KR20180101991A - Creative thinking support apparatus and creative thinking support method - Google Patents

Abstract

An objective of the present invention is to support creative thinking of a user in accordance with writing or speaking by the user while the user writes or speaks. In order to achieve the objective, a creative thinking support apparatus maintains nodes showing elements making up a proposition included in a document defining knowledge, information showing a first graph structure including edges showing relations between the elements shown by the proposition, and information showing an appearance order of relations in the document, receives input of a proposition, extracts elements included in the proposition and relations between the elements shown by the proposition, includes nodes showing the extracted elements, generates a second graph including the relations between the extracted elements as edges, and specifies relations appearing after relations shown by the second graph in the document if the corresponding relations are determined to be included in the first graph.

Description

TECHNICAL FIELD [0001] The present invention relates to an apparatus and a method for supporting an idea,

The present invention relates to an idea support apparatus and an idea support method.

BACKGROUND ART [0002] Japanese Patent Application Laid-Open No. 2007-241901 (Patent Document 1) is known as a background art in this technical field. This publication discloses a technique for extracting a plurality of points related to the theme among a group of comment texts related to an inputted theme, and a method for extracting a plurality of points of interest related to the theme from among the plurality of points of view, A degree of importance calculation means for calculating a degree of opinion representing a ratio of opinions having an admission; a degree of importance calculation means for calculating a degree of importance of the opinion including the one admission for each of the plurality of topics; A representative opinion selection means for selecting a plurality of representative opinions in an affirmative or negative opinion for each of the plurality of issues; And an interface means for outputting the degree of importance and the importance. &Quot;

Japanese Patent Application Laid-Open No. 2007-241901

For example, it is assumed that the user performs writing or uttering about a given theme. The technique described in Patent Document 1 supports the writing and uttering by the user by extracting the points related to a given theme in advance.

However, when a user proceeds writing or uttering, each issue extracted by the technique described in Patent Document 1 can not be said to be a matter suitable to the context of the composition or the conversation concerned. Therefore, one aspect of the present invention aims at supporting the user's idea in accordance with the context of writing or uttering by the user while the user is writing or uttering.

In order to solve the above problems, one aspect of the present invention employs the following configuration. 1. An idea support apparatus for supporting a user's idea, comprising: a processor and a memory, the memory comprising: a node representing an element constituting a proposition contained in a document defining knowledge; The graph information indicating a structure of a first graph including a first kind of edge indicating a relation between the elements representing the first graph and context information indicating a sequence of appearance of a relation in the document, Extracts a relation between an element included in the proposition in which the input has been accepted and an element between the element in which the input has been accepted and an element that represents the extracted element and includes a node indicating the extracted element, Generating a second graph including a relation between elements as the first type edge, referring to the graph information, And when it is determined that the relation represented by the second graph is included in the first graph, referring to the context information, specifies a relation appearing after the relation in the document, and outputs information indicating the specified relation The idea support device.

One aspect of the present invention can support the user's idea in accordance with the context of writing or uttering by the user while the user is writing or uttering.

The problems, the constitution and the effects other than the above are clarified by the following description of the embodiment.

1 is a block diagram showing a configuration example of an idea support apparatus according to the first embodiment;
Fig. 2 is a flowchart showing an example of processing performed by the idea support apparatus in the present embodiment. Fig.
3A is an example of a knowledge graph showing knowledge in a document in the first embodiment;
3B is an example of a knowledge graph showing knowledge in a document in the first embodiment;
3C is an example of a knowledge graph showing knowledge in a document in the first embodiment;
FIG. 3D is an example of a node table in the first embodiment; FIG.
3E is an example of an edge table in the first embodiment;
4A is an example of a knowledge graph showing knowledge in a document in the first embodiment;
4B is an example of a knowledge graph showing knowledge in a document in the first embodiment;
4C is an example of a node table in the first embodiment;
4D is an example of an edge table in the first embodiment;
5A is an example of a knowledge graph showing knowledge in a document in the first embodiment;
5B is an example of a knowledge graph showing knowledge in a document in the first embodiment;
5C is an example of a knowledge graph showing knowledge in a document in the first embodiment;
5D is an example of a knowledge graph showing knowledge in a document in the first embodiment;
5E is an example of a knowledge graph showing knowledge in a document in the first embodiment;
6A is an example of a knowledge graph showing knowledge in a document in the first embodiment;
6B is an example of a node table in the first embodiment;
6C is an example of an edge table in the first embodiment;
7A is an example of a knowledge graph showing knowledge in a document in the first embodiment;
FIG. 7B is an example of a follow-up knowledge table in the first embodiment; FIG.
8A is an example of a knowledge graph showing knowledge in a document in the first embodiment;
8B is an example of a follow-up knowledge table in the first embodiment;
FIG. 9A is an example of a knowledge graph generated by the ignition understanding processing in the first embodiment. FIG.
FIG. 9B is an example of a knowledge graph generated by the ignition understanding processing in the first embodiment; FIG.
9C is an example of the transition of the screen displayed on the display device in the first embodiment;

Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. It should be noted that this embodiment is only one example for realizing the present invention, and does not limit the technical scope of the present invention.

Writing exercises are conducted at schools, for example, in essay writing, to develop expressive and thinking skills. Writing is different from the knowledge that can be obtained from problem books, so it is difficult to learn frequently and there are many cases where individual instruction by a teacher is needed to improve the writing skills of students. In addition, there is a problem that the student does not know what to write at the stage which is not familiar with writing first, and there are many cases that need a polite instruction by the teacher to solve the problem concerned. In addition, there are many cases in which conversation training is conducted in a foreign language, in which situations such as not knowing what to talk about are likely to occur.

