JP6056563B2 - Example part specifying device, example part specifying method, and example part specifying program - Google Patents

Description

  The present invention relates to a technical field such as an apparatus and method for generating a summary of a document and organizing a main point.

  Conventionally, software for generating document summaries and organizing key points has been developed. For example, in Patent Document 1, information such as the position of a summary constituent unit (a unit that minimizes a section including a predicate), a clue word, a title, the importance of a word, and a keyword are used as clues in the text. Calculates the importance of each summary component in, extracts from the summary component units with the highest importance in order based on the importance for each summary component until the summary length requested by the user is reached. A technique is disclosed in which the summary constituent units extracted by the output means are arranged in the order of appearance in the original text and output as a summary result. In addition to such a technique, for example, Patent Document 2 discloses a technique for dividing text into classified discourse constituent elements and combining the discourse constituent elements with the structural representation of the discourse.

JP 2009-146447 A Japanese Patent No. 4706227

  By the way, among the constituent parts (constituent elements) constituting the document, the exemplified part is particularly effective for knowledge organization and memory fixing. This is because, for example, it is considered that the author is trying to convey the contents of the document to the reader in an easy-to-understand manner using examples. As a method for identifying an example portion from a document, it is conceivable to use a clue word (such as “for example”). However, when a clue word is not included in the document, this method cannot identify an example portion from the document. In addition, as a method for identifying an example part from a document, when a specific item such as an event or state described in the sentence A is presented in the sentence B in a group of a plurality of sentences, the sentence B may be used as the example part. Conceivable. However, in this method, only the relationship between sentences is taken into consideration, and therefore the example part cannot be specified when the example part appears independently.

  Therefore, the present invention has been made in view of the above points and the like, and even when a clue word such as a conjunction is not present in a sentence or when a connection relation between sentences cannot be obtained, an example portion is specified. It is an object of the present invention to provide an example part specifying device, an example part specifying method, and an example part specifying program that can be performed.

  In order to solve the above-mentioned problem, the invention according to claim 1 is an exemplary partial specifying device capable of referring to a broader term database that registers a broader term and a narrower term of a plurality of events or things in association with each other. The document acquisition means for acquiring the document data from a predetermined storage means, the subject extraction means for extracting the subject of the document from the document related to the document data acquired by the document acquisition means, and the broader term database Then, an example of extracting a constituent part of a document including words registered in the higher-level and lower-word database from the document related to the document data acquired by the document acquisition unit as an example part candidate of the document The number of words contained in the partial candidate extraction unit and the exemplary partial candidate extracted by the exemplary partial candidate extraction unit is set as the first appearance number, and the extracted by the subject extraction unit A ratio calculating means for calculating a ratio of the second number of occurrences to the first number of occurrences when the number of at least one of the subject and the broader term of the subject is the second number of appearances; And an example portion specifying unit for specifying an example portion of the document from the example portion candidates extracted by the example portion candidate extracting unit based on the ratio calculated by the ratio calculating unit.

  According to a second aspect of the present invention, in the exemplary part specifying apparatus according to the first aspect, the subject of the document extracted by the subject extraction unit is limited to words registered in the broader term database. Features.

  According to a third aspect of the present invention, there is provided the example part specifying device according to the first or second aspect, wherein the example part specified by the example part specifying unit or a summary of the example part is displayed on a predetermined display unit. It further comprises display data output means for outputting display data.

  According to a fourth aspect of the present invention, in the exemplary part specifying apparatus according to the third aspect, the display data is a component part of the document related to the document data acquired by the document acquisition unit or a summary of the component part. The component part related to the example part specified by the example part specifying means or the summary of the component part is data for displaying in association with the example part or the summary of the example part. .

  According to a fifth aspect of the present invention, in the exemplary portion specifying apparatus according to the fourth aspect, the display data is an instruction from a user to display whether or not the exemplary portion specified by the exemplary portion specifying means or a summary of the exemplary portion is to be displayed. The data is switchable according to the above.

