JP5749232B2 - Information retrieval apparatus and operation method thereof - Google Patents

Description

  The present invention relates to an information retrieval apparatus and an operation method thereof.

  Conventionally, a technique for extracting relevance of a plurality of objects from information on the web is known (Non-Patent Document 1).

  According to this technology, a search for “Mr. A”, “Mr. B”, “Mr. A and B” is performed on a plurality of objects on the web, for example, Ms. A and B, and the number of hits (hit (a), hit ( b), hit (a, b)), the strength of the relationship between A and B (hit (a, b) / (hit (a) + hit (b) -hit (a, b)) ( Jaccard coefficient))) can be found. Further, by classifying the pages hit by the search “Mr. A and Mr. B”, the type of the relationship between Mr. A and Mr. B can be extracted. Also, Mr. A and Mr. B can be replaced with an organization or a thing.

  Also, a technique is known that extracts notations indicating “motion”, “target”, “evaluation”, “emotion”, and “location” from blog data as elements constituting human experience, and analyzes their correlation. (Non-patent document 2).

  According to the present technology, it is possible to analyze a relationship such as “many people who saw autumn leaves are impressed”.

  Moreover, it is possible to analyze the correlation between time and human experience by taking advantage of the feature of the blog that describes the date and time of posting.

Keisuke Ishida, Yutaka Matsuo, Yuki Yasuda “Network Making You: Automatic Network Extraction System Using Web Mining”, Artificial Intelligence Society, 21st National Convention (JSAI 2007), June 2007, [August 7, 2012 search ] Internet <URL: http://www.ai-gakkai.or.jp/jsai/conf/2007/data/pdf/100249.pdf> Takeshi Kurashima, Ko Fujimura, Hidenori Okuda “Experience Mining from Large Scale Texts” The Institute of Electronics, Information and Communication Engineers, 19th Data Engineering Workshop (DEWS2008), March 2008, [searched August 7, 2012], Internet <URL: http://www.ieice.org/~de/DEWS/DEWS2008/proceedings/files/a1/a1-4.pdf>

  A conventional information retrieval system retrieves information itself and shows it as it is. Therefore, for example, in keyword search, all the information related to the keyword appears in the result, and the user needs to determine the relevance such as similarity and difference of a lot of information.

  In fact, we know that users will focus on the top information in the search results. For information such as store locations, business hours, and error code handling, Although sufficient results can be obtained with only individual information, what is the individual information in terms of information such as product usage and policy recommendations that reflect the position and opinions of the provider and cannot be determined whether the correctness is correct? It is necessary to grasp from the whole whether it is based on proper positioning and intention, and it is difficult to obtain sufficient results.

  To improve this, there is a method of grasping the position and intention of a smaller number of individuals or organizations describing the information rather than individual information.

  The technology of Non-Patent Document 1 can extract the intent and position as the relationship between the individual or organization describing the information and any object, but it uses the co-occurrence of the object in the information Therefore, it is necessary to widely use information described by other than the target, and the relevance cannot be extracted only from information described by an individual or an organization.

  In addition, the technology of Non-Patent Document 2 can extract the intention or position of the target individual or organization as “behavior” information that is commonly described in blogs or web pages. It is not possible to extract “actions” that are not described. In addition, information other than “behavior” cannot be extracted.

  The present invention has been made in view of the above problems, and the purpose of the present invention is to enable a user of an information search apparatus to determine the relevance such as similarity and difference of curators that are fewer than information. In other words, it is possible to relatively easily determine relevance such as similarity and difference of information.

  A first aspect of the present invention relates to a knowledge notation DB in which knowledge information including knowledge notation used for articles and knowledge values related to the knowledge notation is stored, a scene notation used for articles, and the scene notation. Scene notation DB in which scene information including scene values is stored, emotion notation used in articles and emotion notation DB in which emotion information including emotion values related to the emotion notations are stored, article IDs of articles, articles Article DB that stores article information including curator name and article content, article ID, article curator name, knowledge notation appearing in article content, knowledge value related to knowledge notation, article article A scene DB for storing scene information that includes scene notation appearing in the content, scene value scene value related to the scene notation, emotion notation appearing in the article content of the article, emotion value related to the emotion notation, and seat ID When, For a set of knowledge notation, scene notation and emotion notation in the article content of the article information stored in the article DB, the knowledge value in the knowledge information including the knowledge notation is read from the knowledge notation DB, and from the scene notation DB The scene value in the scene information including the scene notation is read, the emotion value in the emotion information including the emotion notation is read from the emotion notation DB, the article ID of the article information, the knowledge notation, the knowledge value, the scene notation, A notation extraction unit for generating search result information including a scene value, an emotion notation, and an emotion value; and reading an article ID, the knowledge notation, a knowledge value, a scene notation, a scene value, an emotion notation, and an emotion value from the search result information; The curator name in the article information including the article ID is read from the article DB, and the article ID, curator name, knowledge notation, knowledge value, scene notation, scene value, emotion notation, feeling Generates perspectives information including the perspectives ID indicating values and these, and solutions with a information retrieval system, characterized in that it comprises a Perspectives extraction unit to be stored in the Perspective DB.

  According to the second aspect of the present invention, the article ID of the article, the curator name of the article, the knowledge notation appearing in the article content of the article, the knowledge value related to the knowledge notation, the scene notation appearing in the article content of the article, and the scene notation Related scene value, emotion notation appearing in the article content of the article, view DB that stores the viewpoint value including the emotion value and the seat ID related to the emotion notation, and article evaluation information including the evaluation value of the article content of the article Is read from the article evaluation DB and the viewpoint DB, the first viewpoint information group which is all the viewpoint information including one curator name, and all the viewpoint information including one article ID from the viewpoint DB. The second visual information group is read out, and the visual information that does not include the knowledge value that matches the knowledge value of the visual information in the second visual information group is excluded from the first visual information group, and the second visual information group, Within the first visual information group There is seat information including an emotion value that does not match the emotion value of the seat information, or there is seat information including an emotion value that does not match the emotion value of the seat information in the second seat information group in the first seat information group. If the condition is not satisfied, article evaluation information including an evaluation value indicating that the curator with the curator name positively evaluates the article with the article ID is stored in the article evaluation DB, and if the condition is satisfied, the first viewing information First information that does not include an emotion value that does not match the emotion value of the seat information in the second seat information group, the seat information in the second seat information group that includes the emotion value that does not match the emotion value of the seat information in the group. The first visual point that includes an emotion value that does not match the emotion value of the visual point information in the second visual point information group by excluding those including a scene value different from the scene value in the visual point information in the visual point information group from the second visual point information group It is the information on the viewpoint in the information group, Exclude from the first visual information group the ones that contain scene values different from the visual values in the visual information in the second visual information group that do not include the emotion values that do not match the emotion values of the visual information in the one visual information group. There is seat information including an emotion value that does not match the emotion value of the seat information in the first seat information group in the two seat information group, or the emotion of the seat information in the second seat information group in the first seat information group. If the condition that there is the viewpoint information including the emotion value that does not match the value is not satisfied, the article evaluation information including the evaluation value indicating that the curator with the curator name partially evaluates the article with the article ID is positively described. An article evaluation unit that stores article evaluation information including an evaluation value indicating that the curator with the curator name negatively evaluates the article with the article ID if the condition is satisfied and stored in the evaluation DB. To prepare An information search device characterized by the above is used as a solution means.