Thus, the present embodiment supports the idea of writing a writing by presenting a hint or a comment to a user's writing. Likewise, in the present embodiment, in a foreign language conversation, a hint is given for the next utterance, and the conversation is assisted. As a result, the user can become accustomed to conversation in a foreign language, and further, it becomes easy to practice conversation with a person. The idea support apparatus of the present embodiment supports the user's idea by giving the user a hint of the next utterance or giving a comment in writing or conversation, and further supports the writing and conversation of the user.

[Example 1]

1 is a block diagram showing a configuration example of an idea support apparatus according to the present embodiment. The inventive idea supporting apparatus of the present embodiment includes an input device 202, a display device 203, a communication device 204, a computing device (CPU) 205, a memory 206, and an auxiliary storage device 207 ).

The input device 202 accepts input of a command or the like from a user, and is, for example, a keyboard and a mouse. The input device 202 accepts input of commands and the like to be executed for control of a program executed by the computing device (CPU) 205 and control of devices connected to the idea supporting device 201. [ The communication device 204 transmits processing contents by the idea supporting device 201 to an external device or receives information from an external device, for example, according to a predetermined protocol.

A computing device (CPU) 205 includes a processor and executes a program stored in the memory 206. [ The memory 206 includes ROM, which is a nonvolatile memory element, and RAM, which is a volatile memory element. The ROM stores an immutable program (e.g., BIOS) and the like. The RAM is a high-speed and volatile memory element such as a dynamic random access memory (DRAM), and temporarily stores data to be used by the processor and programs executed by the processor.

The auxiliary memory 207 is a large-capacity nonvolatile memory such as a magnetic storage device (HDD), a flash memory (SSD), and the like. And stores data used at the time of execution. That is, the program is read from the auxiliary memory 207, loaded into the memory 206, and executed by the computing device (CPU) 205.

The program executed by the computing device (CPU) 205 is provided to the inventive idea support device 201 via a removable medium (CD-ROM, flash memory or the like) or a network and is stored in a nonvolatile auxiliary memory And is stored in the device 207. To this end, the idea support apparatus 201 may have an interface for reading data from the removable medium.

The idea support apparatus 201 is a computer system that is physically configured on one computer or on a plurality of computers configured logically or physically and may operate as separate threads on the same computer and may be operated on a plurality of physical computer resources It may be operated on the virtual computer.

The CPU 205 includes a knowledge graph generation unit 211, a context calculation unit 212, a speech understanding unit 213, a context recognition unit 214, and a phase recognition unit 215. For example, the computing device (CPU) 205 functions as the knowledge graph generating section 211 by operating in accordance with the knowledge graph generating program loaded in the memory 206, and outputs to the context calculating program And thus functions as the context calculation unit 212 by operating. The same applies to other units included in the computing device (CPU) 205.

The knowledge graph generation unit 211 generates a graph (knowledge graph) expressing knowledge represented by a document of the document DB 103, which will be described later. The context calculation unit 212 calculates the appearance order of the knowledge represented by the knowledge graph held by the knowledge graph DB 105, which will be described later, in the document of the document DB 103. [ The utterance understanding unit 213 accepts input of a writing or a conversation from a user and creates text data of the composition or conversation.

The context recognition unit 214 calculates the appearance order of knowledge in the text data created by the speech understanding unit 213 and compares the appearance order with the appearance order of the knowledge calculated by the context calculation unit 212 do. The phase recognizing unit 215 recognizes the user's writing or conversation phase by judging whether or not the predetermined content required for writing or conversation is insufficient for the text inputted by the user.

The auxiliary storage device 207 holds the document DB 103, the knowledge graph DB 105, and the context DB 107. [ The document DB 103 maintains a document written in a natural language, in particular, a sentence. Since the document held by the document DB 103 is a document for defining knowledge in the idea support apparatus 201, that is, a document serving as an information source, the document DB 103 is a document to be discussed , It is desirable to keep a document describing contents related to consultation.

Web documents, news articles, internal reports, academic papers, public announcements from public authorities and international organizations, and research documents such as think tanks are examples of documents maintained by document DB 103. In addition, the history of past utterances by the user is an example of a document maintained by the document DB 103. Further, the document DB 103 may further maintain bibliographic information of each document.

The knowledge graph DB 105 stores information indicating the structure of the knowledge graph generated by the knowledge graph generation unit 211. [ The context DB 107 stores information indicating the appearance order of knowledge calculated by the context calculation unit 212. [

In the present embodiment, the information used by the idea support apparatus 201 does not depend on the data structure, but may be represented by any data structure. For example, a properly selected data structure from a table, list, database or queue can store information. In this embodiment, each data held by the auxiliary storage device 207 is shown in a table structure.

The idea support apparatus 201 may include at least one of the input device 202, the display device 203, and the communication device 204. When the idea support apparatus 201 does not have the input device 202, for example, the communication device 204 accepts input of a command or the like from an external device. When the idea support apparatus 201 does not have the display apparatus 203, for example, the communication apparatus 204 transmits the processing result generated by the idea support apparatus 201 to the external apparatus.

Each processing unit may execute input / output to another processing unit through the memory 206 or the auxiliary memory 207. [ For example, when the knowledge graph generating unit 211 outputs the processing result to the context calculating unit 212, the knowledge graph generating unit 211 actually outputs the processing result to the memory 206 or the auxiliary storage device 207, and the context calculation unit 212 may acquire the output result stored in the memory 206 or the auxiliary storage 207 as an input.

Fig. 2 is a flowchart showing an example of processing executed by the idea support apparatus 201 in the present embodiment. The processing executed by the idea support apparatus 201 can roughly be divided into two parts: a learning process (S110) and an operation process (S120). The learning process includes the processes of steps S111 to S112, and the operation phase includes the processes of steps S121 to S125. First, the processing of the learning process (S110) will be described. First, the knowledge graph generation unit 211 generates a knowledge graph to be stored in the knowledge graph DB 105 by converting the information of a document stored in the document DB 103 into a graph type knowledge (S111). The details of step S111 will be described below.