  The invention according to claim 6 is an exemplary portion specifying method executed by a computer capable of referring to a broader term database that registers and registers the broader terms and narrower terms of a plurality of events or things. A document acquisition step for acquiring a document from a predetermined storage means, a subject extraction step for extracting a subject of a document from documents related to the document data acquired by the document acquisition step, and referring to the broader term database, Example part candidate extraction that extracts a constituent part of a document that includes a word registered in the broader term database from documents related to the document data acquired by the document acquisition step as an example part candidate of the document The number of words included in the example portion candidates extracted in the step and the example portion candidate extraction step is the first occurrence number, When the number of at least one of the subject extracted in the extracting step and the broader term of the subject is set as the second number of appearances, the ratio of the second number of appearances to the first number of appearances is calculated. A ratio calculating step; and an example portion specifying step for specifying an example portion of the document from the example portion candidates extracted by the example portion candidate extraction step based on the ratio calculated by the ratio calculation step. It is characterized by that.

  The invention according to claim 7 is a document acquisition for acquiring document data from a predetermined storage means for a computer capable of referring to a broader term database that registers and registers the broader terms and narrower terms of a plurality of events or things. Means for extracting the subject of the document from the document related to the document data acquired by the document acquisition means, and relates to the document data acquired by the document acquisition means with reference to the broader term database Extracted from the document by the example part candidate extracting means for extracting the constituent part of the document containing the word registered in the broader term database as the example part candidate of the document, extracted by the example part candidate extracting means The number of words included in the example portion candidate is the first occurrence number, and at least one of the subject extracted by the subject extraction means and the broader term of the subject And the ratio calculation means for calculating the ratio of the second number of appearances to the first number of appearances, and the ratio calculated by the ratio calculation means, , And functioning as an example part specifying means for specifying an example part of a document from the example part candidates extracted by the example part candidate extracting unit.

  According to the present invention, even when a clue word such as a conjunction is not present in a sentence, or even when a connection relationship between sentences cannot be acquired, an exemplary portion can be specified.

It is a figure which shows the example of a schematic structure of the document information provision system which concerns on this embodiment. 5 is a flowchart illustrating an example of an exemplary part specifying process in a control unit 23. It is a figure which shows the example of the example e of each of the example partial candidates 1-3. (A) is a figure which shows an example of the document content browsing screen displayed on the display of the user terminal 1. FIG. (B) is a figure which shows an example of the important point organization screen displayed on the display of the user terminal 1. FIG. (A) is a figure which shows an example of the document content browsing screen displayed on the display of the user terminal 1. FIG. (B) is a figure which shows an example of the important point organization screen displayed on the display of the user terminal 1. FIG. It is a figure which shows the example which a user sets a title contribution. It is a figure which shows the example which a user sets the threshold value of appearance frequency. It is a figure which shows the example which the user sets threshold value (theta) of the example degree e.

  DESCRIPTION OF EMBODIMENTS Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings. The embodiment described below is an embodiment when the present invention is applied to a document information providing system.

  First, the configuration and function of the document information providing system according to the present embodiment will be described with reference to FIG. FIG. 1 is a diagram illustrating a schematic configuration example of a document information providing system according to the present embodiment. As shown in FIG. 1, the document information providing system S includes a user terminal 1, a document information providing server 2, and the like. Examples of the user terminal 1 include a tablet, a mobile phone, a smartphone, and a personal computer. The user terminal 1 can communicate by accessing the document information providing server 2 via the network NW. The network NW includes, for example, the Internet, a dedicated communication line (for example, CATV (Community Antenna Television) line), a mobile communication network (including a base station), a gateway, and the like.

  The document information providing server 2 includes a communication unit 21, a storage unit 22 (an example of a storage unit), a control unit 23, and the like. The communication unit 21 is connected to the network NW and controls the communication state with the user terminal 1. The storage unit 22 is configured by, for example, a hard disk drive or the like, and stores an operating system and an exemplary part specifying program of the present invention. The example portion specifying program may be distributed from a predetermined server or the like via the network NW, or may be recorded on a recording medium such as a CD (Compact Disc) or a DVD (Digital Versatile Disc). It may be provided.

