Knowledge information generation device and program
The knowledge information generation device and program address the challenge of integrating non-educational content with curriculum guidelines by extracting and structuring relevant words to create a knowledge graph, facilitating the search for related content that aligns with educational themes and everyday topics.
Patent Information
- Application Number
- JP2023078879
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-05-11
- Publication Date
- 2024-11-21
AI Technical Summary
Existing technologies struggle to efficiently search for and utilize content not originally produced for educational purposes in a way that aligns with curriculum guidelines, particularly in relating everyday or familiar topics to learning elements, due to the lack of formal computer-processable data expressing relationships between elements and subjects.
A knowledge information generation device and program that extracts and structures theme and content data to generate knowledge information, using methods like TF-IDF and morphological analysis to identify 'element' and 'subject' words, and creates a graph-structured knowledge graph to facilitate the search for related content.
Enables the automatic and efficient generation of knowledge information, allowing for the discovery of related content that aligns with curriculum guidelines and everyday topics, enhancing the educational value of non-educational content.
Smart Images

Figure 2024162900000001_ABST
Abstract
Description
[Technical field]
[0001] The present invention relates to a knowledge information generating device and a program. [Background technology]
[0002] A technology is used for searching for related content related to a content, based on the content already selected, etc. The technology for searching for related content is an important technology used for recommendations to users, etc.
[0003] For example, in the field of education, learning services that utilize the distribution of content over the Internet are becoming increasingly important. Furthermore, the field of education is emphasizing the need for learning services that do not just follow a uniform learning order, but provide content that matches the learner's understanding and progress. Therefore, methods are being researched that automatically present content (learning content) to users (learners) and instructors.
[0004] Patent Document 1 discloses a technology that defines a learning area based on a curriculum guideline and recommends teaching materials to a user from within that area based on the grade, subject, or textbook unit.
[0005] Non-patent document 1 discloses a method for describing the relationship between learning content and other (non-learning) video content in the data format of RDF (Resource Description Framework), which supports the Semantic Web, and for explicitly presenting the relationship between the two. [Prior art documents] [Patent documents]
[0006] [Patent Document 1] JP 2015-018159 A [Non-patent literature]
[0007] [Non-Patent Document 1] Makoto Urakawa, Hiroshi Fujisawa, "Verification of the Use of Video Content Linked to Educational Applications Using Structured Data," ITE Technical Report, Vol. 42, No. 11, BCT2018-38, pp. 1-4 Summary of the Invention [Problem to be solved by the invention]
[0008] For example, in the field of education, it is recognized that it is effective to use content that was not originally produced for educational purposes for educational purposes. In order to use content that was not produced for educational purposes in this way, it is desirable to select appropriate content based on the relevance between the content.
[0009] The Japanese government's curriculum guidelines also emphasize the need to relate each learning topic to everyday life and familiar topics, but the guidelines themselves do not provide specific examples of everyday life or familiar topics.
[0010] The above "specific examples of everyday and familiar topics" are called "topics." Topics are information contained in the content that is useful for understanding the "elements" of systematized information (in education and learning, learning elements). In other words, topics are useful as information for understanding elements.
[0011] With the technologies described in Patent Document 1 and Non-Patent Document 1, it is difficult or impossible to present, as related content to a given piece of content, content that is not produced for educational purposes but is a topic of everyday life or familiar topics related to a learning item (including elements).
[0012] As explained above using education as an example, the objective of the present invention is to make it possible to search for related content that includes appropriate "topics", not limited to content created for a specific purpose. Furthermore, in order to search for such related content, knowledge about the subject field is required. In other words, computer-processable formal data (called knowledge information) that shows the relationship between "elements" and "topics" is required.
[0013] The present invention has been made based on the above problem recognition, and aims to provide a knowledge information generating device and a program capable of efficiently generating the above knowledge information. [Means for solving the problem]
[0014] [1] In order to solve the above problem, a knowledge information generation device according to one embodiment of the present invention includes a theme data acquisition unit that acquires theme data including a pair of a theme code for identifying a theme and theme text information associated with the theme code; a content data acquisition unit that acquires content data including a pair of content identification information for identifying a content and content text information associated with the content identification information; an element and subject extraction unit that determines "element words" from among words extracted from the content text information based on whether or not the words are included in the theme text information, calculates a score related to importance for each of the words extracted from the content text information, and determines "subject words" from among the words extracted from the content text information based on the score; and a structuring processing unit that generates knowledge information by associating the theme code with the "element words" and the "subject words" extracted from the content text information related to the theme code.
[0015] [2] Moreover, one aspect of the present invention is that in the knowledge information generation device of [1] above, the element and subject extraction unit determines a predetermined number of words having the top importance scores among the words extracted from the content text information as the "subject words".
[0016] [3] In accordance with another aspect of the present invention, in the knowledge information generation device of [2] above, the element and topic extraction unit determines a score relating to the importance of each of the terms extracted from the content text information by using a PositionRank method or a TF-IDF (term frequency and inverse document frequency) method.
[0017] [4] Moreover, one aspect of the present invention is that in the knowledge information generation device of [2] or [3] above, the element and subject extraction unit determines, among the words extracted from the content text information, to exclude from the "subject words" those words which are in the top predetermined number of importance scores and which have been determined to be the "element words".
[0018] [5] Moreover, one aspect of the present invention is that in any of the knowledge information generation devices described above in [1] to [4], the structuring processing unit generates the knowledge information as graph-structured data having a node corresponding to the theme code, a node corresponding to the "element word", and a node corresponding to the "topic word", a first hierarchical link between the node corresponding to the theme code and a node corresponding to the "element word" related to the node corresponding to the theme code, and a second hierarchical link between the node corresponding to the "element word" and a node corresponding to the "topic word" related to the node corresponding to the "element word".
[0019] [6] Also, one aspect of the present invention is a program for causing a program to function as a knowledge information generating device, the program including: a theme data acquiring unit that acquires theme data including a pair of a theme code for identifying a theme and theme text information associated with the theme code; a content data acquiring unit that acquires content data including a pair of content identification information for identifying a content and content text information associated with the content identification information; an element and subject extraction unit that determines "element words" from among words extracted from the content text information based on whether or not the words are included in the theme text information, calculates a score related to importance for each of the words extracted from the content text information, and determines "subject words" from among the words extracted from the content text information based on the score; and a structuring processing unit that generates knowledge information by associating the theme code with the "element words" and the "subject words" extracted from the content text information with respect to the theme code.
[0020] [Reference Aspect 1] A related content search device according to one reference aspect includes a knowledge information acquisition unit that acquires knowledge information obtained by associating a theme code for identifying a theme, a "element word" which is a word related to content related to the theme code, and a "topic word" which is a word related to content related to the theme code, a content data acquisition unit that acquires content data including content identification information for identifying content and a keyword list including one or more keywords associated with the content identification information, a content linking processing unit that links the content identification information of the content data to the "topic word" associated with the theme code when a keyword included in the keyword list of the content data matches the "topic word" associated with a predetermined theme code of the knowledge information, and an "original content identification information" which is a basis for searching for related content. a related content search unit that, when "original content identification information" is acquired, searches the knowledge information to (A) acquire, as related content identification information, content identification information of related content other than the "original content identification information," which has in the keyword list a common "subject word" with the keyword in the keyword list associated with the "original content identification information" and which is associated with a specific theme code in the knowledge information, or (B) acquires, as related content identification information, content identification information of related content other than the "original content identification information," which has in the keyword list a common "subject word" with the keyword in the keyword list associated with the "original content identification information," which is associated with a specific theme code in the knowledge information, and outputs the acquired related content identification information as a search result.
[0021] [Reference Aspect 2] Also, in one reference aspect, in the related content search device of [Reference Aspect 1] above, the knowledge information includes information indicating whether or not, for a specific theme code, one of the "element words" has a direct relationship with one of the "topic words", and the content linking processing unit, when linking one content identification information to one "topic word", (1) if the keyword list associated with the content identification information includes the "element word" having the direct relationship with the "topic word" as a keyword, links the content identification information as content identification information of "content in which an element can be understood using the topic as an example", and (2) if the keyword list does not include the "element word" having the direct relationship with the "topic word" as a keyword, links the content identification information as content identification information of "topic content".
[0022] [Reference Aspect 3] Also, in one reference aspect, in the related content search device of [Reference Aspect 2] above, the related content search unit (B) performs a process of acquiring content identification information of related content other than the "original content identification information" that has a common "element word" in the keyword list with the keyword in the keyword list associated with the "original content identification information" and that is associated with a specific theme code in the knowledge information, as related content identification information, limited to cases where the "original content identification information" is content identification information of "content whose elements can be understood by taking a subject as an example" and the related content identification information is content identification information of "content whose elements can be understood by taking a subject as an example."
[0023] [Reference Mode 4] In addition, one reference mode is a related content search device in any of [Reference Mode 1] to [Reference Mode 3] above, further comprising a presentation unit that presents related content identified by the related content identification information output by the related content search unit, or information regarding the related content, based on the related content identification information output by the related content search unit.