3A to 3C are examples of a knowledge graph representing knowledge in a document. In the present embodiment, an example of a document written in Japanese is used, but the same processing is performed in another language such as English. Hereinafter, a node represented by an ellipse of a single line in the graph in the figure is a node indicating an entity representing a person, an object, an organization, or an abstract concept or the like. In addition, the characters described in accordance with the arrows between the entities indicate relations between the entities.

The knowledge graph 301 is a knowledge graph generated from a sentence in which the knowledge graph generating unit 211 finds that "the employer of a school uniform in the school increases the discipline" in the document of the document DB 103 .

The knowledge graph generation unit 211 first extracts an entity from a sentence in the generation of the knowledge graph. The knowledge graph generating unit 211 extracts an entity using, for example, machine learning. That is, a learning document set labeled with an entity to be extracted is prepared in advance, and the knowledge graph generating unit 211 creates a classifier for extracting an entity, and uses the generated identifier to generate an entity in the sentence . It is assumed that the entities in the sentence of Fig. 3A are "expert", "uniform", and "discipline".

Further, the knowledge graph generation unit 211 may extract an entity by referring to a prepared word dictionary or an ontology of a word. Since most of the entities are represented by phrases corresponding to nouns and adjectives, the knowledge graph generating unit 211 may perform a syntax analysis to extract nouns and adjectives as entities.

Next, the knowledge graph generation unit 211 calculates relations between entities. "Uniform" and "discipline" are given by the argument ("uniform") of the predicate which is a predicate, and "discipline" is given in the sentence "the expert raises the discipline" Object). The knowledge graph generation unit 211 extracts, as a relation between entities, a predicate using two entities as arguments, for example, by syntactic analysis.

The knowledge graph generating unit 211 generates the knowledge graph 301 by connecting the two entities and the relations therebetween by an edge. The knowledge graph generation unit 211 generates an edge based on a node pointing to a given entity as an origin point and a node pointing to an entity as a predicate as an endpoint in generation of the knowledge graph. In addition, the knowledge graph may be an irregular graph.

In this example, the knowledge graph generation unit 211 has shown an example of extracting a predicate having a factor as an argument as an argument, but it is also possible to extract another relation between entities. For example, the knowledge graph generation unit 211 may extract a connection relationship between entities as a relation, or may extract a distance in a sentence between entities as a relation. The knowledge graph generating unit 211 may extract the relation using, for example, Semantic Role Labeling.

The knowledge graph 302 is a knowledge graph generated from a sentence in which the knowledge graph generator 211 determines that "the uniform at school is inhibiting freedom" in the document of the document DB 103. The knowledge graph generating unit 211 extracts two entities " uniform " and " freedom " and a relation " inhibit " between entities from the sentence in the same manner as in the example of Fig. ).

The knowledge graph generation unit 211 thus extracts the entities from each sentence in the given document and describes the relations in the sentences between the entities using the knowledge graph.

3C is an example of a knowledge graph incorporating two knowledge graphs. For example, when there are the same nodes in a plurality of knowledge graphs, the knowledge graph generating unit 211 may integrate the plurality of knowledge graphs by identifying the nodes. The knowledge graph 303 is a knowledge graph generated by the knowledge graph generating unit 211 identifying a "uniform" node in the knowledge graph 301 and a "uniform" node in the knowledge graph 302 .

The knowledge graph generating unit 211 identifies nodes having the same surface layer (surface layer) representation, for example. When a node is created by using a word dictionary or an ontology, the knowledge graph generation unit 211 may treat the same expression as the same node, for example, on a word dictionary or an ontology. Specifically, for example, the knowledge graph generation unit 211 may identify the nodes of synonyms. The knowledge graph generating unit 211 may identify the edge indicating the same relation among the edges connecting the two nodes on the graph.

The knowledge graph generating unit 211 displays information of the generated knowledge graph in a table shown in, for example, FIGS. 3D and 3E, and stores the table in the knowledge graph DB 105. FIG. FIG. 3D is an example of a node table stored in the knowledge graph DB 105. FIG. The node table 304 is a table for holding information of nodes of the knowledge graph. The node table 304 in Fig. 3D holds the information of nodes of the knowledge graph 303. Fig.

The knowledge graph generating unit 211 assigns an ID for identifying a node on the knowledge graph to a node of the generated knowledge graph, and stores the assigned ID in the node table 304. [ The knowledge graph generation unit 211 may include other information about the node in the node table 304. [ In the example of FIG. 3D, the node table 304 contains information of the surface layer representation of each node. For example, the node table 304 may include bibliographic information or the like of a document in which a surface expression corresponding to each node is extracted.

3E is an example of an edge table stored in the knowledge graph DB 105. FIG. The edge table 305 is a table including information of edges of the knowledge graph. In the example of FIG. 3E, the edge table 305 holds edge information of the knowledge graph 303. FIG.

The knowledge graph generation unit 211 assigns IDs for identifying the edges on the knowledge graph to the edges of the generated knowledge graphs and associates the IDs of the respective edges, the IDs of the starting nodes and the IDs of the end nodes And stores it in the edge table 305. The knowledge graph generation unit 211 may include other information about the edge in the edge table 305. [ In the example of FIG. 3E, the edge table 305 includes information of the surface representation of each edge. For example, the edge table 305 may include bibliographic information or the like of a document in which a surface expression corresponding to each edge is extracted. The structure of the knowledge graph is defined by the node table 304 and the edge table 305.

Hereinafter, another example of the knowledge graph generation process will be described. 4A to 4C are examples of knowledge graphs. The sentence in Fig. 4A is the same as the sentence in Fig. 3A. The knowledge graph 301 represents information showing that the sentence "increases the discipline" indicated by the sentence. However, the knowledge graph 301 expresses information indicating that the "expert said" (that is, information on a relation called "said") indicated by "the uniform increases the discipline" indicated by the sentence It is not.