  The storage unit 22 stores a plurality of document data that can be provided to the user terminal 1. The document data is electronic data of documents such as books and magazines. Each document data is given a unique document ID (identifier) for each document data. In addition, an element ID, a summary (main point), and a semantic tag are assigned to each component (component) of a document (document represented by document data) related to document data, and these are associated with the document ID. Is remembered. Here, the component part of a document is a part which comprises a document, for example, the sentence in a document corresponds. The element ID is a unique identifier for each component. The summary is, for example, a sentence obtained by shortening the constituent part by a known method or a characteristic word (for example, a noun) included in the constituent part. The meaning tag is a tag indicating the type of the component part, and includes, for example, “conventional”, “current state”, “example”, and the like. Further, in the storage unit 22, data indicating the connection relationship between the constituent parts of the document (for example, data in which element IDs of the constituent parts related to each other are associated) is associated with the document ID of the document. Remembered. For example, the connection relationship between the component part “conventionally distributed by paper ...” and the component part “currently digital ...” is “contrast”. Note that the summary of the component “conventionally distributed by paper ...” is “distributed by paper”, and its meaning tag is “conventional”. On the other hand, the summary of the component “currently digital ...” is “digital distribution”, and the meaning tag is “current”.

In the storage unit 22, a broader term database 221 is constructed. The broader term database 221 registers the broader terms and the narrower terms of a plurality of specific events or things in association (relation). Specific examples of broader terms and narrower terms of events or things include the following (Relational Example 1) to (Relational Example 3).
(Relational Example 1) <Higher> People → Scholar → Physicist → Geophysicist →… <Lower>
(Relationship example 2) <Upper> Food → Fermented food → Cheese → Cheddar cheese →… <Lower>
(Relationship example 3) <Higher> Sports → Sport competition → Ball games → Baseball →… <Lower>

  As the broader term database 221, an existing one may be used or a new one may be generated. The related examples 1 to 3 are merely examples, and the broader term database 221 includes the broader terms and the narrower terms of a large number of events or things. Moreover, it is desirable that the broader terms and narrower terms of events or things are nouns. Further, the broader terms and the narrower terms of events or things may be words or compound words. The broader term database 221 does not have to be in the document information providing server 2 and may be constructed in another server that can be accessed by the document information providing server 2 via the network NW.

  In the present embodiment, among the meaning tags described above, in particular, focusing on “exemplary”, a method for specifying an exemplary portion of a document from a plurality of constituent portions constituting the document will be described in detail.

  The control unit 23 includes a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), and the like. The control unit 23 is an example of an exemplary part specifying device and a computer according to the present invention. The control unit 23 functions as a document acquisition unit, a subject extraction unit, an example part candidate extraction unit, a ratio calculation unit, an example part specifying unit, a display data output unit, and the like in the present invention by executing the example part specifying program. The example portion specifying process is executed. The control unit 23 refers to the broader term database 221 in the example part specifying process.

  By the way, a sentence including a word or compound word representing a specific event or thing is easy to illustrate, but even such a word or compound word may not necessarily be illustrated. Therefore, in the example part specifying process according to the present embodiment, the broader and lowerer words (higher and lower concepts) of the word are compared with the subject matter of the document to determine whether the word is an example.

  FIG. 2 is a flowchart illustrating an example of the exemplary portion specifying process in the control unit 23. The process shown in FIG. 2 is started when there is an instruction from an operator's terminal, for example. When the processing shown in FIG. 1 is started, the control unit 23 acquires unprocessed document data from the storage unit 22 or storage means of another device (for example, reads it into a RAM) (step S1).

  Next, the control unit 23 extracts the subject of the document from the document related to the document data acquired in step S1 (step S2). For example, the control unit 23 determines one or more words included in the document title as the subject of the document. The subject may be extracted from a word included in a title attached to each section such as a chapter or a section in a document. Moreover, you may comprise so that a theme may be extracted from the word contained in the table of contents or index in a document. Moreover, you may comprise so that a theme may be extracted from a word (frequency word) with high appearance frequency in a document. In this case, in order to determine whether the appearance frequency is high, a threshold value for the appearance frequency is set in advance. The subject is a word included in the title of the document (may include words included in the title of each section, such as chapters and sections in the document, and words included in the table of contents or index in the document). Among them, it may be configured to be extracted from words having a high appearance frequency in the sentence. Or words included in the title of the document (may include words included in titles assigned to sections such as chapters and sections in the document, and words included in the table of contents or index in the document) and sentences The number of themes (themes of the subject) determined by distinguishing from words with high appearance frequency and increasing or decreasing the contribution of the words included in the title (weight given to the words included in the title) The number of types) may be changed.