[0024] [Reference Aspect 5] Also, one reference aspect includes a knowledge information acquisition unit that acquires knowledge information formed by associating a theme code for identifying a theme, a "element word" which is a word related to content related to the theme code, and a "topic word" which is a word related to content related to the theme code; a content data acquisition unit that acquires content data including content identification information for identifying content and a keyword list including one or more keywords associated with the content identification information; a content linking processing unit that links the content identification information of the content data to the "topic word" associated with the theme code when the keyword included in the keyword list of the content data matches the "topic word" associated with a predetermined theme code of the knowledge information; and a content linking processing unit that links the content identification information of the content data to the "topic word" associated with the theme code when the "original content identification information" that is the basis for searching for related content is acquired. Therefore, the program causes a computer to function as a related content search device that includes: (A) acquiring, as related content identification information, content identification information of related content other than the "original content identification information" that has a common "subject word" in the keyword list with the "subject word" that is a keyword in the keyword list associated with the "original content identification information" and that is associated with a specific theme code in the knowledge information; or (B) acquiring, as related content identification information, content identification information of related content other than the "original content identification information" that has a common "subject word" in the keyword list with the "element word" that is a keyword in the keyword list associated with the "original content identification information" and that is associated with a specific theme code in the knowledge information; and outputting the acquired related content identification information as a search result. Effect of the Invention
[0025] According to the present invention, knowledge information can be generated automatically and efficiently. [Brief description of the drawings]
[0026] [Figure 1] 1 is a block diagram showing a schematic configuration example of a knowledge utilization system according to an embodiment of the present invention; [Diagram 2] 1 is an entity relationship diagram (ER diagram) for explaining the configuration of information used in the embodiment. [Diagram 3] 2 is a block diagram showing a schematic functional configuration of a knowledge information generation device according to the embodiment. FIG. [Figure 4] 2 is a schematic diagram showing an example of the configuration of content data (learning content data) to be processed in the embodiment. FIG. [Diagram 5] 2 is a schematic diagram showing an example of the configuration of theme data (course of study data) to be processed in the embodiment. FIG. [Figure 6] FIG. 11 is a schematic diagram showing an example of a result of sorting nouns extracted from text of content data (learning content data) in descending order of score in the embodiment. [Figure 7] 13 is a schematic diagram showing an example of a knowledge graph obtained as a result of a structuring processing unit processing information passed from an element and subject extraction unit in the knowledge information generation device of the embodiment. FIG. [Figure 8] FIG. 2 is a schematic diagram showing an example of content data (first content data) having information about content (learning content) in the embodiment. [Figure 9] FIG. 2 is a schematic diagram showing an example of content data (second content data) having information about content (learning content) in the embodiment. [Figure 10] 2 is a schematic diagram showing an example of the configuration of theme data (course of study data) to be processed in the embodiment. FIG. [Figure 11]FIG. 9 is a schematic diagram showing a list of nouns extracted from text contained in the content data (learning content data) shown in FIG. 8 in descending order of importance score (in order of importance) in the embodiment. [Figure 12] FIG. 10 is a schematic diagram showing a list of nouns extracted from texts contained in the content data (learning content data) shown in FIG. 9 in descending order of importance score (in order of importance) in the embodiment. [Figure 13] FIG. 11 is a schematic diagram showing an example of one knowledge graph (knowledge information) generated by integrating information contained in two content data (FIGS. 8 and 9) in the embodiment. [Figure 14] 11 is a flowchart showing a processing procedure in which the knowledge information generation device according to the embodiment generates a knowledge graph (knowledge information). [Figure 15] 2 is a block diagram showing a schematic functional configuration of a related content search device according to the embodiment. FIG. [Figure 16] 4 is a schematic diagram showing an example of the configuration of content data acquired by a content data acquisition unit and passed to a content link processing unit in the related content search device according to the embodiment. FIG. [Figure 17] 2 is a schematic diagram showing an example of theme data (course of study data) for a theme code to be processed by the related content search device according to the embodiment. FIG. [Figure 18] FIG. 2 is a schematic diagram showing an example of a knowledge graph generated by the knowledge information generation device for a specific theme code (teaching guideline code) in the embodiment. [Figure 19] 13 is a decision table showing a method in which the related content search device according to the embodiment links content to a knowledge graph. [Figure 20] 13 is a schematic diagram showing an example in which the content linking processing unit of the related content search device according to the embodiment links content to a knowledge graph corresponding to a specific theme code (course of study code). FIG. [Figure 21]10 is a schematic diagram showing a first pattern of search by a related content search unit of the related content search device according to the embodiment. FIG. [Figure 22] 13 is a schematic diagram showing a second pattern of search by the related content search unit of the related content search device according to the embodiment. FIG. [Diagram 23] 13 is a flowchart showing a processing procedure in which a content linking processing unit of the related content search device according to the embodiment links content to a knowledge graph. [Figure 24] 13 is a flowchart showing the overall processing procedure of the related content search device according to the embodiment. [Diagram 25] FIG. 2 is a block diagram showing an example of the internal configuration of a knowledge information generating device and a related content searching device, which are elements of the knowledge utilization system in the embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0027] Next, an embodiment of the present invention will be described with reference to the drawings. This embodiment assumes the existence of systematized information, and formalizes knowledge extracted from that information as knowledge information. This embodiment also associates individual contents with the above-mentioned formalized knowledge information. Furthermore, this embodiment searches for related contents related to a given specific content by searching for knowledge information in which a large number of contents are associated (linked or linked).
[0028] In this embodiment, a curriculum guideline is used as an example of the above-mentioned systematized information. A curriculum guideline is a standard for a country's standard educational curriculum (education course) established by a government agency. In modern curriculum guideline, a code is given for each theme, such as the contents or units. This code is called a curriculum guideline code. In this way, a curriculum guideline code is given for each theme, and a more generalized concept of the curriculum guideline code is a "theme code." In other words, a curriculum guideline is systematized information that includes text information for each theme code (curriculum guideline code). In this embodiment, knowledge is extracted from the curriculum guideline to generate it as formalized knowledge information. The knowledge information can be expressed, for example, as information in a graph structure (a structure that can be defined as a set of nodes and links). This is called a knowledge graph. However, the method of expressing knowledge information is not limited to a graph structure (knowledge graph).
[0029] The LOD of the Course of Study can be referenced, for example, at the following URL: https: / / jp-cos.github.io / LowerSecondary / 2017 / Science
[0030] In this embodiment, a certain concept may be represented by a word or the like. Furthermore, the word or the like represents a certain concept. The word or the like associated with a concept in this manner is typically a noun or a noun phrase. However, the word or the like associated with a concept may be a word or phrase of another part of speech.
[0031] In this embodiment, the "element words" (words include phrases. Typically, they are nouns or noun phrases, but may be expressions of other parts of speech) are expressions in natural language corresponding to the main concepts constituting the theme. The "topic words" (same as above for words) are information that the content has as useful information for understanding the above-mentioned "elements" in the theme. It is not necessary for the "topic words" to appear directly in the description of the theme. In this embodiment, the "element words" and "topic words" are automatically extracted from the text in the process in which the device generates knowledge information. In other words, the "element words" and "topic words" can be extracted from the information on the theme or the information on the content by a predetermined processing procedure. Typically, the "element words" and "topic words" are extracted by a computer executing a program that realizes the above-mentioned processing procedure. In other words, on the premise that predetermined data exists in advance, the "element words" and "topic words" can be extracted by utilizing natural laws.
[0032] The method of this embodiment can also be applied to fields other than learning (information other than the curriculum guidelines). Even in the case of information other than the curriculum guidelines, this embodiment extracts knowledge based on systematized information and expresses it as knowledge information in the form of formal data. Furthermore, this embodiment links the knowledge information with content information, thereby making it possible to use the knowledge information to search for related content that is related to specific content. Specific examples of application other than the curriculum guidelines will be described later as "variant examples."
[0033] As terms used in this embodiment, ID is an abbreviation for "identifier" (identification information, identifier), and URL is an abbreviation for "Uniform Resource Locator."
[0034] FIG. 1 is a block diagram showing an example of a schematic configuration of a knowledge utilization system according to this embodiment. As shown in the figure, the knowledge utilization system 1 is configured to include a knowledge information generation device 20 and a related content search device 30. The knowledge information generation device 20 and the related content search device 30 can be realized, for example, by a computer and a program. Furthermore, each of the knowledge information generation device 20 and the related content search device 30 has a storage means as necessary. The storage means is, for example, a variable in a program or a memory allocated by the execution of a program. Furthermore, non-volatile storage means such as a magnetic hard disk device or a solid state drive (SSD) may be used as necessary. Furthermore, at least a part of the functions of each functional unit may be realized as a dedicated electronic circuit rather than a program.
[0035] The knowledge information generating device 20 extracts knowledge based on document data organized in a predetermined manner, and generates knowledge information that expresses the knowledge as formal data. An example of a document is a curriculum guideline. In the curriculum guideline, an identifiable code (called a theme code or curriculum guideline code) is assigned to each content (unit, etc.). The curriculum guideline also includes a description in text (sentence) that expresses the content of each code. In this embodiment, the knowledge information generated by the knowledge information generating device 20 is a knowledge graph. The knowledge information generating device 20 passes the generated knowledge information (knowledge graph) to the related content search device 30. A more detailed configuration and processing example of the knowledge information generating device 20 will be described in detail from FIG. 3 onwards.
[0036] The related content search device 30 receives the knowledge information generated by the knowledge information generating device 20, and searches for related content related to a specific content using the knowledge information. As an example of a specific processing method, the related content search device 30 associates a large number of pieces of content information (content attribute information, called content data) with the knowledge information in advance. In this embodiment, since the knowledge information is a knowledge graph having a graph structure, for example, the related content search device 30 links each piece of content to a node in the graph (knowledge graph). Then, the related content search device 30 searches for the knowledge information associated with the content, thereby searching for related content related to a specific content. The related content search device 30 can present information about the related content, which is the search result, or the related content itself (video content, etc.) to the user. Note that a more detailed configuration and processing example of the related content search device 30 will be described in detail in FIG. 15 and subsequent figures.
[0037] 2 is an entity relationship diagram (ER diagram) for explaining the configuration of information used in this embodiment. The diagram includes two entities, a theme and a content. Note that the diagram is drawn focusing on only these two entities, and each of these two entities may have a relationship with another entity (not shown).
[0038] A theme corresponds to an individual item contained in a document (e.g., a curriculum guideline). Each theme is identified by a theme code. When the document is a curriculum guideline, each theme (learning item such as a unit) is identified by the curriculum guideline code. That is, one specific example of a theme code is the curriculum guideline code. That is, the primary key for identifying a theme is the theme code (curriculum guideline code). The attributes of a theme include the theme code (curriculum guideline code) and theme item text. Theme item text is text such as an overview that describes the theme. In this embodiment, a case will be described where the document is a curriculum guideline. That is, the theme code is the curriculum guideline code, and the theme data that holds the attributes of the theme is curriculum guideline data.
[0039] Examples of theme data (course of study data) that holds attributes of themes will be described later with reference to Figs. 5, 10, 17, etc.
[0040] A content is a collection of information. The content can be in any of various forms, such as video, audio, drama, literature, images (photos, etc.), comics, animation, computer games, etc. The content may be distributed to a user's terminal device via a communication line (such as the Internet), may be distributed to a user's broadcast receiver as a broadcast signal, or may be recorded on a recording medium such as an HDD (hard disk drive) or SSD (solid state drive). Each piece of content is identified by a content ID. Alternatively, the content may be identified by other information. For example, if the title of the content functions as identification information, each piece of content may be identified by the title (character string). Content attributes include a title, a content summary, and a theme code (curriculum guideline code). The title is a character string (text) that represents the title (subject) of the content. The content summary is a text summary that summarizes the purpose and content of the content. The theme code (curriculum guideline code) is as explained in the entity "theme". In addition, text related to content, such as titles and content summaries, often contains key concepts (words) that characterize the content. In other words, knowledge can be extracted from text related to content, such as titles and content summaries.