  The third aspect of the present invention relates to a knowledge notation DB for storing knowledge information including knowledge notation used for articles and knowledge values related to the knowledge notation, and a scene notation used for articles and the scene notation. Scene notation DB in which scene information including scene values is stored, emotion notation used in articles and emotion notation DB in which emotion information including emotion values related to the emotion notations are stored, article IDs of articles, articles Article DB that stores article information including curator name and article content, article ID, article curator name, knowledge notation appearing in article content, knowledge value related to knowledge notation, article article A scene notation for storing the scene notation appearing in the content, the scene value related to the scene notation, the emotion notation appearing in the article content of the article, the emotional value related to the emotion notation and the seat ID, and The above For a set of knowledge notation, scene notation, and emotion notation in the article content of the article information stored in the DB, the knowledge value in the knowledge information including the knowledge notation is read from the knowledge notation DB, and the scene from the scene notation DB Read the scene value in the scene information including the notation, read the emotion value in the emotion information including the emotion notation from the emotion notation DB, the article ID of the article information, the knowledge notation, the knowledge value, the scene notation, the scene value A notation extraction unit for generating search result information including emotion notation and emotion value, and reading out the article ID, the knowledge notation, knowledge value, scene notation, scene value, emotion notation and emotion value from the search result information, and article DB Reads the curator name in the article information including the article ID from the article ID, curator name, knowledge notation, knowledge value, scene notation, scene value, emotion notation, emotion value And a seat group ID indicating the seat information generated by an article by any of one or more curators, and a seat group ID which generates seat information including seat IDs indicating these and stored in the seat DB. A seat group DB that stores the seat group information including the seat ID included in the seat information generated by any one of the articles by the curator and the seat information generated by any one or more curators. It is generated by an article by any one of one or more curators and a representative seat DB that stores representative seat information including the seat group ID indicating information and the seat ID included in the seat information generated by the article by the representative curator. And the curator name of the representative curator, From the representative curator DB storing representative curator information including the curator attribute for the curator, the article evaluation DB for storing the article evaluation information including the evaluation value of the article content of the article, and the viewpoint DB, one curator The first visual information group which is all visual information including names is read, the second visual information information group which is all visual information including one article ID is read from the visual DB, and the first visual information is read from the first visual information group. The viewpoint information that does not include the knowledge value that matches the knowledge value of the viewpoint information in the two viewpoint information group is excluded, and the emotion value that does not match the emotion value of the viewpoint information in the first viewpoint information group is excluded from the second viewpoint information group. If the condition that there is seat information that includes emotion information that does not match the emotion value of the seat information in the second seat information group is not satisfied in the first seat information group, Article evaluation information including an evaluation value indicating that the name curator positively evaluates the article with the article ID is stored in the article evaluation DB, and if the condition is satisfied, the emotion of the seat information in the first seat information group The seat information in the first seat information group that is the seat information in the second seat information group that includes the emotion value that does not match the value, and does not include the emotion value that does not match the emotion value in the seat information in the second seat information group. The visual information in the first visual information group including the emotion value that does not match the emotion value of the visual information in the second visual information group is excluded from the second visual information group. From the first visual information group, the information including the scene value different from the scene value in the visual information in the second visual information group that does not include the emotion value that does not match the emotion value of the visual information in the first visual information group. Excluded, and the second visual information group, the visual information in the first visual information group There is a condition that there is seat information that includes an emotion value that does not match the emotion value, or that there is seat information that includes an emotion value that does not match the emotion value of the seat information in the second seat information group in the first seat information group. If not satisfied, article evaluation information including an evaluation value indicating that the curator with the curator name partially evaluates the article with the article ID is stored in the article evaluation DB. If the condition is satisfied, the curator name An article evaluation unit that stores article evaluation information including an evaluation value indicating that the curator negatively evaluates an article with an article ID in the article evaluation DB, and a viewpoint from the viewpoint DB based on a query input from the outside. Information is retrieved, and for each seat ID in the seat information, the seat group ID in the representative seat information including the seat ID is read from the representative seat DB, and the representative curator DB The representative curator name that is the curator name in the representative curator information including the view group ID and the representative curator attribute that is the curator attribute are read from the article DB, and the article name in the article information that includes the curator name that matches the representative curator name from the article DB The representative curator article name and the representative curator article content that is the article content are read out, and the similar curator name that is the curator name other than the representative curator name in the visual group information including the visual group ID is read from the visual group DB, Read the contrasted article name that is the article name in the article evaluation information including the representative curator name and the evaluation value indicating that the article is negatively evaluated from the article evaluation DB, and the same as the contrasted article name from the article DB Contrast with curator name in article information including article name And a curation article including the query, representative curator name, representative curator attribute, representative curator article name, representative curator article content, similar curator name, contrast article name, and contrast curator name. An information retrieval apparatus including a curated article display unit is used as a solving means.

  The fourth aspect of the present invention is an operation method of an information search apparatus, wherein the information search apparatus stores knowledge notation used for articles and knowledge information including knowledge values related to the knowledge notation. A scene notation DB for storing scene notation used for articles and scene information including scene values related to the scene notations, and an emotion notation used for articles and emotions including emotion values related to the emotion notations Appears in the emotion notation DB that stores information, the article DB that stores article information including article ID, article curator name, and article content, and article ID, article curator name, and article article content. Knowledge notation, knowledge value related to the knowledge notation, scene notation appearing in the article content of the article, scene value related to the scene notation, emotion notation appearing in the article content of the article, emotion value related to the emotion notation And perspective A visual field DB for storing visual field information including D, and in the operation method, the notation extraction unit of the information search device includes knowledge notation and scene notation in the article content of the article information stored in the article DB. And a knowledge value in the knowledge information including the knowledge notation from the knowledge notation DB, a scene value in the scene information including the scene notation from the scene notation DB, and from the emotion notation DB. Reading the emotion value in the emotion information including the emotion notation, and generating search result information including the article ID of the article information, the knowledge notation, the knowledge value, the scene notation, the scene value, the emotion notation, and the emotion value; The viewpoint extraction unit of the information search device reads the article ID, the knowledge expression, the knowledge value, the scene expression, the scene value, the emotion expression, and the emotion value from the search result information, from the article DB. Read the curator name in the article information including the article ID, and generate the seat information including the article ID, curator name, knowledge notation, knowledge value, scene notation, scene value, emotion notation, emotion value and the seat ID indicating these And a step of storing the data in the visual field DB.

  The fifth aspect of the present invention is an operation method of an information search apparatus, wherein the information search apparatus is an article ID of an article, an article curator name, a knowledge notation appearing in the article content of the article, and knowledge related to the knowledge notation. Information, scene notation appearing in the article content of the article, scene value related to the scene notation, emotion notation appearing in the article content of the article, emotion value related to the emotion notation and seat information including the seat ID are accumulated A viewpoint DB, and an article evaluation DB for storing article evaluation information including an evaluation value of the article content of the article. In the operation method, the article evaluation unit of the information search device is configured to receive one item from the viewpoint DB. Read the first visual information group which is all visual information including curator names, read the second visual information group which is all visual information including one article ID from the visual DB, from the first visual information group, Second view A step of excluding visual information that does not include a knowledge value that matches the knowledge value of visual information in the information group, and the article evaluation unit includes the emotion value of the visual information in the first visual information group in the second visual information group If there is seat information that includes an emotion value that does not match, or there is seat information that includes an emotion value that does not match the emotion value of the seat information in the second seat information group in the first seat information group. The article evaluation information including an evaluation value indicating that the curator with the curator name positively evaluates the article with the article ID is stored in the article evaluation DB, and if the condition is satisfied, the viewing information in the first viewing information group In the first visual information group that does not include an emotion value that does not match the emotion value of the visual information in the second visual information group, A place different from the scene value in the viewpoint information Information including a value is excluded from the second visual information group, and the visual information in the first visual information group includes an emotion value that does not match the emotion value of the visual information in the second visual information group. The second visual information is excluded from the first visual information group by including the scene value different from the scene value in the visual information in the second visual information group that does not include the emotion value that does not match the emotion value of the visual information in the group. The group has visual information that includes an emotion value that does not match the emotion value of the visual information in the first visual information group, or the first visual information group matches the emotion value of the visual information in the second visual information group If the condition that there is seat information including the emotion value not to be satisfied is not satisfied, the article evaluation information including the evaluation value indicating that the curator of the curator name partially evaluates the article of the article ID in the article evaluation DB. Memorize and curator if condition is met And storing the article evaluation information including an evaluation value indicating that the name curator negatively evaluates the article with the article ID in the article evaluation DB. And