The knowledge graph generating unit 211 can generate the knowledge graph 402 in addition to the knowledge graph 301, for example, when a statement such as " the uniform is to increase the discipline " is regarded as an entity. A statement is a proposition indicating a plurality of elements constituting a sentence and a relation between the plurality of elements. Each element may be, for example, an entity, a clause or a sentence. In addition, each element may be a statement.

Since the knowledge graph 301 and the knowledge graph 402 are independent knowledge graphs, it is not possible to express the relationship between two knowledge graphs represented by the sentence in FIG. 4A. In addition, when each of a plurality of related knowledge is represented by such an independent knowledge graph, the number of entities becomes large.

Thus, the knowledge graph generation unit 211 generates the knowledge graph 403 shown in FIG. 4B in order to express such information. An example of the generation process of the knowledge graph 403 will be described. As described above, when at least one of the objects leading to the relation in the sentence of the document is a statement, the knowledge graph generating unit 211 regards the statement as an entity provisionally, as in the knowledge graph 402, .

When it is determined that there is a node including all the information of a statement represented by another knowledge graph generated from the same document (that is, when there is a node in a nesting structure) ), And a node (hereinafter also referred to as a representative node) representing this statement is generated from the other knowledge graph.

In the example of Fig. 4A, since the " increase uniformity " node of the knowledge graph 402 includes all of the information of the statement represented by the knowledge graph 301 generated from the same document as the knowledge graph 402, The graph generating unit 211 generates a representative node. The knowledge graph generating unit 211 changes the node to "raise" the edge of "raise" representing the relation of the knowledge graph 301, for example, and decides it as the representative node. The knowledge graph generating unit 211 generates a knowledge graph using the representative node as the starting point and the edge that has the representative node as the starting point and the starting point as the starting point of the " Quot; discipline " node as an end point. As a result, a portion surrounded by a dotted line of the rectangle of the knowledge graph 403 is generated.

As noted above, the "raise" node, represented by the double line in the knowledge graph 403, is a representative node that represents one statement that contains three nodes ("uniform", "raise", "discipline"). A representative node representing N (N is an integer equal to or greater than 1) statement is called an (N + 1) th node. That is, the " raise " node represented by the double line of the knowledge graph 403 is a secondary node. A node representing the entity itself is called a primary node. That is, when a relation between N-order nodes is converted into a representative node, the representative node is an N + 1-order node.

The edges indicated by the double lines in the knowledge graph 403 do not represent the relations between the nodes but represent the statements composed of the three nodes included in the portion surrounded by the dotted line of the rectangle of the knowledge graph 403, It is an edge for representing representative nodes. Hereinafter, such an edge is referred to as a statement edge.

As described above, in the knowledge graph 402, the "Expert" node as the starting point and the node "Increase discipline" as the end point are connected by an edge indicating the relation "said". Therefore, the knowledge graph generation unit 211 sets the "increase" node representing the statement "to increase the discipline" as the end point, the two nodes as the "start" , The knowledge graph 403 is generated.

Accordingly, the knowledge graph generation unit 211 can express a statement that "the uniform increases the discipline" and "the expert" says "the uniform increases the discipline" by one knowledge graph 403. If the knowledge graph including the statement itself as an entity remains after the integration processing of the graph as described above, the knowledge graph generation unit 211 may delete the knowledge graph.

4C is an example of the node table 304. FIG. The node table 304 may further maintain the degree of the node. That is, the knowledge graph generation unit 211 associates the node ID with the degree of the node, and stores it in the node table 304. [

Fig. 4D is an example of the edge table 305. Fig. The edge table 305 may hold a flag indicating whether or not the edge is a statement edge. That is, the knowledge graph generation unit 211 associates the edge ID with a flag indicating whether or not the edge is a statement edge, and stores it in the edge table 305. [ Since the statement edge does not represent a surface expression, a null value is stored in the surface expression corresponding to the statement edge in the edge table 305. [

The knowledge graph generating unit 211 can express not only relations between simple entities but also more complicated information as a knowledge graph by creating a knowledge graph having a graph structure including a secondary node indicating a statement in this manner.

5A to 5C are examples of the knowledge graph. The knowledge graph 501 is a knowledge graph prepared from the sentence "The uniform increases the discipline, but hinders the freedom". 3C, the knowledge graph generating unit 211 first obtains from the sentence a knowledge graph 303 indicating that "the uniform increases the discipline" and "the uniform inhibits the uniform" .

The knowledge graph generating unit 211 refers to, for example, a dictionary prepared in advance, and, when judging that the sentence has a superficial expression representing a paradox, the knowledge graph generating unit 211 generates a statement saying " , And a statement that the subordinate clause " uniformity hinders freedom ". For example, the knowledge graph generating unit 211 generates a representative node " raise " from the main verse and a representative node " inhibit " from the subordinate clause in the same manner as the above-described method. Each of these representative nodes is a secondary node.

Then, the knowledge graph generating unit 211 generates a knowledge graph using an edge that has a "uniform" node as a starting point and a representative node as an end point, which is the starting point of an edge, and a "discipline &Quot; node as an end point. Likewise, the knowledge graph generating unit 211 generates the knowledge graph generating unit 211 based on the "uniform" node that was the starting point of the "inhibition" edge, the edge whose representative node is the end point, Quot; free " Each of these generated edges is a statement edge.