[0041] A relationship exists between the entity "theme" and the entity "content". This relationship is related by the theme code (curriculum guideline code). In other words, the value of the attribute "theme code (curriculum guideline code)" that the entity "content" has specifies which "theme" the "content" is related to. The entity "theme" and the entity "content" correspond 1 to N (where 0≦N). In other words, a specific "theme" is related to zero or more "contents".
[0042] 3 is a block diagram showing a schematic functional configuration of the knowledge information generation device according to this embodiment. As shown in the figure, the knowledge information generation device 20 includes a content data acquisition unit 201, a theme data acquisition unit 202, an element and subject extraction unit 203 (a functional unit that extracts elements and subjects), and a structuring processing unit 204. As described above, the functions of the knowledge information generation device 20 can be realized by a computer, an electronic circuit, or the like.
[0043] The content data acquisition unit 201 acquires content data (learning content data) 1100 and passes the content data (learning content data) 1100 to the element and subject extraction unit 203. The content data acquisition unit 201 includes a means for storing the content data 1100 at least temporarily. The content data is information about the content (learning content). The content data includes a theme code (learning course code) to which the content (learning content) belongs and text data explaining the content of the content (learning content). The theme code (learning course code) and the text data are associated with each other. In this embodiment, since the theme code is a learning course code, the content data acquisition unit 201 acquires learning content data, which is information about the content (learning content) belonging to the learning course code. The text data explaining the content of the content (learning content) includes, for example, a text showing the title (title) of the content (learning content) and a text of an outline of the content (learning content). The content data acquisition unit 201 acquires content data by reading the content data from a recording medium such as a magnetic hard disk device or a magnetic tape, or by receiving the content data from an external device through communication. The content data acquisition unit 201 may also acquire content data through a user's input operation. Examples of learning content data will be described later with reference to Figs. 4, 8, 9, and 16.
[0044] That is, the content data acquisition unit 201 acquires content data including a pair of content identification information for identifying a content and content text information associated with the content identification information.
[0045] The theme data acquisition unit 202 acquires theme data (curriculum guideline data) and passes the theme data to the element and subject extraction unit 203. The theme data (curriculum guideline data) includes a theme code (curriculum guideline code) and text data. In the theme data (curriculum guideline data), the theme code (curriculum guideline code) and the text data are mutually associated. The text data included in the theme data is text such as an explanatory sentence corresponding to the theme code (curriculum guideline code). The theme data acquisition unit 202 acquires the theme data by reading the theme data from a recording medium such as a magnetic hard disk device or a magnetic tape, or by receiving the theme data from an external device through communication. The theme data acquisition unit 202 may also acquire the theme data by a user's input operation, as in the case of the above-mentioned content data.
[0046] That is, the theme data acquiring unit 202 acquires theme data including a pair of a theme code for identifying a theme and theme text information associated with the theme code.
[0047] The element and subject extraction unit 203 extracts element words and subject words based on text data (text data included in the learning content data) regarding the content (learning content) acquired from the content data acquisition unit 201 and text data regarding the theme code (learning guideline code) acquired from the theme data acquisition unit 202.
[0048] Here, an "element" is an important concept that constitutes a particular theme (such as a unit in a curriculum guideline). An element is expressed as a word (particularly a noun). An "element word" is a word (noun, etc.) that represents an element. Also, a "subject" is material that leads to an understanding of the above-mentioned "element." A "subject word" is a word (noun, etc.) that represents that subject. The "element noun (noun that is a learning element)" explained later is the same as or similar to the "element word" here. Also, the "subject noun" explained later is the same as or similar to the "subject word" here.
[0049] The element and subject extraction unit 203 determines "element words" from among the words extracted from the content text information based on whether or not the words are included in the theme text information, and also calculates a score relating to the importance of each of the words extracted from the content text information and determines "subject words" from among the words extracted from the content text information based on the score.
[0050] The element and subject extraction section 203 may determine a predetermined number of words having top importance scores among the words extracted from the content text information as the "subject words".
[0051] Also, the element and subject extraction unit 203 may use a PositionRank method or a TF-IDF (term frequency and document inverse frequency) method to obtain a score related to the importance of each of the words extracted from the content text information. Note that the element and subject extraction unit 203 may use a method other than PositionRank or TF-IDF to obtain a score related to the importance of each word.
[0052] In addition, the element and subject extraction unit 203 may determine to exclude from the "subject words" those words that are among the words extracted from the content text information and that are among the top predetermined number of words in terms of importance score and have been determined to be "element words."
[0053] The structuring processor 204 generates knowledge information by associating a theme code with "element words" and "topic words" extracted from the content text information related to the theme code. As described above, a specific example of knowledge information generated by the structuring processor 204 is a knowledge graph having a graph structure. Knowledge graphs will be described later with reference to Figs. 7 and 13.
[0054] The structuring processing unit 204 may generate the knowledge information as data of a graph structure having a node corresponding to a theme code, a node corresponding to an "element word", and a node corresponding to a "topic word", a first hierarchical link between the node corresponding to the theme code and a node corresponding to the "element word" related to the node corresponding to the theme code, and a second hierarchical link between the node corresponding to the "element word" and a node corresponding to the "topic word" related to the node corresponding to the "element word". The knowledge graphs illustrated in Figs. 7 and 13, which will be described later, are configured to have such first hierarchical links and second hierarchical links. The presence or absence of the second hierarchical link is information indicating the presence or absence of a direct relationship between the "element word" and the "topic word".
[0055] FIG. 4 is a schematic diagram showing an example of the configuration of content data (learning content data). As shown in the figure, the content data (learning content data) 1100 is configured to include a content ID 1104, text 1101 (content text information), and a theme code (learning guideline code) 1102. The content ID 1104 is uniquely assigned to each piece of content, and is identification information that can identify the content. The content data (learning content data) 1100 is data describing a specific piece of content (in this embodiment, a learning content). The content data shown in the figure is data describing one piece of content (learning content). This content data (learning content data) 1100 may exist for each piece of content (learning content). The text 1101 has text information of the title (title) and content summary of the content (learning content). Note that the text 1101 may have text information other than the title (title) and content summary. The text 1101 may have other text information related to the content (subtitle text, for example). The theme code (curriculum guideline code) 1102 is data of the theme code (curriculum guideline code) with which the content is associated (belongs). In the illustrated example, the value of the theme code (curriculum guideline code) 1102 is a 16-digit decimal number, "8361233512100000." The theme code (curriculum guideline code) is code information given to the theme, and may be, for example, a number or a character string of any number of digits.
[0056] The content data (learning content data) shown in FIG. 4 is acquired by the content data acquisition unit 201 and passed to the element and subject extraction unit 203.
[0057] 5 is a schematic diagram showing an example of the configuration of theme data (curriculum guideline data). As shown in the figure, theme data (curriculum guideline data) 1200 is configured to include a theme code (curriculum guideline code) 1201, theme item text 1202 (theme text information), school type 1203, and subject 1204. Of these, school type 1203 and subject 1204 are data items specific to the case where the theme data is curriculum guideline data.
[0058] Theme data (curriculum guideline data) exists for each theme code (curriculum guideline code). In the theme data (curriculum guideline data), the theme code (curriculum guideline code) 1201 is a primary key. As described in FIG. 4, the value of the theme code (curriculum guideline code) 1201 is a 16-digit decimal number in this embodiment, but the format (number of digits, etc.) of the value of the theme code (curriculum guideline code) may be different. The theme item text 1202 is a text of an explanation of the theme code (curriculum guideline code). In the illustrated example, the theme item text 1202 is a text describing "speed and direction of motion" in the curriculum guideline (science). The school type 1203 is data indicating the type of school (elementary school, junior high school, high school, etc.) that the curriculum guideline (theme) targets. The subject 1204 is data indicating the subject that the curriculum guideline (theme) targets ("science" in this example).
[0059] [Process of extracting elements and subjects] Next, a description will be given of a processing method in which the element and subject extraction unit 203 generates knowledge information based on the content data (learning content data) 1100 and the theme data (learning guideline data) 1200. The element and subject extraction unit 203 processes a combination of the content data (learning content data) 1100 and the theme data (learning guideline data) 1200 associated with the content data (learning content data) 1100. This association is performed by the values of the theme codes (learning guideline codes) (1102 and 1201).
[0060] The element and subject extraction unit 203 performs a process of extracting nouns from the text 1101 of the content data (learning content data) 1100. The extraction of the nouns is performed by morphological analysis. The element and subject extraction unit 203 also extracts nouns from the theme item text 1202 of the theme data (learning guideline data) 1200 by morphological analysis. Next, the element and subject extraction unit 203 performs scoring according to importance for each of the nouns extracted from the text 1101 of the content data (learning content data) 1100. Methods such as PositionRank and TF-IDF (term frequency - inverse document frequency) can be used as a method of scoring nouns. PositionRank is a method aimed at extracting key phrases from the title and abstract of an academic paper, and can be applied to the scoring of nouns in this embodiment. Note that both PositionRank and TF-IDF are existing technologies.
[0061] In this embodiment, keywords and key phrases are extracted from the title and summary text using the PositionRank method, but keywords and key phrases may be extracted from other data using other methods. For example, keywords and key phrases may be extracted from subtitle text data or the like that accompanies video content (including broadcast content).
[0062] Next, the element and subject extraction unit 203 rearranges (sorts) the nouns extracted from the text 1101 of the content data (learning content data) 1100 in descending order of score (i.e., in order of increasing importance) based on the result of the scoring. The element and subject extraction unit 203 acquires element nouns (nouns that are learning elements) and subject nouns from the nouns in this rearranged order. The maximum number of subject nouns acquired by the element and subject extraction unit 203 is N (N≧0). The value of N may be a configurable parameter.
[0063] Specifically, the element and subject extraction unit 203 acquires, as element nouns (nouns serving as learning elements), nouns that match any of the nouns extracted from the text 1101 of the content data (learning content data) 1100 and that are extracted from the theme item text 1202 of the theme data (learning guideline data) 1200. Also, the element and subject extraction unit 203 acquires, as subject nouns, nouns that do not match any of the nouns extracted from the theme item text 1202 of the theme data (learning guideline data) 1200 and that are ranked in the top N in terms of importance.