  Since the subject generally represents a specific event or thing, the subject extracted in step S2 may be limited to words (higher term or lower term) registered in the broader term database 221. desirable.

  Next, the control unit 23 refers to the high-order low-order word database 221 and configures a document that includes words registered in the high-order low-order word database 221 from among the documents related to the document data acquired in step S1. The part is extracted as an example part candidate of the document (step S3). That is, the control unit 23 checks, for each constituent part of the sentence, whether or not a word contained in the constituent part of the document is registered in the higher-order word database 221, and selects one of the words contained in the constituent part of the document. If any one of them is registered in the broader term database 221, the constituent part is extracted as an example part candidate.

  Next, the control unit 23 sets the number of words included in the exemplary candidate portion extracted in step S3 as the first occurrence number (frequency) n1, and at least one of the subject extracted in step S2 and the broader term of the subject When one of the numbers is the second number of appearances n2, the ratio (n2 / n1) of the second number of appearances n2 to the first number of appearances n1 is calculated. The degree e (= 1− (n2 / n1): 0 ≦ e ≦ 1) is calculated for each exemplary portion candidate (step S4). Here, the ratio indicates the degree of whether or not the subject is along, and a high value of the ratio indicates that the subject is along the subject. The example portion e of the example portion candidate indicates the degree of association between the upper and lower concepts of the words included in the example portion candidate and the subject matter of the document. When the example degree value is high, the example portion candidate is exemplified in the document. It is highly possible that

  Next, the control unit 23 specifies an example part of the document from the example part candidates extracted in step S3 based on the example degree e calculated in step S4 (that is, based on the ratio (n2 / n1)). (Determine) (step S5). For example, in order to determine whether or not the example e is high, an example threshold value θ (0 ≦ θ ≦ 1) is set in advance. Then, the control unit 23 determines whether or not the example e is equal to or greater than the threshold θ (e ≧ θ), and identifies an example part candidate whose example e is equal to or greater than the threshold θ as an example part of the document. On the other hand, an example part candidate having an example e of less than the threshold θ (e <θ) is not specified as an example part.

  FIG. 3 is a diagram illustrating an example of the degree of example e for each of the example portion candidates 1 to 3. The example e of each of the example partial candidates 1 to 3 shown in FIG. 3 is calculated by the equation (1) shown in FIG. In this case, for example, when the threshold θ is set to “0.3”, only the example portion candidate 3 is specified as the example portion among the example portion candidates 1 to 3 shown in FIG. Such a threshold value θ may be configured to be arbitrarily set by a system operator or a system user.

  When the example part is specified in this way, the meaning tag “example” is assigned to the specified example part (corresponding to the element ID of the example part), and the assigned meaning tag “example” is assigned. It is stored in the storage unit 22.

  As described above, according to the example part specifying process, even when a clue word such as a conjunction is not present in the sentence or a connection relationship between sentences cannot be acquired, the example part is efficiently specified. be able to.

  Next, a usage example of the semantic tag given to the example part specified by the example part specifying process will be described. As a premise, the user terminal 1 downloads desired document data (a document ID and an element ID are added) from the document information providing server 2, and a document content browsing screen including a part of the document related to the document data is displayed on the user terminal 1. Are displayed on the display (an example of display means).

  FIG. 4A is a diagram illustrating an example of a document content browsing screen displayed on the display of the user terminal 1. In the display state of the document content browsing screen shown in FIG. 4A, when the user of the user terminal 1 designates a desired sentence (an example of a component part of the document) with a finger or the like as shown in FIG. The user terminal 1 accesses the document information providing server 2 and transmits a request including the element ID of the designated sentence. In response to this, the document information providing server 2 displays other text (including the specified example part specified above) related to the text (element ID) designated by the user or a summary of the text on the display. (Including data indicating a connection relationship between the designated sentence and other sentences) is returned (output) to the user terminal 1. This display data is a component part of the document related to the document data (or a summary of the component part), and a component part (or a summary of the component part) related to the identified example part is designated as an example part (or This is data for display in association with the summary of the example portion. When the user terminal 1 receives the display data transmitted from the document information providing server 2, the user terminal 1 displays a main point arrangement screen on the display.