The knowledge graph generating unit 211 generates a paradoxical edge indicating the relation of the paradox with the representative node "raise" of the main routine as a starting point and the representative node "inhibit" of the dependent flag as the end point. By this processing, the knowledge graph generating unit 211 can generate the knowledge graph 501. [

Hereinafter, an example in which the knowledge graph generating unit 211 generates the knowledge graph 503 shown in Fig. 5C will be described from the sentence " The uniform increases the discipline according to the investigation "shown in Fig. 5B. The knowledge graph generating unit 211 first generates the knowledge graph 301 as in the example of FIG. 4A. Also, the knowledge graph generating unit 211 generates the knowledge graph 502. An example of a method of generating the knowledge graph 502 will be described.

The knowledge graph generating unit 211 refers to a dictionary prepared in advance, and when it is judged that a word ("search" in FIG. 5B) indicating an information source (information source) is present in the sentence, And extracts the information indicated by the information source (" Increase discipline " in Fig. 5B).

The knowledge graph generation unit 211 regards a statement that "the uniform increases the discipline" as an entity. The knowledge graph generation unit 211 generates an "information source" edge as a relation between the two entities, starting from the "irradiation" node and ending with the node "increase uniformity" as the end point. Accordingly, the knowledge graph generating unit 211 can generate the knowledge graph 502 shown in FIG. 5C.

The knowledge graph generator 211 integrates the knowledge graph 301 and the knowledge graph 502 to generate the knowledge graph 503 in a manner similar to that described with reference to Figures 4A and 4B .

Hereinafter, the knowledge graph generating unit 211 generates the knowledge graph 505 shown in Fig. 5E from the sentence " Expert says that the uniform increases the discipline, but hinders freedom ". An example is given. The knowledge graph generating unit 211 generates the knowledge graph 501 from the portion of the sentence that says, "The uniform increases the discipline, but hinders freedom", by the method described with reference to FIG. 5A. Further, the knowledge graph generating unit 211 regards the statement " the uniform increases the discipline but inhibits the freedom " as an entity, and generates the knowledge graph 504.

Since the node of the knowledge graph 504 increases the discipline, but inhibits the freedom, the node includes all of the information of the statement represented by the knowledge graph 501 generated from the same sentence, Quot; paradoxical " node representing the " paradoxical " node indicating the " paradoxical " node. The degree of the representative node "paradox" is three.

Then, the knowledge graph generating unit 211 generates a knowledge graph using a node that has a starting point of a "paradoxical" edge as a starting point, a statement edge whose end point is a "paradoxical" node, Quot; edge " node which is the end point of the " edge " edge.

The knowledge graph generation unit 211 also has a "paradox" node representing the statement "uniform is high in discipline, but inhibits freedom" as an end point, and two nodes To the edge indicating the relation "said", thereby generating the knowledge graph 505.

As described above, the knowledge graph generation unit 211 of the present embodiment expresses nested information as a graph type knowledge by using a special node (a statement edge) connecting the node with a higher order (higher order) node. Accordingly, the knowledge graph generating unit 211 can represent not only simple information but also a knowledge graph with a small complex context. In other words, although the idea support apparatus 201 of the present embodiment can handle a complicated context, it suppresses the data amount of the knowledge graph DB 105, and also controls the comparison of the knowledge graph in the speech understanding process Processing can be performed at a high speed.

Returning to the description of FIG. Subsequently, the context calculation unit 212 executes a context calculation process for calculating the order in which information appears in the sentence using the knowledge graph held by the knowledge graph DB 105 (S112).

As described above, the knowledge graph generating unit 211 graphs the individual information extracted from the sentence, and then, based on a predetermined criterion (e.g., surface expression, word dictionary, or position on the ontology) The knowledge graph shown in Figs. 3C, 4B, 5A, 5C, 5D, and the like is generated.

6A is an example of a knowledge graph. Sentence (601) is an example of an essay contesting the introduction of school uniforms. In the sentence (601), the opinion to oppose the introduction of the uniform is expressed, and the merit that the discipline increases afterwards is described, and the demerit that the student's freedom is inhibited is described. Further, in the sentence 601, the opinion of the viewpoint that the fairness of the market is hindered by the adoption of the specific uniform is described.

As such, writing, conversation, essay writing, etc. have a flow of information presentation (context). The context calculation unit 212 of the present embodiment calculates the context of a sentence in the document DB 103 and stores the calculation result in the context DB 107. [ The context calculation unit 212 performs context calculation by calculating the order of knowledge appearing in a sentence.

The idea supporting apparatus 201 of the present embodiment recognizes the context of writing and conversation by the user by using the knowledge graph DB 105 and the context DB 107 and generates appropriate hints, And the like to the user. Accordingly, the idea support apparatus 201 can deepen the accident of the user and support the idea.

Hereinafter, the details of the context calculation processing (S112) will be described. The knowledge in the context calculation process is the knowledge represented by the relation included in the statement, that is, the knowledge represented by an edge other than a node or a statement edge that is second or higher order. The knowledge graph 602 of FIG. 6A is an example of the knowledge graph generated from the sentence 601. FIG. 6B is an example of the node table 304 corresponding to the knowledge graph 602. FIG. 6C is an example of an edge table 305 corresponding to the knowledge graph 602. FIG.

For each document included in the document DB 103, for example, the context calculation unit 212 calculates the order of knowledge displayed in the document on the knowledge graph. The context calculator 212 calculates the ID of the node and the edge related to the knowledge in the knowledge graph 602 to increase the discipline, , E1, N2, E2, N3). According to this method, for example, the knowledge that "the uniform increases the discipline, but inhibits the freedom" is (N1, E1, N2, E2, N3, E3, N4, E5, N1, E6, N6, E7, N7).

In addition, it is preferable that the context calculation unit 212 express knowledge more concisely using a method described below. More specifically, the context calculation unit 212 expresses knowledge by designating an edge that is not a statement edge and a node that is second or higher order. Accordingly, the context calculation unit 212 can concisely treat complex knowledge including complex sentences, Chinese sentences, and conjunctions, such as " uniforms increase discipline, but inhibits freedom " Hereinafter, a specific example will be used to explain the present invention.