  A sixth aspect of the present invention is an operation method of an information search apparatus, wherein the information search apparatus stores knowledge notation used for articles and knowledge information including knowledge values related to the knowledge notation. A scene notation DB for storing scene notation used for articles and scene information including scene values related to the scene notations, and an emotion notation used for articles and emotions including emotion values related to the emotion notations Appears in the emotion notation DB that stores information, the article DB that stores article information including article ID, article curator name, and article content, and article ID, article curator name, and article article content. Knowledge notation, knowledge value related to the knowledge notation, scene notation appearing in the article content of the article, scene value related to the scene notation, emotion notation appearing in the article content of the article, emotion value related to the emotion notation And perspective The knowledge DB includes the knowledge DB for storing the knowledge DB, the scene notation, and the emotion notation in the article content of the article information stored in the article DB. Reading the knowledge value in the knowledge information, reading the scene value in the scene information including the scene notation from the scene notation DB, reading the emotion value in the emotion information including the emotion notation from the emotion notation DB, and the article information A notation extraction unit for generating search result information including the article ID, knowledge notation, knowledge value, scene notation, scene value, emotion notation, and emotion value, and article ID, knowledge notation, knowledge value, Read the scene notation, scene value, emotion notation, and emotion value, read the curator name in the article information including the article ID from the article DB, and read the article ID, curator name, knowledge According to an article by any one of the viewpoint extraction unit that generates notation, knowledge value, scene notation, scene value, emotion notation, emotion value, and the viewpoint ID including the viewpoint ID indicating these and stores them in the viewpoint DB and one or more curators A seat group ID including a seat group ID including the seat ID included in the seat information generated by the curator name of any of the curators and the seat information generated by any of the articles by the curator. Representative seat information including DB, seat group ID indicating seat information generated by an article by one or more curators and seat ID included in seat information generated by an article by the representative curator is stored. Visual field showing visual field information generated by an article by either the representative visual field DB or one or more curators A representative curator DB that stores representative curator information including the curator group ID, the curator name of the representative curator, and the curator attribute for the curator, and an article for storing article evaluation information including an evaluation value of the article content of the article A first visual information group that is all visual information including one curator name is read from the evaluation DB and the visual DB, and second visual information that is all visual information including one article ID is read from the visual DB. The group is read out, and the seat information that does not include the knowledge value that matches the knowledge value of the seat information in the second seat information group is excluded from the first seat information group, and the second seat information group includes the first seat information group. There is seat information including an emotion value that does not match the emotion value of the seat information of the first seat, or the emotion information value that does not match the emotion value of the seat information in the second seat information group is included in the first seat information group If the condition that there is seat information is not satisfied, article evaluation information including an evaluation value indicating that the curator with the curator name positively evaluates the article with the article ID is stored in the article evaluation DB, and if the condition is satisfied Emotion value that is the seat information in the second seat information group including the emotion value that does not match the emotion value of the seat information in the first seat information group and does not match the emotion value in the seat information in the second seat information group Emotion values that do not match the emotion value of the visual information in the second visual information group are excluded from the second visual information group that include scene values different from the scene values in the visual information in the first visual information information group Different from the scene value in the seat information in the second seat information group that does not include the emotion value that does not match the emotion value of the seat information in the first seat information group. Is the first viewpoint information group including the scene value? Exclude and the second visual information group includes visual information that includes an emotion value that does not match the emotion value of the visual information in the first visual information group, or the first visual information group includes the visual information in the second visual information group Article evaluation including an evaluation value indicating that the curator with the curator name partially evaluates the article with the article ID if the condition that there is the seat information including the emotion value that does not match the emotion value of the seat information is not satisfied Information stored in the article evaluation DB, and if the condition is satisfied, the article evaluation information including an evaluation value indicating that the curator with the curator name negatively evaluates the article with the article ID is stored in the article evaluation DB. An evaluation unit, wherein the selection unit of the information search device searches for visual point information from the visual point DB based on a query input from the outside, and for the visual point ID in the visual point information, The visual field group ID in the representative visual field information including the visual field ID is read from the representative visual field DB, and the representative curator name and the curator attribute that is the curator name in the representative curator information including the visual field group ID are read from the representative curator DB. The curator attribute is read, the representative curator article name that is the article name in the article information including the curator name that matches the representative curator name and the representative curator article content that is the article content are read from the article DB, and Article evaluation including an evaluation value that reads a similar curator name that is a curator name other than the representative curator name in the viewpoint group information including the group ID and negatively evaluates the representative curator name and the article from the article evaluation DB Contrasting note that is the article name in the information Reading the name, reading the contrast curator name that is the curator name in the article information including the same article name as the contrast article name from the article DB, and the curation article display unit of the information search device includes the query And displaying a curation article including a representative curator name, a representative curator attribute, a representative curator article name, a representative curator article content, a similar curator name, a contrasting article name, and a contrasting curator name. The operation method of the search device is used as a solving means.

  According to the present invention, a user of an information search apparatus can determine the relevance such as similarity or difference of curators, which is smaller than information, and can easily determine the relevance of information similarity or difference. can do.

It is a figure which shows an example of a structure of the information search device which concerns on this Embodiment. It is a figure which shows an example of a structure of knowledge notation DB7. It is a figure which shows an example of a structure of scene description DB8. It is a figure which shows an example of a structure of emotion description DB9. It is a figure which shows an example of an article. It is a figure which shows an example of a structure of article DB4. It is a figure which shows an example of a structure of curator DB3. It is a figure which shows an example of a structure of view DB5. It is a figure which shows a mode that the value Xmig which shows the grade like article content is a group based on all the article contents and all the knowledge values using LDA. It is a figure which shows a mode that the value Yig which shows the probability that a knowledge value will appear in a group is calculated | required based on all the article contents and all the knowledge values using LDA. It is a figure which shows a mode that the value Xmjg which shows the grade that the article content seems to be a group is obtained based on all the article contents and all emotion notation using LDA. It is a figure which shows a mode that the value Yjg which shows the probability that an emotion notation appears in a group will be calculated | required based on all the article contents and all the emotion notations using LDA. It is a figure which shows an example of a structure of viewpoint group DB11. It is a figure which shows an example of a structure of representative visual point DB12. It is a figure which shows an example of a structure of representative curator DB13. 5 is a flowchart showing a flow of operations of an article evaluation unit 14. It is a figure which shows a mode that article evaluation information is produced | generated based on information P1, P2 and article content using LDA, and is memorize | stored in article evaluation DB15. It is a figure which shows an example of a structure of article evaluation DB15. 7 is a flowchart showing a flow of operations of a selection unit 16 and a curation article display unit 17. It is a figure which shows the example of a display of a curation article. It is a figure which shows another example of a display of curation article.

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

  FIG. 1 is a diagram illustrating an example of the configuration of the information search apparatus according to the present embodiment.

  An information search device according to the present embodiment (hereinafter simply referred to as an information search device) includes an article collection unit 1, a viewpoint extraction unit 2, a curator database (hereinafter abbreviated as DB) 3, an article DB4, a viewpoint DB5, Notation extraction unit 6, knowledge notation DB7, scene notation DB8, emotion notation DB9, relationship analysis unit 10, view group DB11, representative view DB12, representative curator DB13, article evaluation unit 14, article evaluation DB15, selection unit 16, and curation article A display unit 17 is provided.

  The information retrieval apparatus includes information (hereinafter referred to as an article) including the article content written by the curator based on his / her viewpoint and the curator's own information as meta information (hereinafter referred to as an article), and a keyword (for example, “ "Yokosuka Chuo") is input from the outside. The information search apparatus displays a curation article that is an article presented based on a viewpoint that represents the viewpoint group.

  FIG. 2 is a diagram illustrating an example of the configuration of the knowledge notation DB 7.

  The knowledge notation DB 7 stores in advance knowledge information including knowledge notation (morphological information of nouns, verbs, adjectives, adverbs) used for articles posted on blogs, web pages, and the like and knowledge values related to the knowledge notation. ing. The knowledge value may be, for example, a concept level, a combination of notations, a context, or a relationship of a plurality of notations (notation similar information, superordinate concepts, subordinate concepts, inclusion relations, etc.). For example, knowledge “battleship” as a superordinate concept may be defined for the knowledge notation “Mikasa”.

  FIG. 3 is a diagram illustrating an example of the configuration of the scene notation DB 8.

  In the scene notation DB 8, scene information including scene notation (time and place) used for articles and scene values related to the scene notation is stored in advance. For example, the scene value “= # article reference date” is defined for the scene notation “today”, and the scene value “sunny” is defined for the scene notation “sunny”. In order to express the scene value at the concept level, the article itself, the semantic paragraph itself, and plural notation relationships (notation similar information and proximity information, superordinate concepts, subordinate concepts, inclusion relations, etc.) may be defined. For example, as an indication that an article itself can be processed as a scene, a scene value “= # article” or the like may be defined for the scene notation “#article”.

  FIG. 4 is a diagram illustrating an example of the configuration of the emotion notation DB 9.

  The emotion notation DB 9 stores emotion information including emotion notation (emotion word) used for articles and emotion values related to the emotion notation in advance. For example, an emotion value “joy” and an emotion value “anticipation” are defined for the emotion notation “exciting”. For example, an emotion value “joy” or the like is defined for the emotion notation “feeling rising to the sky”. The emotion value may define a combination of emotion notation and a context.

(Operation of article collection unit 1)
Returning to FIG. 1, the article collection unit 1 receives articles described from individuals and organizations (collectively referred to as curators) from the outside.

  FIG. 5 is a diagram illustrating an example of an article.

  The article includes, for example, the blog title “B's blog”, curator attribute (profile) “military critic (naval)”, article name (article title) “Mikasa's brave figure”, article content (article text) “Today, Mikasa I went to the park ... blessed with fine weather ... excited ... ", including the article reference date" xxxx year x month xx day ". For example, the blog title “Mr. B's blog” includes the curator name “Mr. B”.