[0064] FIG. 6 is a schematic diagram showing an example of the result of sorting nouns extracted from text 1101 of content data (learning content data) 1100 in descending order of score. The result shown in FIG. 6 includes eight nouns, which are "skydiving," "video," "diver," "speed," "camera," "movement," "motion," and "relativity" in order of importance score. These nouns were extracted from the title "Skydiving Shooting" and the content summary "Video of skydiving divers. In a video taken with a camera that is falling at the same speed, it does not look like they are falling. Observe the skydiving motion and notice the relativity of motion." of text 1101 in the example of content data (learning content data) 1100 shown in FIG. 4.
[0065] Of these eight nouns, the fourth noun "speed" and the seventh noun "motion" match the nouns extracted from the theme item text 1202 of the theme data (learning guideline data) 1200. Therefore, the element and subject extraction unit 203 acquires these "speed" and "motion" as element nouns (nouns that become learning elements). In addition, the element and subject extraction unit 203 acquires the nouns with the top N importance scores from the remaining six nouns excluding the fourth noun "speed" and the seventh noun "motion" as the nouns that become the subject. That is, for example, when N=1, the element and subject extraction unit 203 acquires "skydiving", which has the first (highest) importance score, as the noun that becomes the subject. The remaining nouns, namely the second "image", the third "diver", the fifth "camera", the sixth "movement", and the eighth "relativity", are not acquired as element nouns (nouns that are learning elements) or as subject nouns.
[0066] In the case of the learning content in this example, the noun "skydiving" is not described in the curriculum guidelines when it comes to teaching about motion and speed in science, but it can be considered a topical noun that is a familiar and everyday topic. In other words, topical nouns such as "skydiving" appear as topics in the content due to the teaching know-how of the learning content creator.
[0067] As described above, the element and subject extraction unit 203 extracts element nouns (nouns that are learning elements) and subject nouns from the text 1101 of the examples of the content data (learning content data) 1100 by referring to the theme item text 1202 of the theme data (learning guideline data) 1200. The element and subject extraction unit 203 The theme code (learning guideline code) and information on the extraction results (information on element nouns (nouns that are learning elements) and subject nouns), i.e., in the example of Figure 6, "speed," "motion," and "skydiving," are passed to the structuring processing unit 204.
[0068] [Process of structuring knowledge] Next, the process of structuring knowledge will be described. The structuring processor 204 performs structuring processing based on the above information received from the element and subject extraction unit 203, and outputs the result as a knowledge graph.
[0069] 7 is a schematic diagram showing an example of a knowledge graph that is a result of the structuring processing unit 204 processing the information passed from the element and subject extraction unit 203. That is, the structuring processing unit 204 receives element nouns (nouns that are learning elements), subject nouns, and theme codes (learning guideline codes) passed from the element and subject extraction unit 203, and structures them as a set.
[0070] In the illustrated embodiment, the knowledge graph (knowledge information) is a graph that represents a hierarchical structure. This knowledge graph has four nodes that correspond to the theme code (curriculum guideline code) "8323233411200000", the element nouns (nouns that are learning elements) "motion" and "speed", and the subject noun "skydiving". These nodes belong to one of three hierarchies: theme code (curriculum guideline code), element, and subject. That is, in this example, the node "8323233411200000" belongs to the theme code (curriculum guideline code) hierarchy. Also, each of the nodes "motion" and "speed" belongs to the element hierarchy. Also, the node "skydiving" belongs to the subject hierarchy. The theme code (curriculum guideline code)-element-subject represents a parent-child relationship (a hierarchical relationship). The links between the nodes represent this parent-child relationship. That is, from the theme code (learning guideline code) node "8323233411200000" which is the parent, there are links to the child element nodes "motion" and "speed". Furthermore, from the element node "speed" there is a link to its child subject node "skydiving". Note that in this example, the parent of the subject "skydiving" is the node of the word with the highest score among the element (learning element) nodes. Note that the element (learning element) nodes connected to the subject node may be selected in a different manner.
[0071] [Example of integrating and outputting knowledge extracted from multiple contents (learning contents) corresponding to one theme code (curriculum code)] In addition, when there are multiple contents (learning contents) for one theme code (learning guideline code), the structuring unit 204 may structurize all the results of extracting nouns from the content data (learning content data) for the multiple contents together. An example of this will be described below.
[0072] 8 and 9 are schematic diagrams showing examples of content data for two contents (learning contents) having a common theme code (curriculum code). The first content data (learning content data) 1100 shown in FIG. 8 is content data for a content (learning content) having a title of "Global Warming". In this content data (learning content data) 1100, the value of the theme code (curriculum code) is "8323233411200000". The second content data (learning content data) 1100 shown in FIG. 9 is content data for a content (learning content) having a title of "Deterioration of the Global Environment". In this content data (learning content data) 1100, the value of the theme code (curriculum code) is "8323233411200000", which is the same as the value shown in FIG. 8.
[0073] Fig. 10 is a schematic diagram showing an example of theme data (curriculum guideline data) 1200 corresponding to each of the content data (learning content data) 1100 shown in Fig. 8 and Fig. 9. The value of the theme code (curriculum guideline code) 1201 of the theme data (curriculum guideline data) 1200 shown in Fig. 10 is "8323233411200000". The theme data (curriculum guideline data) 1200 in Fig. 10 is associated with the content data (learning content data) 1100 in Fig. 8 and Fig. 9 by the value of this theme code (curriculum guideline code). In the theme data (curriculum guideline data) 1200 in Fig. 10, the value of the school type 1203 is "junior high school" and the value of the subject 1204 is "social studies".
[0074] FIG. 11 is a schematic diagram showing a list of nouns extracted from text 1101 of content data (learning content data) 1100 shown in FIG. 8, sorted in descending order of importance score (in descending order of importance). As shown in the figure, the nouns extracted from text 1101 shown in FIG. 8 are eight nouns, namely, "global warming", "problem", "serious", "global environment", "severe", "modern", "mechanism", and "warming". Of these nouns, only the fourth noun, "global environment", matches the noun extracted from theme item text 1202 of theme data (curriculum guideline data) 1200 in FIG. 10. The element and subject extraction unit 203 acquires this "global environment" as a noun to be an element. The element and subject extraction unit 203 acquires the top N nouns of the remaining seven nouns as nouns to be subjects. 11, since N=1, the element and subject extraction unit 203 acquires only the first noun, "global warming," as a subject noun. The other nouns (the second, third, fifth, sixth, seventh, and eighth nouns) are neither element nouns nor subject nouns.
[0075] FIG. 12 is a schematic diagram showing a list of the results of sorting the nouns extracted from the text 1101 of the content data (learning content data) 1100 shown in FIG. 9 in descending order of importance score (in order of importance). The nouns (or noun phrases) extracted from the text 1101 shown in FIG. 9 are six, namely, "global environment", "ozone layer destruction", "global warming", "desertification", "forest", and "serious". Of these nouns, only the first one, "global environment", matches the noun extracted from the theme item text 1202 of the theme data (learning guideline data) 1200 shown in FIG. 10. The element and subject extraction unit 203 acquires this "global environment" as a noun to be an element. The element and subject extraction unit 203 acquires the top N of the remaining five nouns (or noun phrases) as nouns to be subjects. 12, N=1, and the first noun, "global environment," is extracted as an element noun, so element and subject extraction unit 203 acquires only the second noun, "ozone layer destruction," as a subject noun (noun phrase). The other nouns (the third, fourth, fifth, and sixth nouns) are neither element nouns nor subject nouns.
[0076] The structuring processor 204 performs processing for integrating and structuring both the element nouns (nouns that are learning elements) and topic nouns extracted in Fig. 11 and the element nouns (nouns that are learning elements) and topic nouns extracted in Fig. 12. In other words, the structuring processor 204 combines information extracted from two pieces of content data associated with a common theme code (curriculum guideline code) to generate a knowledge graph.
[0077] FIG. 13 is a schematic diagram showing an example of a knowledge graph (knowledge information) generated by the structuring processing unit 204 by integrating both the information extracted from the content data (learning content data) 1100 shown in FIG. 8 (information extracted in FIG. 11) and the information extracted from the content data (learning content data) 1100 shown in FIG. 9 (information extracted in FIG. 12). The knowledge graph shown in FIG. 13 has four nodes. The node "8323233411200000" is a node representing a theme code (learning guideline code) and belongs to the theme code hierarchy. The node "global environment" is a node representing an element noun commonly extracted from the content data (learning content data) 1100 shown in FIG. 8 and the content data (learning content data) 1100 shown in FIG. 9. The noun "global environment" is extracted from these two content data, but since it is the same noun (same concept), it is integrated on the knowledge graph. The node "global environment" belongs to the element hierarchy. The node "global warming" is a node representing a subject noun extracted from the content data (learning content data) 1100 shown in FIG. 8. The node "ozone layer destruction" is a node representing a subject noun extracted from the content data (learning content data) 1100 shown in FIG. 9. The nodes "global warming" and "ozone layer destruction" both belong to the subject hierarchy.
[0078] The knowledge graph shown in Figure 13 has the following three links. There is a link from the parent node "8323233411200000" (theme code) to the child node "Global Environment" (element noun). In addition, there is one link each leading to the child node "Global Warming" and the child node "Ozone Layer Depletion" from this parent node "Global Environment" (element noun).
[0079] As described above, the knowledge information generation device 20 generates one knowledge graph (knowledge information) based on content data about a plurality of contents corresponding to a common theme code (curriculum guideline code).
[0080] 14 is a flowchart showing the procedure of the process in which the knowledge information generating device 20 generates a knowledge graph (knowledge information). Note that this flowchart shows the flow of the process from a state in which the content data (learning content data) of one piece of content (learning content) is the target and the theme data (curriculum guideline data) corresponding to the content data (learning content data) has already been acquired. The operation procedure will be explained below with reference to this flowchart.
[0081] First, in step S001, the element and subject extraction unit 203 extracts nouns from the text 1101 of the content data (learning content data) 1100. As described above, it is possible to extract nouns from the text by processing such as morphological analysis, which is an existing technique.
