selection area 211 of FIG. 3, and the name of each divided item is a disease name, Corresponds to the category name, such as subject, measurement target. And the information extracted from the said literature according to the category is stored in the investigation result information column 304 corresponding to the item of each category for every literature. Here, the information extracted from the document refers to a term that is marked by the user specifying the category in the text 122 displayed in the text display area 114 of the search instruction screen 101 in FIG.

  Therefore, a plurality of terms or information may be stored in each survey result information column 304. In this case, at the time of file output, the rows or columns of the file may be increased and displayed in parallel. Alternatively, when outputting a file, a screen may be displayed that allows the user to select which term or information to select and display, or to display all.

  As described above, in the present embodiment, if the user selects the research field and category to which the information that the user wants to belong belongs, the information obtained in the keyword search, among the information described in each document, Search result information 28 in which search target terms included in each selected category are associated with the title and bibliographic information of each document can be obtained. Therefore, the user can quickly grasp the main points of the description contents of each document by browsing the survey result information 28. Therefore, according to the present embodiment, it is possible to improve the efficiency of research work of prior research.

  FIG. 8 is a diagram showing an outline of the processing flow of the main processing executed in the prior research research system 1. The processing device 10 of the prior research research system 1 first displays the search instruction screen 101 on the display device 50, and acquires the items that the user selects from the pull-down menu 102, thereby selecting the document DB 31 to be searched (step). S11).

  Next, as a process of the keyword search unit 11, the processing device 10 reads a keyword input in the text box 103, and performs a keyword search of the selected document DB 31 using the keyword (step S12). In this keyword search, the technical term dictionary 21 selected from the pull-down menu 104 is used as appropriate.

  Next, the processing apparatus 10 generates a search document list 22 sorted in order of importance based on the keyword search result, and displays it in the document list display area 107 of the search instruction screen 101 (step S13). Note that the importance here is determined by the frequency of keyword hits.

  Next, the processing device 10 performs the feature word extraction in parallel with the keyword search as the processing of the feature word extraction / display unit 12 and displays the extracted feature words 111 in the feature word display area of the search instruction screen 101. 110 (step S14). Note that the term “feature word” refers to a technical term that frequently appears in the document. In addition, the processing device 10 receives an input in which the user selects one or more feature words 111 from the displayed feature words 111 and clicks the search button 112, that is, in a case where a search is performed again (in step S15). Yes), the process from step S12 is re-executed.

  In addition, when the search button 112 is not clicked and a specific document is checked from the search document list 22 displayed in the document list display area 107 and an input for clicking the text display button 106 is accepted (when the search is not performed again: In step S15, the processing device 10 selects the checked document as the process of the document text display unit 13, and displays the text of the selected document in the text display area 114 of the search instruction screen 101 (step S15). S16).

  Subsequently, the processing device 10 performs a process of dividing the text of the document into chapters and the like (step S17), and further performs a process of excluding quotes and the like that are determined not to contain important information (step S17). Step S18). The specific processing method will be described in detail with reference to the drawings.

  Next, in the processing device 10, as the processing of the research field / category selection unit 14, information (study target term) that the user wants to know is based on operation information for the user to select the items 116 and 118 of the pull-down menus 115 and 117. The research field and category to which it belongs are selected (step S19). However, at this time, if “addition” is selected in the items 116 and 118 (Yes in step S20), the processing device 10 executes a research field / category addition process (step S21). Note that the research field / category addition processing in step S21 will be described in detail with reference to a separate drawing.

  On the other hand, when a predetermined research field and category are selected from the items 116 and 118 in step S19 (when “addition” is not selected: No in step S20), the processing device 10 uses the search priority table. The search priority corresponding to the selected research field and category is acquired from 27, and based on the search priority, the search order or the like is determined for the text of the document selected in step S16 (step S22).

  Next, the processing device 10 associates with the selected research field and category in the search target term dictionary 24 from the document text 23 as processing of the tagged text generation unit 16 and the search target term mark display unit. The retrieved term (information to be known) is searched, a tag is attached to the term hit at this time, and the tagged text 25 is generated and displayed (step S23). At this time, in the text 122 displayed in the text display area 114, a mark 123 is added to a term hit in the search, that is, a term with a tag.

  Next, the processing apparatus 10 displays an output instruction screen 201 as processing of the survey result information output unit 19, reads information input by the user via the output instruction screen 201, and based on the information input by the user The output item is selected (step S24). Subsequently, the processing device 10 outputs the survey result information 28 as shown in FIG. 7 to the storage device 20 or the display device 50 (step S25).