  When the article collection unit 1 receives an article, it reads the curator name, article name, article content, etc. of the curator from the article, generates article information, and stores it in the article DB 4.

  FIG. 6 is a diagram illustrating an example of the configuration of the article DB 4.

  The article information includes article identification information indicating the corresponding article (hereinafter, identification information is referred to as ID), curator name (such as “Mr. B” in the figure), article name (such as “Mikasa's brave figure” in the figure), article content and Includes the article reference date.

  Returning to FIG. 1, if the curator name of the received article is a new curator name, the article collection unit 1 generates curator information and stores it in the curator DB 3.

  FIG. 7 is a diagram illustrating an example of the configuration of the curator DB 3.

  The curator information includes a curator name (such as “Mr. B” in the figure) and a curator attribute (such as “military critic (navy)” in the figure).

(Operation of the visual point extraction unit 2)
Returning to FIG. 1, for example, when article information is newly stored in the article DB 4, the viewpoint extraction unit 2 reads the article information from the article DB 4 and transmits it to the notation extraction unit 6.

  The notation extraction unit 6 receives the article information, reads the article ID from the article information, and includes the knowledge notation from the knowledge notation DB 7 for each of the knowledge notation, scene notation, and emotion notation in the article content of the article information. The knowledge value in the knowledge information is read, the scene value in the scene information including the scene notation is read from the scene notation DB 8, and the emotion value in the emotion information including the emotion notation is read from the emotion notation DB 9.

  The notation extraction unit 6 generates search result information including the read article ID, knowledge notation, knowledge value, scene notation, scene value, emotion notation, and emotion value for each set of knowledge notation, scene notation, and emotion notation. Each search result information is transmitted to the visual point extraction unit 2.

  The viewpoint extraction unit 2 reads the article ID, knowledge notation, knowledge value, scene notation, scene value, emotion notation, and emotion value from each search result information, and reads the curator name in the article information including the article ID from the article DB4. Perspective information including the article ID, curator name, knowledge notation, knowledge value, scene notation, scene value, emotion notation, emotion value and the seat ID indicating these is generated and stored in the view DB 5.

  FIG. 8 is a diagram illustrating an example of the configuration of the visual point DB 5.

  Perspective information includes article ID (such as “36” in the figure), curator name (such as “Mr. B” in the figure), knowledge notation (such as “Mikasa” in the figure), knowledge value (such as “Mikasa” in the figure), scene Notation (such as “article“ Mikasa's brave figure ”), scene values (such as“ article “Mikasa's brave figure”), emotion notation (such as “exciting” in the figure), emotion value (“joy” in the figure) ")") And sight ID. In the seat information, a value Im [%] and a value Jm [%] are stored later.

  In calculating the values Im and Jm, the viewpoint extraction unit 2 reads all curator names from the curator DB 3 and divides them into, for example, three groups. The number of groups is set in advance and may be a different number.

  As shown in FIG. 9, the viewpoint extraction unit 2 reads all article contents from the article DB 4 and reads all knowledge values from the viewpoint DB 5 by using a technique called LDA (Lent Dirichlet Allocation). For example, a value Xmig “80%” indicating the degree of the article content that seems to be a group is obtained for the set of the article content of number 1 and the group of number 1. The viewpoint extraction unit 2 obtains the value Xmig for the other sets.

  As shown in FIG. 10, the viewpoint extraction unit 2 uses the LDA, based on all the article contents and all the knowledge values, for example, the knowledge value for the group of the knowledge value “Mikasa” and the number 1 group. A value Yig “30%” indicating the probability that “Mikasa” appears in the group is obtained. The viewpoint extraction unit 2 obtains the value Yig for the other sets.

  As shown in FIG. 11, the viewpoint extraction unit 2 reads all emotion expressions from the viewpoint DB 5 by using LDA, and, for example, the article contents and numbers of number 1 are based on all article contents and all emotion expressions. For a group of one group, a value Xmjg “50%” indicating the degree of article content that seems to be a group is obtained. The viewpoint extraction unit 2 obtains the value Xmjg for the other sets.

  As shown in FIG. 12, the viewpoint extraction unit 2 uses the LDA, based on all the article contents and all the emotion expressions, for example, for the group of the emotion expression “exciting” and number 1 A value Yjg indicating the probability that the notation “exciting” appears in the group is obtained. The viewpoint extraction unit 2 obtains the value Yjg for the other sets.

  The viewpoint extraction unit 2 obtains the value Xmig (for example, 80% in FIG. 9), the knowledge value, and the group obtained for the article content and the group for one article content, one knowledge value, and one group. A product Zmig (24%) with the value Yig (for example, 30% in FIG. 10) is obtained.

  The viewpoint extraction unit 2 obtains the product Zmig for other sets having the same article content and knowledge value but different groups.

  The visual point extraction unit 2 calculates a value Im of the sum of products Zmig. That is, the viewpoint extraction unit 2 obtains a value Im for a set of one article content and one knowledge value.

  The viewpoint extraction unit 2 obtains a value Im for another set in which at least one of the article content and the knowledge value is different.

  The viewpoint extraction unit 2 includes each value Im in the viewpoint information including the knowledge value corresponding to the value Im and including the article ID of the article content corresponding to the value Im.

  The viewpoint extraction unit 2 calculates the value Xmjg (for example, 50% in FIG. 11) obtained for the article content and the group, and the emotion expression and the group for one article content, one emotion expression, and one group. A product Zmjg (20%) with the value Yjg (for example, 40% in FIG. 12) is obtained.

  The viewpoint extraction unit 2 obtains the product Zmjg for other pairs having the same article content and emotion notation but different groups.

  The viewpoint extraction unit 2 obtains a sum value Jm of products Zmjg. That is, the viewpoint extraction unit 2 obtains a value Jm for a set of one article content and one emotion notation.

  The viewpoint extraction unit 2 obtains the value Jm for other sets in which at least one of the article content and the emotion notation is different.

  The viewpoint extraction unit 2 includes each value Jm in the viewpoint information including the emotion ID corresponding to the value Jm and including the article ID of the article content corresponding to the value Jm.

(Operation of the relationship analysis unit 10)
For example, if a large number and various types of viewpoint information are accumulated in the viewpoint DB 5, the relationship analysis unit 10 has different curator names from the viewpoint DB 5, the knowledge values match, the scene values match, and the emotion values match. Are searched for a set of two pieces of visual point information (hereinafter referred to as visual point information A and B) that satisfy the condition that they match. For example, the curator name of the visual point information A is “Mr. A”, and the curator name of the visual point information B is “Mr. B”.

  The relationship analysis unit 10 generates a seat group ID if there is corresponding seat information A and B.

  The relationship analysis unit 10 searches the visual field information A for other visual field information A including the curator name that matches the curator name in the visual field information A and the scene value that matches the scene value in the visual field information A.

  The relationship analysis unit 10 searches the visual field information B for other visual field information B including a curator name that matches the curator name in the visual field information B and a scene value that matches the scene value in the visual field information B.

  The relationship analysis unit 10 searches the visual field information A for other visual field information A including the curator name that matches the curator name in the visual field information A and the knowledge value that matches the knowledge value in the visual field information A.

  The relationship analysis unit 10 searches the visual field information B for other visual field information B including the curator name that matches the curator name in the visual field information B and the knowledge value that matches the knowledge value in the visual field information B.

  If the knowledge value and the emotion value match for each set of one piece of seat information A and one piece of seat information B, the relationship analysis unit 10 sets the generated seat group ID, the curator name and the seat ID in the seat information A. The viewpoint group information including is generated and stored in the viewpoint group DB 11.

  When the relationship analysis unit 10 generates the sight group information, the relationship analysis unit 10 excludes the combination of the sight information A and B, and performs the same process for other sets in which at least one of the sight information A and B is different.

  The relation analysis unit 10 is different from the pair of the visual point information A and B in that at least one curator name is different and includes the knowledge value, the scene value, and the emotion value that match in the visual point information A and B described above. The same processing is performed for the set of viewpoint information. For example, one curator name of another set of viewpoint information is “Mr. A”, and the other curator information curator name is “Mr. C”. In this case, the seat group ID is the same as the seat group ID generated in the case of the seat information A and B.

  Further, the relationship analysis unit 10 matches the knowledge values and the scene values even when at least one of the knowledge value, the scene value, and the emotion value is different from the set of the viewpoint information A and B. If there is a set of perspective information whose emotion values match, the same processing is performed. In this case, another viewpoint group ID is used.

  FIG. 13 is a diagram illustrating an example of the configuration of the seat group DB 11.