[0082] Next, in step S002, the element and subject extraction unit 203 extracts nouns from theme item text 1202 of theme data (learning course data) 1200 corresponding to the content data (learning content data) 1100 processed in step S001. The content data (learning content data) 1100 and the theme data (learning course data) 1200 are associated with each other by the value of a theme code (learning course code).
[0083] Next, in step S003, the element and subject extraction unit 203 performs a scoring process according to importance for each of the nouns extracted in step S001 from the text 1101 of the content data (learning content data) 1100. The noun scoring process can be performed, for example, using the previously mentioned PositionRank, which is an existing technology. In addition, the element and subject extraction unit 203 performs a process of sorting the nouns extracted from the text 1101 of the content data (learning content data) 1100 in descending order of the score values (i.e., so that the nouns are arranged in order from the nouns with higher importance to the nouns with lower importance).
[0084] Step S004 represents control of the repetitive process. That is, for each of the nouns sorted by the process of step S003, the element and subject extraction unit 203 repeats the processes of steps S005 to S008 in descending order of the score value. That is, for each of the nouns (which may include noun phrases) extracted from the text 1101 of the content data (learning content data) 1100, the element and subject extraction unit 203 performs a process of determining whether the noun is an element noun (a noun that is a learning element), a subject noun, or neither. The process in this repetitive loop is as described in the following steps S005, S006, S007, and S008.
[0085] In step S005, the element and subject extraction unit 203 judges whether the noun currently being judged matches any of the nouns extracted in step S002 from the theme item text 1202 of the theme data (course of study data) 1200. If they match (step S005: YES), the process proceeds to step S006. If they do not match (step S005: NO), the process proceeds to step S007.
[0086] When the process proceeds to step S006, in that step, the element and subject extraction unit 203 acquires the noun currently being judged as a noun that will be an element. After the process of step S006 ends, the process for the noun being judged ends, and the process returns to step S004 to process the next noun (if any remains).
[0087] Next, when the process proceeds to step S007, in this step, the element and subject extraction unit 203 judges whether or not the noun currently being judged should be acquired as a subject noun. In other words, it judges whether or not the number of subject nouns already acquired (number of acquired subjects) is less than N. The value of N is set in advance, as described above. If the number of acquired subjects up to this point is less than N (step S007: YES), the process proceeds to step S008. If the number of acquired subjects up to this point has already reached N (step S007: NO), the noun being judged is not acquired as a subject noun, the process for the noun being judged is terminated, and the process returns to step S004 to process the next noun (if any remains).
[0088] Next, when the process proceeds to step S008, in this step, the element and subject extraction unit 203 acquires the noun currently being judged as the subject noun. After the process of step S008 ends, the process for the noun being judged ends, and the process returns to step S004 to process the next noun (if any remains).
[0089] When the repetitive process starting from step S004 has been completed for all noun candidates, the process ends and proceeds to step S009.
[0090] Next, in step S009, the structuring processor 204 structures the information extracted in the processes up to this point, and generates and outputs a knowledge graph. That is, the structuring processor 204 groups information on theme codes (learning guideline codes), element nouns (nouns that are learning elements), and subject nouns into sets, generates and outputs data (knowledge graph) that expresses the relationships between them in a graph structure.
[0091] Next, a method for searching for related content using the knowledge information generated by the above-described knowledge information generating device will be described.
[0092] 15 is a block diagram showing a schematic functional configuration of a related content search device according to this embodiment. As shown in the figure, related content search device 30 includes a knowledge information acquisition unit 301, a theme data acquisition unit 302, a content data acquisition unit 303, a content link processing unit 304, a content ID acquisition unit 305, a related content search unit 306, and a presentation unit 307. As described above, the functions of related content search device 30 can be realized by a computer, an electronic circuit, or the like.
[0093] In the process executed by the related content search device 30, the content is uniquely identified by a content ID (content identification information). That is, the process of linking a certain content to a specific node (a node of a subject word) in a knowledge graph (knowledge information) is equivalent to the process of linking the content ID of that content to that node. Furthermore, when searching for related content, the content that is the source of the search is identified by the content ID of that content. Furthermore, the process of acquiring related content as a result of searching for related content based on the content that is the source of the search is equivalent to the process of acquiring a content ID that is information that identifies the related content. In other words, in the process executed by the related content search device, the process executed for one piece of content and the process executed for the content ID of that piece of content are equivalent to each other.
[0094] The knowledge information acquisition unit 301 acquires knowledge information representing various knowledge, and passes it to the content linkage processing unit 304. The knowledge information acquisition unit 301 includes a means for at least temporarily storing the knowledge information.
[0095] Specifically, the knowledge information acquisition unit 301 acquires knowledge information (knowledge graph) that is formed by associating a theme code for identifying a theme, "element words" that are words related to content related to the theme code, and "topic words" that are words related to content related to the theme code. Note that the knowledge information may include information indicating whether or not one "element word" has a direct relationship with one "topic word" for a specific theme code. An example of the configuration of the knowledge graph acquired by the knowledge information acquisition unit 301 will be described later with reference to FIG. 18 etc.
[0096] The theme data acquisition unit 302 acquires theme data (curriculum guideline data) 1200, and passes the theme data (curriculum guideline data) 1200 to the content linkage processing unit 304. The theme data acquisition unit 302 includes a means for at least temporarily storing the theme data (curriculum guideline data) 1200. The theme data (curriculum guideline data) 1200 has data items such as a theme code (curriculum guideline code), theme item text, school type, subject, etc.
[0097] The content data acquisition unit 303 acquires the content data 1150 and passes the content data 1150 to the content linkage processing unit 304. The content data acquisition unit 303 includes a means for at least temporarily storing the content data 1150. The content data 1150 includes a content ID, a content title text, a content summary text, a keyword list, a stream URL to be accessed when playing back the content, subtitle text data, and the like.
[0098] Specifically, the content data acquisition unit 303 acquires content data including content identification information for identifying content, and a keyword list including one or more keywords associated with the content identification information.
[0099] The content linkage processor 304 receives a knowledge graph (knowledge information) from the knowledge information acquisition unit 301, receives theme data (learning guideline data) from the theme data acquisition unit 302, and receives content data 1150 from the content data acquisition unit 303. Based on the received data, the content linkage processor 304 performs processing to link (link) the contents within the knowledge graph. The content linkage processor 304 passes the knowledge graph after the linkage processing to the related content search unit 306. Details of the linkage processing by the content linkage processor 304 will be described further below.
[0100] Specifically, when a keyword included in a keyword list held by content data matches a "topic word" associated with a specific theme code held by knowledge information (knowledge graph), the content linking processing unit 304 may perform processing to link the content identification information (content) held by the content data to the "topic word" associated with the theme code.
[0101] In addition, when linking one content identification information (content) to one "topic word," the content linking processing unit 304 may link the content identification information as the content identification information of "content that allows elements to be understood using the topic as an example" if the keyword list associated with the content identification information contains the "element word" that has the direct relationship with the "topic word" as a keyword, and (2) link the content identification information as the content identification information of "topic content" if the keyword list does not contain the "element word" that has the direct relationship with the "topic word" as a keyword (see also Figure 20).
[0102] The content ID acquisition unit 305 acquires a content ID and passes the content ID to the related content search unit 306. The content ID acquisition unit 305 acquires the content ID of a specific content (learning content or other content) based on, for example, the user's operation of inputting the content ID or some other operation performed by the user (for example, an operation of selecting an icon of a specific content on the screen). The content ID is information for uniquely identifying each piece of content. The content ID acquired by the content ID acquisition unit 305 is information for identifying the content that is the source of the related content search unit 306's search for related content. The content ID acquisition unit 305 passes the acquired content ID to the related content search unit 306.
[0103] The related content search unit 306 receives a content ID from the content ID acquisition unit 305. The related content search unit 306 also receives a knowledge graph after content linkage processing from the content linkage processing unit 304, and searches for content related to the content ID received from the content ID acquisition unit 305 based on the knowledge graph. The related content search unit 306 passes information about the content obtained as a search result to the presentation unit 307.
[0104] Specifically, when the related content search unit 306 acquires “original content identification information” that is the basis for searching for related content, it searches knowledge information (knowledge graph) to (A) acquiring, as related content identification information, content identification information of related content other than the “original content identification information” that has a “topic word” in the keyword list that is common to the “topic word” that is the keyword in the keyword list associated with the “original content identification information” and is associated with a specific theme code in the knowledge information, or (B) acquiring, as related content identification information, content identification information of related content other than the “original content identification information” that has in the keyword list a common “element word” with the keyword in the keyword list associated with the “original content identification information” and associated with a specific theme code in the knowledge information; The acquired related content identification information is output as a search result.
[0105] The related content search unit 306 may perform the above process (B) only when the “original content identification information” at the time of search is content identification information of “content that allows an element to be understood by taking a subject as an example” and the related content identification information is content identification information of “content that allows an element to be understood by taking a subject as an example.”
[0106] The search process performed by the related content search unit 306 will be described in detail later.
[0107] The presentation unit 307 presents the content to the user. Specifically, the presentation unit 307 displays the video of the content, which is a moving image content, on a display and outputs the audio from a speaker or an earphone, etc. The presentation unit 307 presents the content having the ID acquired by the content ID acquisition unit 305 and the content that is the search result acquired by the related content search unit 306 to the user.
[0108] That is, based on the related content identification information output by the related content search unit 306, the presentation unit 307 presents the related content identified by the related content identification information or information related to the related content.
[0109] 16 is a schematic diagram showing an example of the configuration of content data acquired by content data acquisition unit 303 and passed to content linking processing unit 304. Content data 1150 shown here has some data items common to content data 1150 shown in FIG. 4. As shown in FIG. 16, content data 1150 includes content ID 1151, title 1152, content summary 1153, keyword list 1154, stream URL 1155, and subtitle text data 1156. Here, the data combining title 1152 and content summary 1153 is information corresponding to text 1101 shown in FIG. 4. The meaning of each item of data is as follows:
[0110] The content ID·1151 is identification information for uniquely identifying each content. The title 1152 is text indicating the title (title) of the content. The content summary 1153 is text such as a sentence explaining the summary of the content. The keyword list 1154 is a list of keywords that characterize the content. The keywords included in the keyword list 1154 may be a list of words extracted from the title 1152 or the content summary 1153, or may be a list of words set separately from the title 1152 or the content summary 1153. The stream URL·1155 is information indicating a location to be accessed when playing the content. The subtitle text data 1156 is the text of subtitles that may be displayed on the screen when playing the content. The subtitle text is one of the text data that appropriately expresses the contents of the content.