  FIG. 4B is a diagram illustrating an example of a main point organizing screen displayed on the display of the user terminal 1. In FIG. 4B, “mobile phone”, “smartphone”, and “tablet” are summaries (essential points) of exemplary portions identified as described above. The display data returned from the document information providing server 2 to the user terminal 1 is data that can switch whether or not to display the example portion related to the text designated by the user or the summary of the example portion according to the user's instruction. It may be. In this case, a key point arrangement screen as shown in FIG. 5B is displayed on the display. In FIG. 5B, the display of the broken line and the portion indicated within the broken line can be switched in accordance with a user instruction. That is, it is possible to adopt a configuration in which the illustration is not displayed and only the present discussion is displayed. Note that the display data returned from the document information providing server 2 to the user terminal 1 may be display data for causing the display to display only the example portion related to the text designated by the user or the summary of the example portion. Good.

  In the above embodiment, the word included in the title of the document is distinguished from the word having a high appearance frequency (frequent word) in the sentence, and the contribution degree of the word included in the title (hereinafter referred to as “title contribution degree”). ) Has been described above. However, the title contribution may be configured so that the user can arbitrarily set the title contribution. Moreover, you may comprise so that a user can set arbitrarily the threshold value of appearance frequency for determining whether appearance frequency is high. Furthermore, the threshold value θ for determining whether or not the example e is high may be configured so that the user can arbitrarily set the threshold value θ.

  FIG. 6 is a diagram illustrating an example in which the user sets the title contribution. 6A, the user of the user terminal 1 moves the slider 51 for changing the title contribution degree to the right or left as shown in FIG. 6A. The title contribution is changed. The title contribution degree thus changed is transmitted from the user terminal 1 to the document information providing server 2 together with the document ID of the document and the element ID of the component whose summary is displayed on the screen. Thereby, the document information providing server 2 performs the example part specifying process shown in FIG. 2, specifies the example part again, and displays display data for displaying the specified example part or its summary on the display to the user terminal 1. By replying, as shown in FIG. 6B or FIG. 6C, the number of display examples increases or decreases.

  FIG. 7 is a diagram illustrating an example in which the user sets the threshold value of the appearance frequency. 7A, the user of the user terminal 1 moves the slider 52 for changing the threshold of appearance frequency to the right or left as shown in FIG. 7A. The threshold of the appearance frequency is changed. The threshold value of the appearance frequency thus changed is transmitted from the user terminal 1 to the document information providing server 2 together with the document ID of the document and the element ID of the component whose summary is displayed on the screen. Thereby, the document information providing server 2 performs the example part specifying process shown in FIG. 2, specifies the example part again, and displays display data for displaying the specified example part or its summary on the display to the user terminal 1. By replying, as shown in FIG. 7 (B) or (C), the number of display examples increases or decreases.

  FIG. 8 is a diagram illustrating an example in which the user sets the threshold value θ of the example e. 8A, the user of the user terminal 1 moves the slider 53 for changing the threshold value θ of the example e to the right or left as shown in FIG. 8A. By doing so, the threshold value θ of the example e is changed. The threshold value θ of the degree of illustration e thus changed is transmitted from the user terminal 1 to the document information providing server 2 together with the document ID of the document and the element ID of the component whose summary is displayed on the screen. Thereby, the document information providing server 2 performs the example part specifying process shown in FIG. 2, specifies the example part again, and displays display data for displaying the specified example part or its summary on the display to the user terminal 1. By replying, as shown in FIG. 8B or FIG. 8C, the number of display examples increases or decreases.

  As described above, if the user can arbitrarily set the title contribution level, the appearance frequency threshold value, or the example degree e threshold value θ, the user can view the summary of the example portion displayed on the summary screen. However, since the display number of examples can be changed as appropriate, the examples more in line with the user's intention can be displayed.