First, the knowledge graph 602 includes the secondary nodes N2 and N6, and the tertiary node N4 as secondary nodes or more. The secondary node N2 "increases" is connected to the node N1 "uniform" and the node N3 "discipline" and the statement edge, indicating knowledge that "the uniform increases the discipline". Likewise, the secondary node N6 " inhibits " indicates knowledge that " the uniform damages the freedom. &Quot; The third node (N4) "paradox" means that N2 is represented as "the uniform increases the discipline" and N6 which represents "the uniform hinders the freedom" leads to the statement edge. , And inhibit freedom ".

Next, the knowledge graph 602 includes E4 and E8 as edges, not a statement edge. The edge E4 is connected to the node N5 "expert" and the node N4 "paradox", and "expert indicates the knowledge that the uniform increases the discipline but inhibits the freedom". Likewise, the edge E8 represents knowledge that " the uniform damages the market fairness ".

The context calculation unit 212 uses the above method to identify the edge of an edge that is not a statement edge, as indicated by N4, for example, a complicated knowledge such as " uniform is high in discipline, Knowledge can be represented concisely using the ID of the second or higher node.

The context calculation unit 212 calculates the order in which the knowledge appears in the document in the context calculation process. An example of the calculation method will be described with reference to Figs. 7A and 7B. 7A is an example of a knowledge graph. The knowledge graph 703 is a knowledge graph generated from a sentence 701 and a sentence 702 included in the document DB 103. [ As described above, when knowledge is expressed by using an edge other than a statement edge and a node having a second degree or higher, knowledge appears in the order of N6, E8, N2, and E9 in the sentence 701. [ Similarly, in the sentence 702, knowledge appears in the order of N6 and E9.

The context calculation unit 212 calculates the number of occurrences of knowledge (hereinafter referred to as " subsequent knowledge ") appearing later than the knowledge in the document DB 103 for each knowledge appearing in the knowledge graph 703 And stores it in the subsequent knowledge table maintained by the context DB 107. [

In other words, the context calculation unit 212 specifies knowledge corresponding to an edge other than a node or a statement edge appearing in the knowledge graph 703. The context calculation unit 212 calculates the degree of similarity between each of the specific knowledge items in the document DB 103 as a knowledge that appears later than the knowledge in the document DB 103, The number of occurrences of knowledge represented by the edge is calculated.

FIG. 7B is an example of the following knowledge table. The subsequent knowledge table 704 holds the node ID or edge ID corresponding to the knowledge, the subsequent knowledge of the knowledge, and the number of occurrences of the subsequent knowledge after the knowledge in association with each other. For example, the second line of the subsequent knowledge table 704 of FIG. 7B is a table in which knowledge E8 is once, knowledge N2 is once, knowledge E9 appears twice after knowledge N6 . Since the knowledge E9 appears after N6 once in each of the sentence 701 and the sentence 702, in the following knowledge table 704 of Fig. 7B, the appearance of E9 as a follower knowledge of N6 The number of times is two.

By storing the subsequent knowledge and the number of occurrences of the subsequent knowledge in the knowledge table 704 in correspondence with each knowledge as described above, the context calculator 212 calculates the order of knowledge in the document and And can maintain the information indicating the relationship before and after.

For each knowledge, the context calculation unit 212 may calculate the distance on the document with the subsequent knowledge and store it in the subsequent knowledge table 704. [ For subsequent knowledge where the number of occurrences is two or more, the context calculation unit 212 stores in the following knowledge table 704 the average value or the minimum value of the distance of the knowledge and the following knowledge on the document. The learning process has been described above.

Next, the operation processing (S120) will be described. In the operation processing, the idea support apparatus 201 understands the context in writing and conversation by the user, and provides appropriate hints during writing and conversation, using the knowledge graph created in the learning process and the follow-up knowledge Thereby urging the user to start thinking.

First, the utterance understanding unit 213 accepts input of a composition or a conversation containing a statement from a user and converts it into text (S121). Specifically, the speech understanding unit 213 may receive text input from an input interface such as a keyboard, for example, or may convert texts input by a microphone or the like into text by speech recognition technology. For example, the speech understanding unit 213 may text-convert a sentence written on a paper by OCR or the like. Further, the utterance understanding unit 213 may text-convert the user's movement using a method such as sign language recognition.

Subsequently, the speech understanding unit 213 performs a speech understanding process to understand the text acquired in step S121 using the knowledge graph (S122). Here, the utterance understanding unit 213 comprehends the utterance of the user by matching the input text with the knowledge graph. 8A, 8B, 9A, 9B, and 9C, the details of the speech understanding processing will be described.

8A is a part of the knowledge graph represented by the node table 304 and the edge table 305 stored in the knowledge graph DB 105. The knowledge graph 801 shown in FIG. 8B is a part of the subsequent knowledge table 704 stored in the context DB 107. [

For example, about the theme "should uniform should be introduced to school", user shall make essay writing in the position of "in favor of introducing uniform to school". It is assumed that the utterance understanding unit 213 receives an input of a sentence such as " School should introduce uniforms " on the display screen 903 shown in Fig. 9C. The speech understanding unit 213 generates the knowledge graph 901 of FIG. 9A from the sentence using the same method as the knowledge graph generation process of step S111.

Next, the utterance understanding unit 213 collates the knowledge graph 901 with the knowledge graph 801, and retrieves knowledge contained in the knowledge represented by the knowledge graph 901 from the knowledge graph 801. Then, Using the knowledge representation method described in the context calculation process, the knowledge E10 represented by the knowledge graph 801 (i.e., the knowledge of the relation between the "school" node and the "uniform" node) 901), the utterance understanding unit 213 extracts E10. Thus, the utterance understanding unit 213 extracts the knowledge contained in the input of the user, and compares the extracted knowledge with the previously learned knowledge graph to understand the user's input.