[0111] FIG. 17 is a schematic diagram showing an example of theme data (learning guideline data) for a theme code (learning guideline code) to be processed by the related content search device 30. In an example of processing by the related content search device 30 described later, the theme code (learning guideline code) "8361233512200000" is processed. The illustrated theme data (learning guideline data) 1200 has this "8361233512200000" as the value of the theme code (learning guideline code) 1201. Also, the illustrated theme item text 1202 is text data including an explanatory text (a text describing a goal in learning related to the code) for the above theme code (learning guideline code) 1201. Also, the school type 1203 is "junior high school" and the subject 1204 is "science."
[0112] Fig. 18 is a schematic diagram showing an example of a knowledge graph (knowledge information) generated by the above-mentioned knowledge information generation device 20 for the theme code (curriculum guideline code) "8361233512200000." The knowledge information acquisition unit 301 acquires knowledge graph data as exemplified in Fig. 18 from the outside (for example, from the knowledge information generation device 20).
[0113] The knowledge graph shown in FIG. 18 has a total of seven nodes. Of these nodes, there is one node that belongs to the hierarchy of theme codes (learning guidelines codes), which is "8361233512200000". There are two nodes that belong to the hierarchy of elements (learning elements), which are "force" and "motion". There are links from the above node "8361233512200000" to the nodes "force" and "motion" that belong to the hierarchy of elements (learning elements). There are four nodes that belong to the hierarchy of subjects, which are "curling", "cart", "ball", and "car". There are links from the above node "force" to the nodes "curling" and "cart" that belong to the hierarchy of subjects. There are links from the above node "motion" to the nodes "ball" and "cart" that belong to the hierarchy of subjects. The relationship between elements (learning elements) and subjects is as follows: That is, a link connecting an element (learning element) node and a topic node indicates that the topic can be cited as a familiar example when learning the element (learning element).
[0114] As illustrated, the knowledge graph of this embodiment is graph data generated for each theme code (learning guideline code). In other words, in the knowledge graph, words (concepts) are clustered (grouped) for each theme code (learning guideline code). In the graph, links between nodes corresponding to concepts represent the association between the two concepts. The related content search device 30 utilizes the graph data representing the above-mentioned know-how. In other words, the content linking processor 304 can link not only educational content but also all kinds of content from the viewpoint of learning. The related content search device 30 uses information linked by the content linking processor 304 to search for other content (learning content or other content) related to a certain content, and makes it possible to present the content to the user.
[0115] The knowledge information (knowledge graph) regarding a specific theme code includes information indicating whether or not one "element word" has a direct relationship with one "topic word." Specifically, the presence or absence of a direct link between a node belonging to the element hierarchy and a node belonging to the topic hierarchy is the above-mentioned "information indicating whether or not there is a direct relationship."
[0116] [Details of processing by the content linkage processing unit 304] Here, the processing of the content linkage processor 304 will be described in further detail mainly with reference to FIGS.
[0117] The content linking processor 304 links the content to the knowledge graphs based on the content data passed from the content data acquisition unit 303 and the multiple knowledge graphs (knowledge graphs for each theme code (learning course code)) passed from the knowledge information acquisition unit 301. In order to estimate which theme code (learning course code) in the multiple clustered knowledge graphs corresponds to the content to be processed, the content linking processor 304 identifies the theme code (learning course code) with the largest number of matching words for the keyword group (keyword list 1154 shown in FIG. 16) in the content data of the content, and acquires the cluster of that theme code (learning course code).
[0118] In other words, the "cluster" referred to here corresponds to a group of words in a knowledge graph relating to one theme code (curriculum guideline code).
[0119] If the keywords included in keyword list 1154 of content data 1150 for a certain piece of content contain a word that matches an element (learning element) in the knowledge graph of the acquired cluster (i.e., the identified theme code (learning guideline code)) and a word that matches a subject in the knowledge graph of that cluster, the content linking processing unit 304 links the content to the knowledge graph of that cluster as "content that allows you to learn the element (learning element) using the subject as an example" (in other words, "content that allows you to understand the element using the subject as an example").
[0120] On the other hand, if the keywords included in the keyword list 1154 of the content data 1150 for that content do not contain any words that match the elements (learning elements) in the knowledge graph of that cluster (however, even in this case, the keywords contain words that match the subject matter in the knowledge graph of that cluster), the content linking processing unit 304 links the content to the knowledge graph of that cluster as "subject matter content."
[0121] FIG. 19 is a decision table showing how to link content to a knowledge graph. In the figure, "content A" is the content to be linked to the knowledge graph. The keyword list for content A is keyword list 1154 in content data 1150 related to content A. Also, "theme code (curriculum guideline code) B" is the target to which content A is linked. In other words, the content A is to be linked to the knowledge graph of theme code (curriculum guideline code) B.
[0122] The first proposition represented by the decision table in Figure 19 is "There is something in the keyword list of content A that matches a 'element word' in the knowledge graph of theme code (curriculum guideline code) B" (vertical axis), and the way in which the content is linked to the knowledge graph changes depending on whether this is true or false. The second proposition is "There is something in the keyword list of content A that matches a 'subject word' in the knowledge graph of theme code (curriculum guideline code) B" (horizontal axis), and the way in which the content is linked to the knowledge graph also changes depending on whether this proposition is true or false. There are four cases depending on the truth of the first and second propositions above. For the sake of convenience, these are represented as the following categories: <Category 11> (first proposition is true and second proposition is true), <Category 10> (first proposition is true and second proposition is false), <Category 01> (first proposition is false and second proposition is true), <Category 00> (first proposition is false and second proposition is false). Note that when linking content A to a knowledge graph, it is assumed that at least a part of the keyword list that content A has exists in at least one of the "element words (words that are learning elements)" or "subject words" in the knowledge graph of theme code (curriculum guideline code) B. If this assumption is made, the above <Category 00> is an impossible category.
[0123] In the case of the above <Classification 11>, the content linking processor 304 links content A to the node of "topic word" in the knowledge graph of theme code (curriculum guideline code) B as "content that allows learning elements (learning elements) using the topic as an example." In the case of the above <Classification 10>, the content linking processor 304 does not link content A to a node in the knowledge graph of theme code (curriculum guideline code) B. In the case of the above <Classification 01>, the content linking processor 304 links content A to the node of "topic word" in the knowledge graph of theme code (curriculum guideline code) B as "topic content." Note that <Classification 00> is not possible as described above, so it does not need to be considered.
[0124] FIG. 20 is a schematic diagram showing an example of content linked (linked) by the content linking processor 304 to a knowledge graph corresponding to a theme code (learning guideline code) of "8361233512200000". The knowledge graph itself in FIG. 20 is the same as the knowledge graph described in FIG. 18. That is, this knowledge graph has a node "8361233512200000" in the theme code hierarchy, and has nodes "force" and "motion" in the element (learning element) hierarchy below that. In addition, this knowledge graph has nodes "curling" and "cart" linked to the node "force" in the subject hierarchy. In addition, this knowledge graph has nodes "ball" and "cart" linked to the node "motion" in the subject hierarchy.
[0125] As shown in FIG. 20, contents 1401 and 1411 are linked to a node "curling" in the hierarchy of the subject matter. Contents 1402 and 1412 are linked to a node "truck" in the hierarchy of the subject matter. Contents 1403 and 1413 are linked to a node "ball" in the hierarchy of the subject matter. Contents 1404 and 1414 are linked to a node "car" in the hierarchy of the subject matter. Keywords (words stored in the keyword list 1154 shown in FIG. 16, e.g., nouns; the same applies below) related to content 1411 include the word "curling" (subject matter) and do not include the word "force" (element (learning element)). Keywords related to content 1401 include the word "curling" (subject matter) and the word "force" (element (learning element)). Keywords related to content 1412 include the word "truck" (subject matter) and do not include the word "force" (element (learning element)). Keywords for content 1402 include the word "cart" (subject) and the word "force" (element (learning element)). Keywords for content 1413 include the word "ball" (subject) and do not include the word "motion" (element (learning element)). Keywords for content 1403 include the word "ball" (subject) and the word "motion" (element (learning element)). Keywords for content 1414 include the word "car" (subject) and do not include the word "motion" (element (learning element)). Keywords for content 1404 include the word "car" (subject) and the word "motion" (element (learning element)).
[0126] In other words, in FIG. 20, the contents 1411, 1412, 1413, and 1414 linked by the content linking processor 304 are "contents serving as subjects" that contain words (nouns, etc.) serving as subjects as their respective keywords, but do not contain words (nouns, etc.) serving as elements (learning elements). Also, the contents 1401, 1402, 1403, and 1404 linked by the content linking processor 304 are "contents for learning elements (learning elements) using the subject as an example" that contain both words (nouns, etc.) serving as subjects and words (nouns, etc.) serving as elements (learning elements) as their respective keywords. The number of "contents serving as subjects" linked to one subject node is an arbitrary number equal to or greater than 0. Also, the number of "contents for learning elements (learning elements) using the subject as an example" linked to one subject node is also an arbitrary number equal to or greater than 0.
[0127] [Details of processing by related content search unit 306] Here, the process of searching for related content by the related content search unit 306 will be described in further detail mainly with reference to FIGS.
[0128] The related content search unit 306 receives information on the content linked to the knowledge graph from the content linking processor 304. The related content search unit 306 also receives a content ID (e.g., an ID of a content specified by a user) that is the basis of the search from the content ID acquisition unit 305. The related content search unit 306 then searches the knowledge graph to which the content is linked, to determine whether the content identified by the content ID passed from the content ID acquisition unit 305 is "content to be used as a subject" or "content for learning elements (learning elements) using the subject as an example." The "content to be used as a subject" refers to the contents 1411, 1412, 1413, and 1414 shown in FIG. 20, etc. The "content for learning elements (learning elements) using the subject as an example" refers to the contents 1401, 1402, 1403, and 1404 shown in FIG. 20, etc.
[0129] Fig. 21 is a schematic diagram showing a first pattern of search by related content search unit 306. The first pattern is a pattern when the content specified by content ID acquisition unit 305 (here referred to as "original content"; in the example shown in Fig. 21, content 1501) is determined to be "subject content", and related content search unit 306 acquires the following two types of content as related content related to the content (content 1501). Note that for each of these two types, related content search unit 306 outputs any number of related contents equal to or greater than 0 as search results.