  The viewpoint group information includes a viewpoint group ID (such as “11” in the figure), a curator name (such as “Mr. B” in the figure), and a viewpoint ID (such as “15” in the figure).

  That is, the viewpoint group information includes the viewpoint group ID indicating the viewpoint information generated by the article by any one or more curators (for example, a plurality of pieces of viewpoint information by “Mr. B”, “Mr. O”, etc.) The curator's curator name (for example, “Mr. B”) and the seat ID included in the seat information generated by any of the articles by the curator are included.

  Next, the relationship analysis unit 10 groups all the viewpoint group information according to the viewpoint group ID. When there is only one type of visual group ID in the visual group DB 11, the number of groups is 1.

  The relationship analysis unit 10 selects, for each group, the most frequent curator name in the group, reads the visual group ID and visual field ID for each visual group information including the curator name in the group, and representative visual information including these Is stored in the representative visual point DB 12.

  FIG. 14 is a diagram illustrating an example of the configuration of the representative visual point DB 12.

  The representative seat information includes a seat group ID (such as “11” in the figure) and a seat ID (such as “15” in the figure).

  In other words, the representative visual point information includes a visual point group ID indicating visual point information generated by an article by one or more curators and a visual point ID included in the visual point information generated by an article by the representative curator.

  Next, for each group, the relationship analysis unit 10 reads the curator attribute in the curator information including the same curator name as the most frequent curator name in the group from the curator DB 3, and the same curator name as the most frequent curator name from the visual group DB 11. The visual group ID in the visual group information including the curator name is read, representative curator information including the visual group ID, curator name, and curator attribute is generated and stored in the representative curator DB 13.

  FIG. 15 is a diagram illustrating an example of the configuration of the representative curator DB 13.

  The representative curator information includes a viewpoint group ID (such as “11” in the figure), a curator name (such as “Mr. B” in the figure), and a curator attribute (such as “military critic (navy)” in the figure).

That is, the representative curator information includes a seat group ID indicating seat information generated by an article by any one or more curators, a curator name of the representative curator, and a curator attribute for the curator. If a large number of various pieces of viewpoint information are newly accumulated in the viewpoint DB 5, the data stored in the viewpoint group DB 11, the representative viewpoint DB 12, and the representative curator DB 13 may be cleared and the process may be performed again.

(Modification)
If the knowledge value and the emotion value match for each set of one piece of seat information A and one piece of seat information B, the relationship analysis unit 10 generates the seat group ID generated, the curator name in the seat information A, and the seat. Although the visual group information including the ID is generated and stored in the visual group DB 11, it may be as follows.

  The relationship analysis unit 10 has the same knowledge value and emotion value for a set of one piece of seat information A and one piece of seat information B, and the value Im in the seat information A and the value Im in the seat information B. If the absolute value of the difference is less than or equal to a predetermined threshold value, similar visual group information may be generated and stored in the visual group DB 11.

  Alternatively, the relationship analysis unit 10 matches the knowledge value and the emotion value for one set of the visual information A and the visual information B, and the value Im in the visual information A and the value in the visual information B If the absolute value of the difference between the Im ratio and the value “1” is equal to or smaller than a predetermined threshold, similar visual group information may be generated and stored in the visual group DB 11.

  Alternatively, the relationship analysis unit 10 matches the knowledge values and the emotion values for a set of one piece of seat information A and one piece of seat information B, and the value Jm in the seat information A and the value in the seat information B If the absolute value of the difference in Jm is equal to or smaller than a predetermined threshold value, similar visual group information may be generated and stored in the visual group DB 11.

  Alternatively, the relationship analysis unit 10 matches the knowledge values and the emotion values for a set of one piece of seat information A and one piece of seat information B, and the value Jm in the seat information A and the value in the seat information B If the absolute value of the difference between the ratio of Jm and the value “1” is equal to or less than a predetermined threshold, similar visual group information may be generated and stored in the visual group DB 11.

  Alternatively, the processing of the relationship analysis unit 10 may be as follows.

  For example, when all curator names are divided into three groups as described above, the relationship analysis unit 10 performs the following processes (1) to (4) for each of a total of nine permutations each composed of two groups. I do. The number of groups is set in advance and may be a different number.

  (1) The relationship analysis unit 10 obtains a product value Xmijg of the product of the value Xmig obtained for one group in the permutation and the value Xmjg obtained for the other group for each article content.

  (2) For example, the relationship analysis unit 10 selects all values Xmijg of 60% or more (which is a preset value and may be another value), generates a sight group ID, and generates a value for each selected value Xmijg. , Searching for all the seat information including the corresponding article ID from the seat DB 5, and for each seat information, generating the seat group information including the generated seat group ID, the curator name and the seat ID in the seat information A, and the seat group. Store in DB11.

  (3) The relation analysis unit 10 selects, for example, the highest value Xmijg, searches the viewing DB 5 for the viewing information including the article ID of the corresponding article, and the representative viewing information including the viewing group ID and the viewing ID therein. Is stored in the representative visual point DB 12.

  Alternatively, instead of (3), the relationship analysis unit 10 generates representative viewpoint information by the method described above and stores it in the representative viewpoint DB 12.

  (4) The relation analysis unit 10 generates representative curator information by the above-described method and stores it in the representative curator DB 13.

(Operation of article evaluation unit 14)
For example, when a new article is received and the update process of the article DB 4 and the viewpoint DB 5 is also completed, the article evaluation unit 14 performs the following process.

  FIG. 16 is a flowchart showing an operation flow of the article evaluation unit 14.

  First, the article evaluation unit 14 reads a first visual information group that is all visual information including the same curator name from the visual DB 5 (S1).

  The article evaluation unit 14 reads out the second viewpoint information group, which is all the viewpoint information including the article ID indicating the article, from the viewpoint DB 5 (S3).

  The article evaluation unit 14 excludes, from the first visual information group, visual information that does not include a knowledge value that matches the knowledge value of the visual information in the second visual information group (S5).

  The article evaluation unit 14 includes, in the second visual information group, visual information K including an emotion value that does not match the emotion value of the visual information in the first visual information group, or the second visual information in the first visual information group. It is determined whether or not the condition that there is the seat information L including the emotion value that does not match the emotion value of the seat information in the information group is satisfied (S9).

  If the condition is not satisfied (S9: NO), the article evaluation unit 14 reads the article name in the article information including the article ID from the article DB 4, reads out the curator attribute in the curator information including the curator name from the curator DB 3, and Evaluation value “3” indicating that the curator of the name positively evaluates the article, the article name, the evaluator name that matches the curator name, the evaluator attribute that matches the curator attribute, and the emotion in the first viewpoint information group Article evaluation information including an evaluator emotion value group made up of values and an article curator emotion value group made up of emotion values in the second viewpoint information group is generated and stored in the article evaluation DB 15 (S11), and the process ends.

  If the condition is satisfied (S9: YES), the article evaluation unit 14 is the seat information K of the second seat information group, and is different from the scene value in the seat information that is not the seat information L of the first seat information group. Are included in the second visual information group, and the visual information L in the first visual information group includes a scene value different from the scene value in the visual information that is not the visual information K in the second visual information group. It excludes from a 1st viewpoint information group (S13).

  That is, the article evaluation unit 14 differs from the scene value in the seat information in the first seat information group that does not include the emotion value that is the seat information K and does not match the emotion value of the seat information in the second seat information group. Those that include the scene value are excluded from the second visual information group, and the visual information in the second visual information group that is the visual information L and does not include an emotion value that does not match the emotion value of the visual information in the first visual information group. Those including scene values different from the scene values in the information are excluded from the first visual point information group (S13).

  The article evaluation unit 14 determines whether or not the condition is satisfied (S15).

  If the condition is not satisfied (S15: NO), the article evaluation unit 14 sets the evaluation value “2” indicating that the curator with the curator name partially evaluates the article partially, the article name of the article, and the curator name. Article curator emotion value consisting of an evaluator attribute that matches the corresponding evaluator name, an evaluator attribute that matches the corresponding curator attribute, an evaluator emotion value group consisting of emotion values in the first visual information group, and an emotion value in the second visual information group Article evaluation information including a group is generated and stored in the article evaluation DB 15 (S17), and the process ends.

  If the condition is satisfied (S15: YES), the article evaluation unit 14 evaluates the evaluation value “1” indicating that the curator with the curator name negatively evaluates the article, the evaluation corresponding to the article name and curator name of the article. Includes an evaluator attribute that matches the name of the person, a corresponding curator attribute, an evaluator emotion value group that consists of emotion values in the first visual information group, and an article curator emotion value group that consists of emotion values in the second visual information group Article evaluation information is generated and stored in the article evaluation DB 15 (S19), and the process ends.