The speech understanding unit 213 uses the same expression method as the context calculation process, for example, to generate a node ID and an edge ID on the knowledge knowledge graph 801 represented by the text input by the user, To the context recognition unit 214. The context recognition unit 214 receives the information indicating the knowledge appearance order in the context recognition unit 214. [ When the sentence indicated by the display screen 903 is input, the utterance understanding unit 213 outputs, for example, E10 to the context recognition unit 214. [ When the sentence 702 of FIG. 7A is input, the speech understanding unit 213 outputs the edge ID to the context recognition unit 214 in the order of, for example, N6 and E9.

By the way, the knowledge graph 901 completely coincides with a part (N10, E10, and N1) of the knowledge graph 801. In the contrast of the knowledge graph, the utterance understanding unit 213 can determine whether or not one relation or entity does not completely coincide with the other relation or entity, for example, when the degree of similarity is equal to or larger than a predetermined value, May be handled as the same.

For example, it is assumed that the speech understanding unit 213 generates the knowledge graph 902 of FIG. 9B from the text input from the user. Although the edge of the knowledge graph 902 is " accepted ", the utterance understanding unit 213 regards E10 " introduce " of the knowledge graph 801 as " 902) corresponds to E10.

The utterance understanding unit 213 calculates the degree of similarity between the entities and the relations based on a similarity relationship in a word dictionary, an ontology, or the like. The utterance understanding unit 213 can increase the knowledge in the knowledge graph DB 105 that matches the knowledge extracted from the user input by treating the relation or entity having the similarity degree equal to or higher than the predetermined degree as agreement, It is more likely to be able to give.

Returning to the description of FIG. Next, the context recognition unit 214 executes context recognition processing for collating the appearance order of knowledge represented by the text of the user input acquired in the speech understanding processing with the subsequent knowledge table 704 (S123). For example, when a text corresponding to the knowledge graph 901 is input from the user, E10 is output in step S122 as described above. At this time, the context recognition unit 214 outputs to the phase recognition unit 215 information represented by the second row (knowledge value E10 row) of the subsequent knowledge table 704 in Fig. 8B.

Next, the phase recognizing unit 215 executes a phase recognizing process for recognizing the user's writing or conversation phase (S124). For example, in a case where a user writes a certain theme (in the example of FIGS. 9A to 9C, "Should a uniform be introduced to the school") in favor or against the case, have.

The expressions of the pros and cons, the issue of the pros and cons, and the reasons and rationale of pros and cons are examples of what is required of the essay. The phase recognizing unit 215 recognizes the user's writing or conversation phase by judging whether or not the predetermined content required for writing or conversation is insufficient for the text inputted by the user.

Fig. 9C shows an example of the transition of the screen displayed on the display device 203. Fig. For example, the phase of the sentence displayed on the display screen 903 is a phase in which the position for the introduction of the uniform is expressed, but there are no issues, or a reason or ground is not presented. The phase recognizing unit 215 specifies the contents not described in the text input by the user from the predetermined contents.

In addition, contents required for composition and conversation are defined in advance. For example, the idea support apparatus 201 holds a document for learning in advance. For example, before starting the operation processing, the inventive support apparatus 201 prepares a label corresponding to the contents and prepares an identifier capable of recognizing the contents by machine learning using the label. The phase recognition unit 215 uses this identifier to determine whether or not the contents are included in the text input by the user. In the example of Fig. 9c, in writing, it is necessary to express the pros and cons of the introduction of the uniform to the school, and to issue arguments for the pros and cons expressed.

In addition, the idea support apparatus 201 may be provided in a place corresponding to the content of the document DB 103 (for example, a document including the content or a text including the content) corresponding to the content of the document DB 103 An annotation indicating that the point is located may be added.

When generating a knowledge graph including knowledge using an annotation part in the knowledge graph generation process, the knowledge graph generation unit 211 generates knowledge graphs indicating that the knowledge represented by the annotation part is represented by the knowledge, And may be provided as additional information. That is, for example, the knowledge graph generation unit 211 adds columns to the node table 304 and the edge table 305, and stores the additional information in the added column.

Returning to the description of FIG. Subsequently, the phase recognizing unit 215 performs a hint presenting process of presenting a hint to the user based on the results of the context recognizing process and the phase recognizing process (S125). The phase recognition unit 215 generates a hint to be presented from the subsequent knowledge represented by the subsequent knowledge table 704. [

As described above, when a sentence displayed on the display screen 903 is input, in the context recognition processing, information of a record in which knowledge of the following knowledge table 704 is E10 is output. In other words, the phase recognizing unit 215 has already obtained information indicating that the subsequent knowledge of knowledge E10 is N2 and E9.

Therefore, the phase recognizing unit 215 determines a statement indicated by the relation of the subsequent knowledge N2 and the subsequent knowledge E9 as a candidate of a hint to be presented to the user. The phase recognizing unit 215 may output all of the candidates to the display device 203 as a hint.

Further, for example, in the phase recognition processing, if there is a candidate corresponding to the knowledge indicating the content determined to be insufficient in the sentence displayed on the display screen 903, the phase recognizing section 215 gives priority to the candidate It may be displayed as a hint on the display device 203. [

The phrase " any candidate is displayed in preference " means, for example, that only the candidate is displayed on the display device 203 and that the candidate is highlighted when all the candidates are displayed on the display device 203 And the like.

In addition, the phase recognizing unit 215 gives priority to the knowledge in which the knowledge of the knowledge table 704 having knowledge is more frequent in the record having E10 (in the example of Fig. 8B, N2 is 10 times and E9 is 5 times) (For example, a predetermined number of candidates may be presented in the order of greater number of occurrences, or all candidates whose number of occurrences is equal to or greater than a predetermined number) may be presented as a hint to the display device 203. Accordingly, the phase recognizing unit 215 can output, as a hint, knowledge that is highly related to the contents inputted by the user.