[0130] The first type is "topic content" (content 1511 in the example shown in FIG. 21) that has as its topic a word that is the same as a keyword in the original content (content 1501). In other words, it is another "topic content" related to the same topic as the original content (content 1501). The related content search unit 306 can search for related content on a topic that interests the user by searching for this first type of content as related content.
[0131] The second type is "content that allows learning elements (learning elements) using a topic as an example" (content 1512 in the example shown in FIG. 21) that has as its topic the same word that the original content (content 1501) has as a keyword. In other words, it is content that allows learning of an element (learning element) based on the same topic as the original content (content 1501). By searching for this second type of content as related content, the related content search unit 306 can provide a natural flow from a topic that interests the user to viewing content that allows learning elements.
[0132] FIG. 22 is a schematic diagram showing a second pattern of search by the related content search unit 306. The second pattern is a pattern when it is determined that the content (herein called "original content"; in the example shown in FIG. 22, content 1521) instructed by the content ID acquisition unit 305 is "content that can learn an element (learning element) using a subject as an example". In the example shown in FIG. 22, the keywords of the content 1521, which is the original content, are "curling" and "force". Among these, the keyword "curling" corresponds to a node belonging to the subject hierarchy in the knowledge graph to which the content is linked. Also, the keyword "force" corresponds to a node belonging to the element hierarchy in the knowledge graph to which the content is linked. The related content search unit 306 acquires the following two types of content as related content related to the content (content 1521). Note that the related content search unit 306 outputs an arbitrary number of related contents, 0 or more, for each of these two types as a search result.
[0133] The first type is "subject content" (content 1531 in the example shown in FIG. 22) that has as its subject a word that is the same as a keyword that is the subject of the original content (content 1521) but does not have any element (learning element). In other words, it is "subject content" that has a common subject with the original content (content 1521). The related content search unit 306 searches for this first type of content as related content, thereby enabling the user to learn while relating the elements (learning elements) that they are learning through the original content (content 1521) to the real world.
[0134] The second type is "content that allows learning an element (learning element) using a subject as an example" (content 1532 in the example shown in FIG. 22) that has as an element (learning element) the same word as the element (learning element) that the original content (content 1521) has as a keyword, and has as a subject a different word that the original content (content 1521) has as a keyword. In other words, it is content that allows learning the same element (learning element) as the original content (content 1532) using a different subject as an example from the original content (content 1532). The related content search unit 306 searches for this second type of content as related content, so that the user can learn the element (learning element) that he or she is currently learning using a different subject as an example.
[0135] The related content search unit 306 may perform the above-mentioned process of "(B) acquiring, as related content identification information, content identification information of related content other than the "original content identification information" that has in the keyword list a common "element word" with the "element word" that is the keyword in the keyword list associated with the "original content identification information" and is associated with a specific theme code in the knowledge information" only when the "original content identification information" at the time of searching is content identification information of "content that allows elements to be understood by taking a subject as an example" and the related content identification information is content identification information of "content that allows elements to be understood by taking a subject as an example." In other words, the reason why content 1521 is used as the source and content 1532, which is related content, is output as a search result is because both content 1521 and content 1532 are "content that allows elements to be understood by taking a subject as an example."
[0136] Next, the procedure of the process performed by the related content search device 30 will be described with reference to flowcharts (FIGS. 23 and 24).
[0137] 23 is a flowchart showing the procedure of processing for linking contents to a knowledge graph by the content linking processing unit 304 of the related content search device 30. The processing procedure will be described below with reference to this flowchart.
[0138] In step S101, the content data acquisition unit 303 acquires content data. The content data acquired by the content data acquisition unit 303 may be, for example, content data corresponding to all the contents to be linked to the knowledge graph. The content data acquisition unit 303 passes the acquired content data to the content linkage processing unit 304.
[0139] Next, in step S102, the knowledge information acquisition unit 301 acquires data of a knowledge graph (knowledge information). This knowledge graph may be generated by the knowledge information generation device 20 (FIG. 3) already described. The knowledge information acquisition unit 301 passes the acquired knowledge graph to the content linkage processing unit 304.
[0140] Next, in step S103, the content cooperation processor 304 starts repeated processing on the content transferred in step S101. That is, the content cooperation processor 304 performs the processing of each step included in steps S104 to S108 on each piece of content.
[0141] In step S104, the content linkage processor 304 identifies the theme code (curriculum guideline code) that matches the largest number of words with respect to the group of keywords (keyword list 1154 in FIG. 16) held by the content data of the content currently being processed. In other words, the content linkage processor 304 identifies the cluster (theme code (curriculum guideline code)) that matches the content currently being processed the most. Then, the content linkage processor 304 sets the knowledge graph of this cluster (theme code (curriculum guideline code)) as the processing target in the following steps S105, S106, S107, and S108. In other words, the knowledge graph of the theme code (curriculum guideline code) identified here is the destination to which the content currently being processed is linked.
[0142] In step S105, the content linking processor 304 determines whether or not there is a word that matches any of the keywords in the content currently being processed among the nouns (nouns that are learning elements) that are elements of the knowledge graph of the theme code (learning guideline code). If there is (step S105: YES), the process proceeds to step S106. If there is not (step S105: NO), the process proceeds to step S108.
[0143] Note that the above "Step S105: YES" is based on the classification explained in FIG. <11> or classification <10> In addition, "Step S105: NO" corresponds to the case where the classification described in FIG. <01> This corresponds to the case where, as mentioned above, the classification shown in Fig. <00> This is because this is not possible (in fact, the flowchart in FIG. 23 does not include any processing for such a case).
[0144] If the process proceeds to step S106, in this step the content linking processor 304 determines whether or not there is a word that matches any of the keywords in the content currently being processed among the nouns that are the subject of the knowledge graph of the theme code (curriculum guideline code). If there is (step S106: YES), the process proceeds to step S107. If there is not (step S106: NO), the process for the content starting from step S104 ends (in this case, the content is not linked to the knowledge graph) and the process moves to the next content (if there is).
[0145] Note that the above "Step S106: YES" is based on the classification explained in FIG. <11> In addition, "Step S106: NO" corresponds to the case of the classification explained in FIG. <10> This corresponds to the case where
[0146] When the process proceeds to step S107, in this step the content linking processor 304 links the currently processed content to the knowledge graph as "content that can learn elements (learning elements) using the subject as an example". At this time, the content linking processor 304 links the content to the relevant subject (subject that was a "matching word" in the determination process of step S106) in the knowledge graph. For example, the contents 1401, 1402, 1403, and 1404 shown in FIG. 20 are contents linked to the node of the subject as "content that can learn elements (learning elements) using the subject as an example". When the process of step S107 ends, the process for the content that starts from step S104 ends, and the process moves to the next content (if there is one).
[0147] When the process proceeds to step S108, in this step the content linking processor 304 links the content currently being processed to the knowledge graph as "content to be the subject". At this time, the content linking processor 304 links the content to the relevant subject in the knowledge graph (the subject that matches the keyword of the content with the subject word in the knowledge graph of the theme code (course of study code)). For example, the contents 1411, 1412, 1213, and 1414 shown in FIG. 20 are contents linked to the node of the subject as "content to be the subject". When the process of step S108 ends, the process for the content that starts from step S104 ends, and the process moves to the next content (if there is one).
[0148] When the processes from steps S104 to S108 are completed for all the contents, the process leaves this repetitive process and proceeds to step S109.
[0149] In step S109, the content linkage processor 304 outputs the content information linked to the knowledge graph. That is, the content linkage processor 304 outputs the knowledge graph in which the content is linked. The knowledge graph output by the content linkage processor 304 is passed to the related content search unit 306.
[0150] 24 is a flowchart showing the overall processing procedure of the related content search device 30. In other words, the processing of this flowchart shows the procedure of the related content search unit 306 searching for related content, etc., on the premise that the content linkage processing unit 304 performs processing to link content with a knowledge graph. The following will be explained with reference to this flowchart.
[0151] In step S201, the related content search device 30 determines whether or not content has been linked, that is, whether or not the content linking processing unit 304 has already completed processing to link content to a knowledge graph. If content has been linked (step S201: YES), the process jumps to step S203. If content linking has not been completed (step S201: NO), the process proceeds to step S202 to perform content linking processing.
[0152] When the process proceeds to step S202, in this step, the related content search device 30 performs a content linking process. The procedure of the content linking process has already been described with reference to Fig. 23. After the completion of this step, the process proceeds to step S203.
[0153] In step S203, the content ID acquisition unit 305 acquires a content ID. This content ID may be, for example, input by the user as a character string or selected via a graphical user interface. The content ID acquired by the content ID acquisition unit 305 is information for identifying content that is the source of a search for related content. The content ID acquisition unit 305 passes the acquired content ID to the related content search unit 306.
[0154] In step S204, the related content search unit 306 determines whether the ID (ID of the original content) acquired in step S203 is an ID of "content to be used as a subject." Specifically, the related content search unit 306 searches the knowledge graphs of all theme codes (learning guideline codes) to determine whether the content of the content ID is linked to the knowledge graph as "content to be used as a subject" or as "content for learning learning elements using the subject as an example." These two types of connections have already been described with reference to FIG. 20, etc.
[0155] If the original content is the ID of the "content to be used as the subject" (step S204: YES), proceed to step S205. If the original content is not the ID of the "content to be used as the subject" but is the ID of the "content for learning elements (learning elements) using the subject as an example" (step S204: NO), proceed to step S206.
[0156] When the process proceeds to step S205, in this step, the related content search unit 306 searches for a knowledge graph in which content has been linked, 1) The content IDs of all "subject content" related to the same subject as the original content, 2) The content IDs of all "content that uses the subject as an example to learn elements (learning elements)" related to the subject of the original content, That is, the related content search unit 306 acquires the content 1511 and the content 1512 based on the content 1501 shown as an example in Fig. 21. After the process of this step ends, the process proceeds to step S207.
[0157] When the process proceeds to step S206, in this step, the related content search unit 306 searches for a knowledge graph in which content has been linked, 1) The content IDs of all "subject content" related to the same subject as the original content, 2) The content IDs of all "content that can learn elements (learning elements) using the subject as an example" that has the same elements (learning elements) as the original content, but has a subject different from the subject of the original content, That is, the related content search unit 306 acquires the content 1531 and the content 1532 based on the content 1521 shown as an example in Fig. 22. After the process of this step ends, the process proceeds to step S207.