  Alternatively, the article evaluation unit 14 performs the following processing.

  The article evaluation unit 14 reads all curator names from the curator DB 3 and divides them into, for example, three groups. The number of groups is set in advance and may be a different number.

  The article evaluation unit 14 reads all the article contents from the article DB 4 and reads all the knowledge values from the viewpoint DB 5, and based on these, the article contents are group-like for one article contents and one group. The indicated value Xmig is determined. The article evaluation unit 14 also obtains the value Xmig for other groups.

  The article evaluation unit 14 reads all emotion expressions from the viewpoint DB 5, and based on all article contents and all emotion expressions, a value indicating the degree that the article contents are a group for one article content and one group. Xmjg is obtained. The article evaluation unit 14 obtains the value Xmjg for the other sets.

  As shown in FIG. 17, the article evaluation unit 14 uses the LDA, information P1 indicating each article content, each group, each value Xmig, and the relationship thereof, each emotion value, each group, each value Xmjg, and the relationship thereof. The following processing is performed based on the information P2 indicating the information and the content of one article. Since the processing of other article contents is the same, the description is omitted.

  First, the article evaluation unit 14 performs the following processing for one group. Since the processing of another group is the same, the description is omitted.

  The article evaluation unit 14 reads the representative curator name and curator attribute of the group from the representative curator DB 13.

  The article evaluation unit 14 reads an emotion value from the information P2.

  The article evaluation unit 14 reads the article name in the article information including the corresponding article content from the article DB 4.

  The article evaluation unit 14 reads the emotion value in the viewpoint information including the article ID of the corresponding article content from the viewpoint DB 5.

  The article evaluation unit 14 obtains a value indicating the degree that the article content is a group.

  If the calculated value is equal to or greater than a predetermined threshold, the article evaluation unit 14 has an evaluation value “3”, an article name, an evaluator emotion value group composed of emotion values read from the information P2, and an article composed of emotion values read from the viewpoint information. Article evaluation information including a curator emotion value group, an evaluator name that matches the read curator name, and an evaluator attribute that matches the read curator attribute is generated and stored in the article evaluation DB 15.

  FIG. 18 is a diagram illustrating an example of the configuration of the article evaluation DB 15.

  The article evaluation information includes, for example, an evaluator name “Mr. B”, an evaluator attribute “military critic (navy)”, an article name “walking Yokosuka”, an evaluation value “1”, an article curator emotion value group “sadness”, an evaluation The person emotion value group “fear, surprise, sadness, anticipation” is included.

(Operations of the selection unit 16 and the curation article display unit 17)
FIG. 19 is a flowchart showing a flow of operations of the selection unit 16 and the curation article display unit 17.

  The selection unit 16 generates one or more keywords (for example, “Mikasa”, “Military Shop”) related to an externally input query (for example, “Yokosuka Chuo”) (S101).

  The selection unit 16 searches the visual point DB 5 for one or more visual point information including a knowledge value, a scene value, an emotion value, etc. that partially or entirely match the keyword (S103).

  The selection unit 16 reads out all the seat group IDs in the representative seat information including the seat ID from the representative seat DB 12 for one or more seat IDs in the searched seat information (S105).

  Here, it is assumed that the first visual group ID, the second visual group ID, and the third visual group ID are read out.

  The selection unit 16 selects a curator name (“Mr. B” in the representative curator information including the first sight group ID from the representative curator DB 13, hereinafter referred to as a representative curator name “Mr. B”), a curator attribute (“military critic” (Navy) "and the representative curator attribute" Military critic (Navy) ") is read out (S111).

  The selection unit 16 reads the article name (referred to as the representative curator article name) and the article content (referred to as the representative curator article content) in the article information including the curator name matching the representative curator name “Mr. B” from the article DB 4 (S113). .

  The selection unit 16 selects a curator name (“Mr. O”) other than the representative curator name “Mr. B” in the gaze group information including the first gaze group ID from the gaze group DB 11, and hereinafter referred to as a similar curator name “Mr. O”. ) Is read (S115).

  The selection unit 16 reads the article name (referred to as the contrasting article name) in the article evaluation information including the representative curator name “Mr. B” and the evaluation value “1” from the article evaluation DB 15 (S117).

  The selection unit 16 reads the curator name (“Mr. P” in the article information including the same article name as the contrasted article name from the article DB 4, and hereinafter referred to as the contrasted curator name “Mr. P”) (S 119).

  Or the selection part 16 performs the following processes, when narrowing down the number of contrasting article names.

  Here, the article curator emotion value groups of the three article evaluation information including the representative curator name are referred to as A, B, and C. The evaluator emotion value group of each article evaluation information is referred to as G.

  The selection unit 16 selects an AG distance D (AG), a BG distance D (BG), a CG distance D (CG), an AB distance D (AB), an AC distance D (AC), and a BC distance D ( BC) is calculated.

The selection unit 16
ACG distance D (ACG) = D (AC) + D (CG) + D (AC)
BCG distance D (BCG) = D (BC) + D (CG) + D (BG)
ABG distance D (ABG) = D (AB) + D (BG) + D (AG) is calculated.

  The selection unit 16 selects the distances D (ACG), D (BCG), and (ABG) in the order set in descending order. For example, if only the distance D (ACG) is selected, A and C are included. Read the article name (contrast article name) from the article evaluation information.

  The selection unit 16 reads a curator name (referred to as a contrast curator name) in article information including the same article name as the contrast article name from the article DB 4.

  The selection unit 16 transmits the query, the representative curator name, the representative curator attribute, the representative curator article name, the representative curator article content, the similar curator name, the contrast article name, and the contrast curator name to the curation article display section 17 (S121). ).

  For example, as shown in FIG. 19, the curation article display unit 17 includes a query “Yokosuka Chuo”, a representative curator name “B”, a representative curator attribute “military critic (navy)”, a representative curator article name “Mikasa's "Yu figure", the contents of the representative curator "I went to Mikasa Park today. I am blessed with fine weather ..., I'm excited ...", similar curator name "O-san", contrasting article name "Yokosuka walk" and A curated article including the contrasting curator name “Mr. P” is displayed (S123).

  Further, as shown in FIG. 19, the selection unit 16 and the curation article display unit 17 also display curation articles obtained by the same processing for the other representative curator name “Mr. C” (S125).

  Further, as shown in FIG. 19, the selection unit 16 and the curation article display unit 17 also display curation articles obtained by the same processing for the other representative curator name “Mr. D” (S125).

  Alternatively, as shown in FIG. 20, the curation article display unit 17 omits the representative curator name “Mr. B” and the representative curator attribute “military critic (navy)”, centering on the image indicating the query “walking Yokosuka”. An image showing “Navy B”, representative curator name “C” and representative curator attribute “curry lover” omitted “curry C”, representative curator name “D” and representative curator attribute “old clothes” A curated article including an image showing “Mr.

  In addition, the curation article display unit 17 receives information related to the representative curator name and the representative curator attribute received from the selection unit 16 around the image indicating the representative curator name and the representative curator attribute, a part thereof, and related knowledge. A curation article including an image showing notation, knowledge value, scene notation, scene value, emotion notation, emotion value, etc. is displayed.

  As described above, the information search device displays not only the information itself but also the individual or organization (curator) describing the information, the curator's perspective, the evaluation by the curator, and the related information. It is possible to determine the relevance such as similarity and difference of curators, which is smaller than information, and it is possible to relatively easily determine the relevance such as similarity and difference of information. In addition, it becomes easy to grasp the position of each piece of information and the original intention.

  That is, the curator's viewpoint can be extracted from the curator's blog and web page, and the representative curator and the representative viewpoint can be found for a specific topic. Also, any blog and web page can be evaluated based on the curator's perspective.

  In this embodiment, the case where text information is extracted from a blog, a web page, or the like is described. However, in the case of music, for example, the feature amount of a phrase is expressed in knowledge, composed, and played for a song. By defining the location, the era, and the song itself as a scene notation, and the phrase that recalls some emotion as an emotion notation, the composer can be evaluated as a curator as well as text information. . Video can be handled in the same way by defining feature values.

  A computer program for causing a computer to function as an information retrieval device can be recorded on a computer-readable recording medium such as a semiconductor memory, a magnetic disk, an optical disk, a magneto-optical disk, or a magnetic tape, and a communication network such as the Internet can be used. And can be widely distributed.