  FIG. 9 is a diagram showing an example of a detailed processing flow of the research field / category addition processing (step S21) in FIG. The processing apparatus 10 reads a new research field and category input via the text box 128 of the research field addition editing screen 101b and the text box 130 of the category addition editing screen 101c, and investigates the new research field and category. A new research field and category are set in the term dictionary 24 (step S31). If a predetermined research field is selected in step S19 (see FIG. 8), the setting of the research field input via the research field additional editing screen 101b is omitted.

  Next, the processing apparatus 10 reads the selection information of the items 133, 136, and 139 of the pull-down menus 132, 135, and 138 on the research field addition editing screen 101b and the input values to the text boxes 134, 137, and 140. Then, the processing device 10 sets priority values for the search location, correlation, and proximity for the new category set via the text box 130 based on the read information, and sets the search priority table. 27 (step S32). At this time, the processing device 10 also sets weights for manual learning, search location, correlation, and proximity.

  Next, the processing apparatus 10 selects the document checked by the check box 108 from the document list displayed in the document list display area 107 of the search instruction screen 101, and the text of the selected document is displayed in the text display area. 114 (step S33). Then, the processing device 10 divides the text of the displayed document into chapters or the like (step S34), and further excludes quotes from the text (step S35).

  Next, the processing apparatus 10 reads the term selected by the user in the text 122 displayed in the text display area 114, and puts a mark such as changing the background color of the term (step S36). Here, it is assumed that the user selects a term suitable for inclusion in the new research field and category set in step S31.

  Next, the processing apparatus 10 obtains information such as in which part (such as chapters) of the marked text the marked term frequently appears and in which feature word it appears, that is, the term Are learned and stored (step S37).

  The processing device 10 repeatedly executes the processing of Step S33 to Step S37 (processing of the learning unit 39) a predetermined number of times (No in Step S38). By repeating this, terms belonging to the research field and category newly set in the search target term dictionary 24 are determined. Further, learning information for setting search priority, correlation, and proximity priority values in the search priority table 27 is obtained. Therefore, the user can change the priority value determined in step S32 to a more appropriate value.

  Next, when the processing unit 10 has executed the processing of the learning unit 39 a predetermined number of times (Yes in step S38), the processing device 10 in the new research field and category of the search priority table 27 determined by the processing so far. Using the priority information, the text search order is determined (step S39). Then, the processing device 10 searches the text of the appropriately selected document with terms belonging to the category set in the search target term dictionary 24, generates tagged text of the searched terms, and displays the text display area 114. (Step S40).

  In the text 122 displayed in the text display area 114, since the tag-added term is marked with a mark 123 or the like, the user determines whether or not an appropriate survey target term is appropriately extracted. be able to. Therefore, when it is determined that the user should repeat the learning (No in step S41), the processing device 10 executes again the processing after step S33 (processing of the learning unit 39).

  FIG. 10 is a diagram illustrating an example of a method for setting a part to be preferentially searched in the text of a document. If the document is an academic article in particular, the text will be abstracted 610, Introduction 620, Materials and Method 630, Results and Discussion, as shown in Figure 10. ), 640, Conclusion 650, Acknowledgment 660, etc. The Abstract (summary) describes the summary of the research, and the Introduction (introduction) describes the previous research related to the research. In the Material and Method (experimental method), a measurement method, a measurement object, and the like are described.

  Therefore, in this embodiment, the text of the literature is divided into chapters such as Abstract 610, Introduction 620, Materials and Method 630, Results and Discussion 640, Conclusion 650, Acknowledgment 660, and the chapter to be preferentially searched is determined for each category of term. The priority value of the priority search for each chapter is stored in the search priority table 27.

  For example, if the terms of the category marked with the mark 601 appear in the order of Abstract 610 and Results and Discussion 640, the priority search priority values are set in that order and stored in the search priority table 27. . Further, if the frequency of terms marked with the marks 602 and 603 appears in the order of Material and Method 630 and Abstract 610, priority search priority values are set in that order and stored in the search priority table 27. Keep it.

  In addition, for example, the introduction 620 describes the previous research of the research paper, and does not describe the content of the research paper. To be judged. Therefore, the Introduction 601 can be excluded from the search target. For chapters excluding search in this way, for example, 0 may be set as the priority value of priority search.

  FIG. 11 is a diagram showing another example of a method for setting a part to be preferentially searched in the text of a document. In this method, as shown in FIG. 11, a plurality of teacher text data 701, 702, 703 in which tags (marks) 705, 706, 707 are added to information (terms belonging to a category prepared in advance) that the user wants to know. Is used. Here, the teacher text data 701, 702, and 703 are data in which all texts of documents previously selected for teacher use are arranged in one line, and are stored in the teacher text 26 of the storage device 20. .