  In the above embodiment, the document information providing server 2 is configured to perform the exemplary portion specifying process. However, as another example, the user terminal 1 that can access the broader term database 221 performs the exemplary portion specifying process. You may comprise as follows. In this case, the control unit (an example of a computer and an information processing device) of the user terminal 1 executes the exemplary part specifying program of the present invention, whereby the document acquisition unit, the subject extraction unit, the exemplary part candidate extraction unit, and the ratio in the present invention It functions as a calculation unit, an example portion specifying unit, a display data output unit, and the like.

1 User terminal 1
2 Document information providing server 21 Communication unit 22 Storage unit 23 Control unit

Claims (7)

An example portion specifying device that can refer to a broader term database that registers and associates broader terms and narrower terms of a plurality of events or things,
Document acquisition means for acquiring document data from a predetermined storage means;
Subject extraction means for extracting the subject of the document from the document related to the document data acquired by the document acquisition means;
With reference to the broader term database, the constituent parts of the document containing the words registered in the broader term database are selected from the documents related to the document data acquired by the document acquisition unit. An example portion candidate extracting means for extracting as an example portion candidate;
The number of words included in the example part candidate extracted by the example part candidate extraction unit is a first appearance number, and the number of at least one of the subject extracted by the subject extraction unit and the broader word of the subject A ratio calculating means for calculating a ratio of the second number of appearances to the first number of appearances,
Based on the ratio calculated by the ratio calculation means, an example part specifying means for specifying an example part of the document from the example part candidates extracted by the example part candidate extraction means;
An example portion specifying apparatus comprising:   The exemplary portion specifying apparatus according to claim 1, wherein the subject of the document extracted by the subject extracting unit is limited to a word registered in the broader term database.   3. The display data output means for outputting display data for causing the predetermined display means to display the example part specified by the example part specifying means or the summary of the example part, further comprising: The example part identification apparatus of description.   The display data is a component part of the document related to the document data acquired by the document acquisition unit or a summary of the component part, and a component part or the component part related to the example part specified by the example part specifying unit The example part specifying device according to claim 3, wherein the summary is data for displaying the example part in association with the example part or the summary of the example part.   5. The example according to claim 4, wherein the display data is data that can switch whether or not to display the example part specified by the example part specifying unit or the summary of the example part according to a user instruction. Partial identification device. An example portion specifying method executed by a computer capable of referring to a broader term database that associates and registers broader terms and narrower terms of a plurality of events or things,
A document acquisition step of acquiring document data from a predetermined storage means;
A subject extraction step of extracting a subject of the document from the document related to the document data acquired by the document acquisition step;
Referring to the broader term database, the constituent parts of the document containing the words registered in the broader term database are selected from the documents related to the document data acquired by the document acquisition step. An example portion candidate extraction step for extracting as an example portion candidate;
The number of words included in the example part candidate extracted by the example part candidate extraction step is a first appearance number, and the number of at least one of the subject extracted by the subject extraction step and the broader term of the subject A ratio calculating step of calculating a ratio of the second number of appearances to the first number of appearances,
Based on the ratio calculated by the ratio calculation step, an example part specifying step for specifying an example part of the document from the example part candidates extracted by the example part candidate extraction step;
An example portion specifying method comprising: A computer capable of referring to a broader term database that registers and associates broader terms and narrower terms of multiple events or things,
Document acquisition means for acquiring document data from a predetermined storage means;
Subject extraction means for extracting the subject of the document from the document related to the document data acquired by the document acquisition means;
With reference to the broader term database, the constituent parts of the document containing the words registered in the broader term database are selected from the documents related to the document data acquired by the document acquisition unit. Example part candidate extraction means for extracting as example part candidates;
The number of words included in the example part candidate extracted by the example part candidate extraction unit is a first appearance number, and the number of at least one of the subject extracted by the subject extraction unit and the broader word of the subject Is a second appearance number, a ratio calculating means for calculating a ratio of the second appearance number to the first appearance number, and
An example characterized in that, based on the ratio calculated by the ratio calculation means, it functions as an example part specifying means for specifying an example part of a document from the example part candidates extracted by the example part candidate extraction means. Partial identification program.