1 article collection part 2 view extraction part 3 curator DB
4 Article DB
5 view DB
6 Notation extraction unit 7 Knowledge notation DB
8 Scene notation DB
9 Emotion expression DB
10 Relation Analysis Department 11 View Group DB
12 Representative view DB
13 Representative Curator DB
14 Article Evaluation Department 15 Article Evaluation DB
16 Selection part 17 Curation article display part

Claims (7)

A knowledge notation DB for storing knowledge notation used for articles and knowledge information including knowledge values related to the knowledge notation;
A scene notation DB in which scene information including scene values used for articles and scene values related to the scene notations are stored;
An emotion notation DB in which emotion information including emotion values used in articles and emotion values related to the emotion notations are accumulated;
An article DB in which article information including the article ID of the article, the curator name of the article, and the article content is accumulated;
Article ID of the article, curator name of the article, knowledge notation appearing in the article content of the article, knowledge value related to the knowledge notation, scene notation appearing in the article content of the article, scene value related to the scene notation, A view DB for storing view information including the emotion notation appearing in the article content, the emotion value related to the emotion notation, and the view ID;
For a set of knowledge notation, scene notation, and emotion notation in the article content of the article information stored in the article DB, a knowledge value in the knowledge information including the knowledge notation is read from the knowledge notation DB, and from the scene notation DB The scene value in the scene information including the scene notation is read, the emotion value in the emotion information including the emotion notation is read from the emotion notation DB, the article ID of the article information, the knowledge notation, the knowledge value, the scene notation, A notation extraction unit that generates search result information including scene values, emotion expressions, and emotion values;
Read the article ID, the knowledge notation, knowledge value, scene notation, scene value, emotion notation and emotion value from the search result information, read the curator name in the article information including the article ID from the article DB, the article ID, A visual point extraction unit that generates visual point information including a curator name, knowledge notation, knowledge value, scene notation, scene value, emotion notation, emotion value and a visual point ID indicating these, and stores it in the visual point DB. Information retrieval device. Article ID of the article, curator name of the article, knowledge notation appearing in the article content of the article, knowledge value related to the knowledge notation, scene notation appearing in the article content of the article, scene value related to the scene notation, A view DB that stores view information including an emotion notation appearing in the article content, an emotion value related to the emotion notation, and a view ID;
An article evaluation DB for storing article evaluation information including an evaluation value of the article content of the article;
From the seat DB, the first seat information group that is all seat information including one curator name is read, and from the seat DB, the second seat information group that is all seat information including one article ID is read. From the first view information group, the view information that does not include the knowledge value that matches the knowledge value of the view information in the second view information group is excluded, and the second view information group includes the view information in the first view information group. The condition that there is seat information that includes an emotion value that does not match the emotion value, or that there is seat information that includes an emotion value that does not match the emotion value of the seat information in the second seat information group is satisfied. If not, article evaluation information including an evaluation value indicating that the curator with the curator name positively evaluates the article with the article ID is stored in the article evaluation DB, and if the condition is satisfied, Perspective information emotion In the visual information in the first visual information group that does not include an emotion value that does not match the emotion value of the visual information in the second visual information group, The visual information in the first visual information group includes the emotional value that does not match the emotional value of the visual information in the second visual information group. In the first visual information group, information that includes a scene value different from the scene value in the visual information in the second visual information group that does not include an emotion value that does not match the emotion value in the visual information in the first visual information group is excluded from the first visual information group. The second visual information group includes visual information that includes an emotion value that does not match the emotion value of the visual information in the first visual information group. Alternatively, the first visual information group includes the visual information in the second visual information group. The condition that there is sight information that includes emotion values that do not match the emotion values of the information is satisfied If so, article evaluation information including an evaluation value indicating that the curator with the curator name partially evaluates the article with the article ID is stored in the article evaluation DB, and if the condition is satisfied, the curator with the curator name An article evaluation unit that stores article evaluation information including an evaluation value indicating that an article with an article ID is negatively evaluated is stored in the article evaluation DB. A knowledge notation DB for storing knowledge notation used for articles and knowledge information including knowledge values related to the knowledge notation;
A scene notation DB in which scene information including scene values used for articles and scene values related to the scene notations are stored;
An emotion notation DB in which emotion information including emotion values used in articles and emotion values related to the emotion notations are accumulated;
An article DB in which article information including the article ID of the article, the curator name of the article, and the article content is accumulated;
Article ID of the article, curator name of the article, knowledge notation appearing in the article content of the article, knowledge value related to the knowledge notation, scene notation appearing in the article content of the article, scene value related to the scene notation, A view DB for storing view information including the emotion notation appearing in the article content, the emotion value related to the emotion notation, and the view ID;
For a set of knowledge notation, scene notation, and emotion notation in the article content of the article information stored in the article DB, a knowledge value in the knowledge information including the knowledge notation is read from the knowledge notation DB, and from the scene notation DB The scene value in the scene information including the scene notation is read, the emotion value in the emotion information including the emotion notation is read from the emotion notation DB, the article ID of the article information, the knowledge notation, the knowledge value, the scene notation, A notation extraction unit that generates search result information including scene values, emotion expressions, and emotion values;
Read the article ID, the knowledge notation, knowledge value, scene notation, scene value, emotion notation and emotion value from the search result information, read the curator name in the article information including the article ID from the article DB, the article ID, A seat extraction unit that generates curator information including curator name, knowledge notation, knowledge value, scene notation, scene value, emotion notation, emotion value and a seat ID indicating these, and stores the information in the seat DB;
A seat group ID indicating seat information generated by an article by any one or more curators, a curator name of any of the curators, and a seat ID included in the seat information generated by any of the articles by the curator. A view group DB in which view group information is stored;
Representative seat DB that stores representative seat information including a seat group ID indicating seat information generated by an article by one or more curators and seat IDs included in seat information generated by an article by the representative curator. When,
A representative curator DB that stores representative curator information including a visual group group ID indicating a visual point information generated by an article by any one or more curators, a curator name of the representative curator, and a curator attribute for the curator;
An article evaluation DB for storing article evaluation information including an evaluation value of the article content of the article;
From the seat DB, the first seat information group that is all seat information including one curator name is read, and from the seat DB, the second seat information group that is all seat information including one article ID is read. From the first view information group, the view information that does not include the knowledge value that matches the knowledge value of the view information in the second view information group is excluded, and the second view information group includes the view information in the first view information group. The condition that there is seat information that includes an emotion value that does not match the emotion value, or that there is seat information that includes an emotion value that does not match the emotion value of the seat information in the second seat information group is satisfied. If not, article evaluation information including an evaluation value indicating that the curator with the curator name positively evaluates the article with the article ID is stored in the article evaluation DB, and if the condition is satisfied, Perspective information emotion In the visual information in the first visual information group that does not include an emotion value that does not match the emotion value of the visual information in the second visual information group, The visual information in the first visual information group includes the emotional value that does not match the emotional value of the visual information in the second visual information group. In the first visual information group, information that includes a scene value different from the scene value in the visual information in the second visual information group that does not include an emotion value that does not match the emotion value in the visual information in the first visual information group is excluded from the first visual information group. The second visual information group includes visual information that includes an emotion value that does not match the emotion value of the visual information in the first visual information group. Alternatively, the first visual information group includes the visual information in the second visual information group. The condition that there is sight information that includes emotion values that do not match the emotion values of the information is satisfied If so, article evaluation information including an evaluation value indicating that the curator with the curator name partially evaluates the article with the article ID is stored in the article evaluation DB, and if the condition is satisfied, the curator with the curator name An article evaluation unit that stores article evaluation information including an evaluation value indicating that an article with an article ID is negatively evaluated in the article evaluation DB;
Based on a query inputted from the outside, the seat information is searched from the seat DB, and the seat group ID in the representative seat information including the seat ID is read from the representative seat DB for the seat ID in the seat information, The representative curator name that is the curator name and the representative curator attribute that is the curator attribute in the representative curator information including the view group ID is read from the representative curator DB, and the article information including the curator name that matches the representative curator name is read from the article DB. The representative curator article name that is the article name and the representative curator article content that is the article content are read out, and a similar curator name that is a curator name other than the representative curator name in the visual group information including the visual group ID is read from the visual group DB Read the relevant price from the article evaluation DB Read the contrasting article name which is the article name in the article evaluation information including the curator name and the evaluation value indicating negative evaluation of the article, and the article information including the same article name as the contrasting article name from the article DB A selection unit for reading a contrast curator name which is a curator name in
A curation article display section for displaying a curation article including the query, representative curator name, representative curator attribute, representative curator article name, representative curator article content, similar curator name, contrast article name, and contrast curator name An information retrieval apparatus characterized by that. An operation method of an information retrieval device,
The information search device includes:
A knowledge notation DB for storing knowledge notation used for articles and knowledge information including knowledge values related to the knowledge notation;