  Here, the processing apparatus 10 uses the teacher text data 701, 702, and 703, and the relationship between the place (chapter and the like) where the term with the tags 705, 706, and 707 appears and the context and part of speech before and after the term. Analyze and learn. When the processing device 10 finishes learning the appearance location for the term tagged for each category, the processing device 10 obtains the frequency of the location where the term belonging to each category appears. Then, based on the frequency and the like, a priority search priority value is determined and stored in the search priority table 27.

  Then, when searching for the search target text 704 with no tag, the processing device 10 obtains the priority of the priority search from the search priority table 27, and prioritizes the high priority part for each category term. Search and put a mark (tag) on the hit term.

  In this example, the number of teacher text data 701, 702, and 703 is three, but the number may be more or less than three. However, the higher the number, the higher the learning accuracy. Through the learning process as described above, the priority accuracy of the priority search stored in the search priority table 27 can be improved.

  FIG. 12 is a diagram schematically illustrating an example of a method for extracting a part to be excluded from a search target from the text of a document. In each of the texts 801 to 804 shown in FIG. 12, the numbers enclosed in [] appearing before the punctuation marks, the superscript numbers, etc., the items up to the punctuation marks are citations of other documents. Is shown. Therefore, the described items are often not the method or conclusion of the document. Therefore, in the present embodiment, such a description portion is excluded from the search target.

  That is, the processing device 10 searches the texts 801, 802, 803, and 804, and when a quotation mark (a number surrounded by [], a superscript number, etc.) is detected before the punctuation mark, Exclude the previous sentence from search. Note that one sentence here means the whole sentence including a punctuation mark when the quotation mark precedes the punctuation mark “.” As in the examples of the texts 802 and 803. In addition, as in the examples of the texts 801 and 804, when the quotation mark is before the punctuation mark “,”, it means the part before the punctuation point in the sentence including the punctuation point.

  Further, as described above, the Introduction 620 portion of each document does not include the contents of the research paper, and can be excluded from the search target. Thus, in this embodiment, the search time by the processing apparatus 10 can be shortened by setting a part not to be searched for the text of each document.

  As described above, according to the present embodiment, it is possible to quickly obtain information that the user wants to know with a simple operation, and thus it is possible to realize an improvement in the efficiency of the research work of the prior research based on the literature research.

  In addition, this invention is not limited to above-described embodiment, Various modifications are included. For example, the above-described embodiment has been described in detail for easy understanding of the present invention, and is not necessarily limited to one having all the configurations described. In addition, a part of the configuration of a certain embodiment can be replaced with the configuration of another embodiment, and the configuration of another embodiment can be added to the configuration of a certain embodiment. In addition, it is possible to add, delete, or replace another configuration for a part of the configuration of each embodiment.

DESCRIPTION OF SYMBOLS 1 Prior research research system 2 Communication network 3 Literature DB server 10 Processing apparatus 11 Keyword search part 12 Feature word extraction display part 13 Reference text display part 14 Research field | category selection part 15 Research field | category addition part 16 Tagged text generation part 17 Search target term mark display unit 18 Learning unit 19 Search result information output unit 20 Storage device 21 Technical term dictionary 22 Search reference list 23 Reference text 24 Search target term dictionary 25 Tagged text 26 Teacher text 27 Search priority table 28 Search result Information 39 Learning unit 40 Input device 50 Display device 101 Search instruction screen 101b Research field addition edit screen 101c Category addition edit screen 201 Output instruction screen

Claims (14)