[0158] In step S207, the related content search unit 306 outputs the content IDs of all the related contents acquired in either step S205 or S206. The related content search unit 306 passes the content IDs of these related contents to, for example, the presentation unit 307. This allows the related content search unit 306 to pass the content IDs of all the related contents related to the "original content" initially selected by the user to the outside. When the presentation unit 307 receives the content IDs of these related contents, the presentation unit 307 can present (for example, have the user view or recommend) the related contents related to the content initially selected (for example, viewed) by the user to the user. The relationship between the original content and the related contents is as described with reference to FIG. 21 and FIG. 22.
[0159] That is, by the procedure described above, the related content search unit 306 searches for related content while referring to the knowledge graph that has been subjected to the content linkage process, and thereby it is possible to obtain content on a subject related to the original content, content that has the same elements as the elements of the original content but is explained using a different subject, and the like. When applied to the field of learning content, the related content search unit 306 can obtain content on a subject related to the original content, content that has the same elements (learning) as the elements of the original content but is explained using a different subject, and the like. That is, the user can deepen his / her understanding of learning elements not only by relying on learning content but also by content other than learning content. That is, the related content search device 30 can present content that is not produced for educational purposes (for example, news, sports, entertainment, etc.) to the user by linking it to educational content.
[0160] Next, some modified examples of the present embodiment will be described. Note that multiple modified examples may be combined as long as they are possible to be combined.
[0161] [Variation 1 Knowledge Information] In the above embodiment, a knowledge graph having a graph structure is used as the knowledge information. In order to realize a function equivalent to that of the above embodiment, knowledge information other than a graph structure may be used. In this case, the knowledge information is information that represents the relationship between a theme code (curriculum guideline code) and a word (noun, etc.) that represents an element (learning element) for the theme code (curriculum guideline code) and a word (noun, etc.) that represents a subject. In the existing technology, there are many forms of data expression that represent a relationship in a manner other than a graph structure. In this modified example, the content linking processor 304 also performs a process of linking the content to the knowledge information. In this case, the content linking processor 304 may also link the content to a "word that represents a subject" in the knowledge corresponding to a specific theme code (curriculum guideline code).
[0162] [Variation 2: Information system to which it applies] In the above description of the embodiment, the method of the embodiment is applied to the curriculum guidelines. As a modified example, the method of the embodiment may be applied to an information system other than the curriculum guidelines.
[0163] One example outside of the curriculum guidelines is the Nippon Decimal Classification (NDC) for books. Books are classified according to the Nippon Decimal Classification. When applied to the Nippon Decimal Classification, knowledge utilization system 1 generates knowledge information based on summaries and titles that correspond to three-digit decimal classification codes (equivalent to theme codes, although there are also codes with four or more digits). Furthermore, knowledge utilization system 1 can use the generated knowledge information to search for related content that is related to a certain content and present it to the user. The content in this case is books, etc. Furthermore, the text of the summary (explanation) of the book, etc. is included in the content data.
[0164] Another example other than the curriculum guidelines is the International Patent Classification (IPC) for patents. Patents are classified according to the International Patent Classification. However, one patent may be associated with multiple International Patent Classification codes. In this case, the knowledge utilization system 1 generates knowledge information based on the abstract, title, etc. corresponding to the International Patent Classification code. Furthermore, the knowledge utilization system 1 can use the generated knowledge information to search for related content related to a certain content and present it to the user. The content in this case is a patent document (e.g., a patent gazette or a published patent gazette), etc. Furthermore, the text of the abstract (e.g., an abstract) for each patent is included in the content data.
[0165] [Realization by computers and programs] FIG. 25 is a block diagram showing an example of the internal configuration of the knowledge information generating device 20 and the related content searching device 30, which are elements of the knowledge use system 1 in the above embodiment (including the modified example). Each of the knowledge information generating device 20 and the related content searching device 30 can be realized using a computer. As shown in the figure, the computer is configured to include a central processing unit 901, a RAM 902, an input / output port 903, input / output devices 904 and 905, etc., and a bus 906. The computer itself can be realized using existing technology. The central processing unit 901 executes instructions included in a program read from the RAM 902, etc. The central processing unit 901 writes data to the RAM 902, reads data from the RAM 902, and performs arithmetic operations and logical operations according to each instruction. The RAM 902 stores data and programs. Each element included in the RAM 902 has an address and can be accessed using the address. It should be noted that RAM is an abbreviation for "random access memory." The input / output port 903 is a port through which the central processing unit 901 exchanges data with external input / output devices, etc. The input / output devices 904 and 905 exchange data with the central processing unit 901 via the input / output port 903. The bus 906 is a common communication path used within the computer. For example, the central processing unit 901 reads and writes data in the RAM 902 via the bus 906. Also, for example, the central processing unit 901 accesses the input / output port 903 via the bus 906.
[0166] At least some of the functions of the knowledge information generating device 20 and the related content searching device 30 in the above-mentioned embodiment can be realized by a computer and a program. In that case, the functions may be realized by recording a program for realizing the functions in a computer-readable recording medium, and reading and executing the program recorded in the recording medium into a computer system. The term "computer system" here includes hardware such as an OS and peripheral devices. The term "computer-readable recording medium" refers to a portable medium such as a flexible disk, an optical magnetic disk, a ROM, a CD-ROM, a DVD-ROM, a USB memory, and a storage device such as a hard disk built into a computer system. In other words, the term "computer-readable recording medium" may be a non-transitory computer-readable recording medium. The term "computer-readable recording medium" may also include a medium that temporarily and dynamically holds a program, such as a communication line when a program is transmitted via a network such as the Internet or a communication line such as a telephone line, and a medium that holds a program for a certain period of time, such as a volatile memory inside a computer system that is a server or client in that case. The above-mentioned program may be a program for realizing some of the above-mentioned functions, or may be a program that can be realized in combination with a program already recorded in the computer system.
[0167] The embodiment (including modifications) has been described above with reference to the drawings. The specific configuration for carrying out the invention is not limited to this embodiment (including modifications), and includes designs within the scope of the gist of the invention.
[0168] According to the above embodiment (including the modified example), the knowledge information generating device 20 can automatically extract element words and subject words based on the content data and theme data, and can generate knowledge information by associating them. The knowledge information is information that can be referred to when searching for related content, for example, and represents the relationship between elements and subjects for each theme. The knowledge information is formatted and is in a form that can be used by a computer or the like. The related content search device 30 performs a process of linking the content with the knowledge information. The related content search device 30 can search for related content related to a given piece of content (identified by a content ID) by using the knowledge information in which the content is linked. The related content search device 30 can present information on the related content obtained as a search result to the user, or present the related content itself to the user.
[0169] Taking the field of education as an example, this embodiment makes it possible to link content produced for educational purposes with content produced for purposes other than education (e.g., news reporting, entertainment, etc.) as related content and present it to the user. [Industrial Applicability]
[0170] The present invention can be used, for example, for managing and using knowledge related to content, but the scope of use of the present invention is not limited to the examples given here. [Explanation of symbols]
[0171] 1 Knowledge-based systems 20 Knowledge information generation device 30 Related content search device 201 Content Data Acquisition Department 202 Theme Data Acquisition Department 203 Element and Subject Extraction Section 204 Structured Processing Unit 301 Knowledge Information Acquisition Department 302 Theme Data Acquisition Department 303 Content Data Acquisition Unit 304 Content Linkage Processing Unit 305 Content ID Acquisition Unit 306 Related Content Search Section 307 Presentation section 901 Central Processing Unit 902 RAM 903 Input / Output Ports 904,905 Input / Output Devices 906 Bus 1100 Content data (learning content data) 1101 Text (content text information) 1150 Content data (learning content data) 1200 Theme Data (Curriculum Guideline Data) 1202 Theme item text (theme text information)
Claims
1. a theme data acquiring unit that acquires theme data including a pair of a theme code for identifying a theme and theme text information associated with the theme code; a content data acquisition unit that acquires content data including a pair of content identification information for identifying a content and content text information associated with the content identification information; an element and theme extraction unit that determines "element words" from among the words extracted from the content text information based on whether or not the words are included in the theme text information, and also determines a score related to importance for each of the words extracted from the content text information and determines "theme words" from among the words extracted from the content text information based on the score; a structuring processing unit that generates knowledge information by associating the theme code with the "element words" and the "topic words" extracted from the content text information in relation to the theme code; A knowledge information generating device comprising:
2. the element and subject extraction unit determines a predetermined number of words having top importance scores among the words extracted from the content text information as the "subject words"; 2. The knowledge information generating device according to claim 1.
3. The element and subject extraction unit determines a score on the importance of each of the extracted words from the content text information by using a PositionRank method or a TF-IDF (Term Frequency and Inverse Document Frequency) method.
3. The knowledge information generating device according to claim 2.
4. the element and subject extraction unit determines, among the words extracted from the content text information, that the words having the top predetermined importance scores and that have been determined to be the "element words", to be excluded from the "subject words"; 3. The knowledge information generating device according to claim 2.
5. The structuring processing unit includes: A node corresponding to the theme code; A node corresponding to the "element word"; a node corresponding to the "topic word"; a first hierarchical link provided between a node corresponding to the theme code and a node corresponding to the "element word" related to the node corresponding to the theme code; a second hierarchical link provided between a node corresponding to the "element word" and a node corresponding to the "topic word" related to the node corresponding to the "element word"; The knowledge information is generated as graph structure data having the following structure:
2. The knowledge information generating device according to claim 1.
6. a theme data acquiring unit that acquires theme data including a pair of a theme code for identifying a theme and theme text information associated with the theme code; a content data acquisition unit that acquires content data including a pair of content identification information for identifying a content and content text information associated with the content identification information; an element and theme extraction unit that determines "element words" from among the words extracted from the content text information based on whether or not the words are included in the theme text information, and also determines a score related to importance for each of the words extracted from the content text information and determines "theme words" from among the words extracted from the content text information based on the score; a structuring processing unit that generates knowledge information by associating the theme code with the "element words" and the "topic words" extracted from the content text information in relation to the theme code; A program for causing a program to function as a knowledge information generation device comprising the above.
Citation Information
Patent Citations
Educational teaching material navigation system
JP2015018159A