A scene notation DB in which scene information including scene values used for articles and scene values related to the scene notations are stored;
An emotion notation DB in which emotion information including emotion values used in articles and emotion values related to the emotion notations are accumulated;
An article DB in which article information including the article ID of the article, the curator name of the article, and the article content is accumulated;
Article ID of the article, curator name of the article, knowledge notation appearing in the article content of the article, knowledge value related to the knowledge notation, scene notation appearing in the article content of the article, scene value related to the scene notation, A view DB for storing view information including an emotion notation appearing in the article content, an emotion value related to the emotion notation and a view ID,
The operation method is as follows:
The notation extraction unit of the information search device includes a knowledge notation, a scene notation, and an emotion notation in the article content of the article information stored in the article DB, in the knowledge information including the knowledge notation from the knowledge notation DB. Read the knowledge value, read the scene value in the scene information including the scene notation from the scene notation DB, read the emotion value in the emotion information including the emotion notation from the emotion notation DB, the article ID of the article information, Generating search result information including the knowledge notation, knowledge value, scene notation, scene value, emotion notation, and emotion value;
The viewpoint extraction unit of the information search device reads the article ID, the knowledge notation, the knowledge value, the scene notation, the scene value, the emotion notation, and the emotion value from the search result information, and includes the article ID from the article DB. The curator name is read out, and the article ID, curator name, knowledge notation, knowledge value, scene notation, scene value, emotion notation, emotion value, and seat information indicating these are generated and stored in the seat DB. And a method for operating the information search apparatus. An operation method of an information retrieval device,
The information search device includes:
Article ID of the article, curator name of the article, knowledge notation appearing in the article content of the article, knowledge value related to the knowledge notation, scene notation appearing in the article content of the article, scene value related to the scene notation, A view DB that stores view information including an emotion notation appearing in the article content, an emotion value related to the emotion notation, and a view ID;
An article evaluation DB for storing article evaluation information including an evaluation value of the article content of the article, and the operation method includes:
The article evaluation unit of the information search device reads from the viewpoint DB the first viewpoint information group that is all the viewpoint information including one curator name, and all the viewpoint information including one article ID from the viewpoint DB. Reading the second visual information group, and excluding visual information from the first visual information group that does not include a knowledge value that matches the knowledge value of the visual information in the second visual information group;
The article evaluation unit includes seat information including an emotion value not matching the emotion value of the seat information in the first seat information group in the second seat information group, or the second seat information in the first seat information group. Article evaluation including an evaluation value indicating that the curator with the curator name positively evaluates the article with the article ID if the condition that there is the seat information including the emotion value that does not match the emotion value of the seat information within the group is not satisfied. If the information is stored in the article evaluation DB and the condition is satisfied, it is the seat information in the second seat information group including the emotion value that does not match the emotion value of the seat information in the first seat information group, Exclude from the second visual information group the one containing a scene value different from the scene value in the visual information in the first visual information group that does not include an emotion value that does not match the emotion value of the visual information in the visual information group, Matches the emotional value of the visual information in the visual information group Scenes in the visual information in the second visual information group that do not include emotion values that do not match the emotion values of the visual information in the first visual information group, which are the visual information in the first visual information group that includes no emotion value Excludes scene values different from the first visual information group, and the second visual information group includes visual information including emotion values that do not match the emotion values of the visual information in the first visual information group, or If the condition that the first visual point information group has visual point information including an emotion value that does not match the emotion value of the visual point information in the second visual point information group is not satisfied, the curator with the curator name partially selects the article with the article ID. Is stored in the article evaluation DB, and if the condition is satisfied, an evaluation indicating that the curator with the curator name evaluates the article with the article ID negatively. Article evaluation information including values Method of operating the information retrieval device, characterized in that it comprises the steps of storing in said article evaluation DB a. An operation method of an information retrieval device,
The information search device includes:
A knowledge notation DB for storing knowledge notation used for articles and knowledge information including knowledge values related to the knowledge notation;
A scene notation DB in which scene information including scene values used for articles and scene values related to the scene notations are stored;
An emotion notation DB in which emotion information including emotion values used in articles and emotion values related to the emotion notations are accumulated;
An article DB in which article information including the article ID of the article, the curator name of the article, and the article content is accumulated;
Article ID of the article, curator name of the article, knowledge notation appearing in the article content of the article, knowledge value related to the knowledge notation, scene notation appearing in the article content of the article, scene value related to the scene notation, A view DB for storing view information including the emotion notation appearing in the article content, the emotion value related to the emotion notation, and the view ID;
For a set of knowledge notation, scene notation, and emotion notation in the article content of the article information stored in the article DB, a knowledge value in the knowledge information including the knowledge notation is read from the knowledge notation DB, and from the scene notation DB The scene value in the scene information including the scene notation is read, the emotion value in the emotion information including the emotion notation is read from the emotion notation DB, the article ID of the article information, the knowledge notation, the knowledge value, the scene notation, A notation extraction unit that generates search result information including scene values, emotion expressions, and emotion values;
Read the article ID, the knowledge notation, knowledge value, scene notation, scene value, emotion notation and emotion value from the search result information, read the curator name in the article information including the article ID from the article DB, the article ID, A seat extraction unit that generates curator information including curator name, knowledge notation, knowledge value, scene notation, scene value, emotion notation, emotion value and a seat ID indicating these, and stores the information in the seat DB;
A seat group ID indicating seat information generated by an article by any one or more curators, a curator name of any of the curators, and a seat ID included in the seat information generated by any of the articles by the curator. A view group DB in which view group information is stored;
Representative seat DB that stores representative seat information including a seat group ID indicating seat information generated by an article by one or more curators and seat IDs included in seat information generated by an article by the representative curator. When,
A representative curator DB that stores representative curator information including a visual group group ID indicating a visual point information generated by an article by any one or more curators, a curator name of the representative curator, and a curator attribute for the curator;
An article evaluation DB for storing article evaluation information including an evaluation value of the article content of the article;
From the seat DB, the first seat information group that is all seat information including one curator name is read, and from the seat DB, the second seat information group that is all seat information including one article ID is read. From the first view information group, the view information that does not include the knowledge value that matches the knowledge value of the view information in the second view information group is excluded, and the second view information group includes the view information in the first view information group. The condition that there is seat information that includes an emotion value that does not match the emotion value, or that there is seat information that includes an emotion value that does not match the emotion value of the seat information in the second seat information group is satisfied. If not, article evaluation information including an evaluation value indicating that the curator with the curator name positively evaluates the article with the article ID is stored in the article evaluation DB, and if the condition is satisfied, Perspective information emotion In the visual information in the first visual information group that does not include an emotion value that does not match the emotion value of the visual information in the second visual information group, The visual information in the first visual information group includes the emotional value that does not match the emotional value of the visual information in the second visual information group. In the first visual information group, information that includes a scene value different from the scene value in the visual information in the second visual information group that does not include an emotion value that does not match the emotion value in the visual information in the first visual information group is excluded from the first visual information group. The second visual information group includes visual information that includes an emotion value that does not match the emotion value of the visual information in the first visual information group. Alternatively, the first visual information group includes the visual information in the second visual information group. The condition that there is sight information that includes emotion values that do not match the emotion values of the information is satisfied If so, article evaluation information including an evaluation value indicating that the curator with the curator name partially evaluates the article with the article ID is stored in the article evaluation DB, and if the condition is satisfied, the curator with the curator name An article evaluation unit that stores article evaluation information including an evaluation value indicating that the article with the article ID is negatively evaluated in the article evaluation DB,
The operation method is as follows:
Based on a query input from the outside, the selection unit of the information search device searches for the seat information from the seat DB, and for the seat ID in the seat information, the representative seat information including the seat ID from the representative seat DB. The representative curator name that is the curator name and the representative curator attribute that is the curator attribute in the representative curator information that includes the visual point group ID is read from the representative curator DB, and the representative curator name is read from the article DB. The representative curator article name that is the article name in the article information including the matching curator name and the representative curator article content that is the article content are read, and other than the representative curator name in the visual group information including the visual group group ID from the visual group DB Read similar curator names that are curator names Read the contrast article name that is the article name in the article evaluation information including the representative curator name and the evaluation value indicating negative evaluation of the article from the article evaluation DB, and the contrast article name from the article DB Retrieving a contrasting curator name, which is a curator name in article information containing the same article name;
The curation article display unit of the information search device includes a curation including the query, representative curator name, representative curator attribute, representative curator article name, representative curator article content, similar curator name, contrast article name, and contrast curator name. And a step of displaying an article. An operation method of an information search apparatus, comprising:   A computer program for causing a computer to function as the information search device according to claim 1.

Images