A communication device communicably connected to a document DB storing the text of the document and its bibliographic data;
A storage device holding a search term dictionary configured by associating a search term that is information that the user wants to know with a research field and category; and
A research field / category selection unit for selecting a research field and a category including a search target term based on information operated by a user via an input device;
Referring to the search term dictionary, obtain the search term associated with the selected research field and category, and search the text of the document selected from the literature DB with the acquired search term A tagged text generation unit that generates a tagged text by tagging the search target term included in the text of the document, and stores the generated tagged text in the storage device;
While displaying the tagged text on a display device, a search target term mark display unit for displaying a mark clearly indicating that the tag has been attached to the search target term,
A prior research research system characterized by having a processing device equipped with The processor is
The text of one or more documents selected from the document DB is searched for the search target terms associated with each of the one or more research fields / categories selected by the research field / category selection unit, and The survey result information output unit for displaying on the display device a list in which the search target terms extracted from the text are associated with the titles and bibliographic data of the documents are further provided. Prior research survey system described. The processor is
Based on the operation information of the user's input device, a new research field and category are set, and terms associated with the new research field and category are selected from the texts of one or more documents selected from the document DB. 2. A research field / category addition unit that selects and adds the selected term to the research term dictionary as a research target term in association with the new research field and category. Prior research survey system described in 1. The storage device
For each research field and category, a search priority table storing search priorities related to the search order of each part when the text of the document is divided into a plurality of parts,
The processor is
When searching for the search target terms belonging to the research field and category in the text of the document, the text of the document is divided into a plurality of parts and is associated with the research field and category from the search priority table. The search order in the text of the document is determined based on the search priority of each part of the text of the document, and the text of the classified document is searched according to the determined search order. Prior research research system described in 1. The processor is
For each research field and category, an input field for setting a search priority of each part of the text of the document is displayed on the display device, and information input by a user through the input field is used. The prior research research system according to claim 4, wherein a search priority of each part of the text of the document is set for each research field and category in the search priority table. The storage device
Tags of one or more documents selected from the document DB are searched with search terms corresponding to one or more research fields and categories selected in advance, and tags are identified in the respective research fields and categories. Further comprising one or more teacher texts that are tagged text generated with
The processor is
Using the one or more teacher texts, for each of the research fields and categories, any part when the search target term associated with each research field and category classifies the teacher text into one or more parts The frequency information of whether or not each of the texts of the document corresponding to the research field and category in the search priority table is set based on the frequency information. Item 5. A prior research research system according to item 4. The processor is
When searching for the text of the document, when a quotation mark is detected immediately before a punctuation mark in the text, a part before the punctuation mark is excluded from a search target in a sentence including the punctuation mark. The prior research research system according to claim 1. It is communicably connected to a document DB that stores the text of the document and its bibliographic data, and includes a processing device, a storage device, an input device, and a display device. A computer that holds a search target term dictionary configured in association with the storage device,
A first step of selecting a research field and a category including a search target term to be searched based on information operated by a user via the input device;
Referring to the search term dictionary, obtain the search term associated with the selected research field and category, and search the text of the document selected from the literature DB with the acquired search term A second step of generating a tagged text by attaching a tag to the search target term included in the text of the document, and storing the generated tagged text in the storage device;
A third step of displaying the tagged text on the display device and displaying a mark clearly indicating that the tag is attached to the search target term to which the tag is attached;
This is a prior research survey method characterized by The computer
The text of one or more documents selected from the document DB is searched for the search target terms associated with each of the one or more research fields / categories selected in the first step, and the texts of the respective documents are searched. 9. The fourth step of displaying on the display device a list formed by associating the extracted search target terms with the titles and bibliographic data of the documents is further executed. Prior research survey method. The computer
Based on the operation information of the user's input device, a new research field and category are set, and terms associated with the new research field and category are selected from the texts of one or more documents selected from the document DB. 9. The fifth step of selecting and adding the selected term to the research term dictionary as a research term in association with the new research field and category is further performed. Prior research survey method described. The storage device includes a search priority table configured by associating search priorities related to the search order of each part when the text of the document is divided into a plurality of parts for each research field and category. Is also retained,
The computer
When searching for the search target terms belonging to the research field and category in the text of the document, the text of the document is divided into a plurality of parts and is associated with the research field and category from the search priority table. The search order in the text of the document is determined based on the search priority of each part of the text of the document, and the text of the classified document is searched according to the determined search order. Prior research investigation method according to 8. The computer
For each research field and category, an input field for setting a search priority of each part of the text of the document is displayed on the display device, and information input by a user through the input field is used. The prior research research method according to claim 11, wherein a search priority of each part of the text of the document is set for each research field and category in the search priority table. In the storage device, texts of one or more documents selected from the document DB are searched for a search target term corresponding to each of one or more research fields and categories selected in advance. One or more teacher texts, which are tagged text generated with tags identified by category, are also retained,
The computer
Using the one or more teacher texts, for each of the research fields and categories, any part when the search target term associated with each research field and category classifies the teacher text into one or more parts The frequency information of whether or not each of the texts of the document corresponding to the research field and category in the search priority table is set based on the frequency information. Item 14. A prior research investigation method according to Item 11. The computer
When searching for the text of the document, when a quotation mark is detected immediately before a punctuation mark in the text, a part before the punctuation mark is excluded from a search target in a sentence including the punctuation mark. The prior research investigation method according to claim 8.

Images