If the distance between the knowledge and the subsequent knowledge is stored in the subsequent knowledge table 704, the phase recognition unit 215 determines that the distance in the record with the knowledge of the following knowledge table 704 is E10 (For example, a predetermined number of candidates may be presented in descending order of distances, or all candidates whose distances are equal to or less than a predetermined value) may be presented on the display device 203 in preference to a small knowledge. Accordingly, the phase recognizing unit 215 can output, as a hint, knowledge that is highly related to the contents inputted by the user.

For example, it should be a matter of or against the knowledge (N2) and the knowledge (E9). At this time, the phase recognizing section 215 gives priority to, for example, the frequency of appearance, and hints that "the uniform will increase the discipline" as displayed on the display screen 904 in FIG. 9C . The phase recognizing unit 215 displays a hint on the display device 203 when it is assumed that the user needs a hint, or when the user has received a hint presentation request from the user. Concretely, for example, the phase recognizing unit 215 may provide a hint when there is no next input even after a predetermined time elapses from the input of the user, or there may be a request from the user such as button pressing by the user You may also provide a hint in this case.

In this example, the phase recognizing unit 215 presents only one hint on the display screen 904, but a plurality of hints may be presented. Since the phase recognition process is a process for narrowing the hint to be presented, it is not necessary to execute the phase recognition process.

The user continues inputting the text as shown on the display screen 905 with reference to the hint presented in the display screen 904, and returns to the speech understanding processing. The user may input text that is not related to the presented hint. Then, for the input text, the speech understanding processing and the context recognition processing are executed again. The outputs in the context recognition processing when the sentence displayed on the display screen 905 is input are E10 and N2.

8B, since the following knowledge common to E10 and N2 is E9, the phase recognizing unit 215 recognizes, in the hint presenting process, for example, the display screen 906 ), "The uniform increases the motivation to learn" as a hint.

According to the above-described processing, the idea support apparatus 201 of the present embodiment can give a proper hint to the user when the user is blocked in the text input or when the user requests the hint. In addition, And can support the next utterance by the user.

The phase recognizing unit 215 may perform the following processing using the knowledge graph in the hint presenting process. As described above, when the user inputs a sentence displayed on the display screen 905, knowledge E10 and knowledge N2 are output in context recognition processing.

In the knowledge graph 801, there is knowledge N6 that is the inverse of the knowledge N2 output in the context recognition process. The phase recognition unit 215 recognizes that the knowledge of the paradox of knowledge inputted from the context recognition unit 214 is included in the knowledge graph and that the knowledge of the paradox is not input from the context recognition unit 214 In this case, a hint corresponding to the knowledge of the paradox may be presented. Concretely, for example, the phase recognizing unit 215 recognizes that the display screen 907 is displayed on the display screen 907, such as " How do you think that the uniform has high opinion on discipline, You may present questions or comments as a hint.

By this processing, the user can know the opinion from the opposite side of his / her own opinion, and can further assist in deepening the user's idea in essay writing, writing, and conversation. Although the example in which the phase recognition unit 215 presents the knowledge of paradoxical knowledge inputted from the context recognition unit 214 as a hint has been described in the above example, the phase recognition unit 215 may be configured to include the context recognition unit 214 may be presented as a knowledge in a relation other than the paradox of knowledge input from knowledge input from the context recognition unit 214 as a hint.

The present invention is not limited to the above-described embodiment, and various modifications are included. For example, the above-described embodiments have been described in detail in order to facilitate understanding of the present invention, and are not limited to those having all the configurations described above. It is also possible to replace some of the configurations of the embodiments with the configurations of other embodiments, and it is also possible to add configurations of other embodiments to the configurations of any of the embodiments. Further, it is possible to add, delete, or substitute another configuration with respect to a part of the configuration of each embodiment.

Further, the above-described components, functions, processing units, processing means, and the like may be realized by hardware by designing a part or all of them with, for example, an integrated circuit. Further, the above-described respective configurations, functions, and the like may be realized by software by analyzing and executing a program that realizes the respective functions of the processor. Information such as a program, a table, and a file that realize each function can be stored in a recording device such as a memory, a hard disk, a solid state drive (SSD), or a recording medium such as an IC card, an SD card, or a DVD.

In addition, the control lines and the information lines are considered to be necessary for explanation, and not necessarily all the control lines and information lines on the product. In practice, it can be considered that almost all configurations are connected to each other.

The present invention relates to an information processing apparatus and an information processing apparatus which are capable of providing a knowledge DB with a knowledge DB and a context DB. 212: context calculation unit, 213: speech recognition unit, 214: context recognition unit, 215: phase recognition unit,

Claims (10)

As an idea support device for supporting user's ideas,
Comprising a processor and a memory,
The memory comprising:
A structure of a first graph including a node representing an element constituting a proposition included in a document defining knowledge and a first kind edge representing a relation between the elements indicated by the proposition, And
Retaining context information indicating a sequence of occurrence of a relation in the document,
The processor comprising:
We accept input of proposition,
Extracts a relation between an element included in the proposition in which the input is accepted and the element indicated by the proposition in which the input is received,
Generating a second graph including a node indicating the extracted element and including a relation between the extracted elements as the first kind edge,
Refers to the graph information to specify a relation appearing after the relation in the document by referring to the context information when it is determined that the relation represented by the second graph is included in the first graph,
And outputs information indicating the specific relation. The method according to claim 1,
The document includes a plurality of propositions,
Wherein the node of the first graph comprises:
A first kind node indicating an entity included in the plurality of propositions,
And a second kind node indicating a relation of a first kind proposition, in which all elements and relations are contained in one element of another proposition in the plurality of propositions, as a proposition contained in the plurality of propositions,
Wherein the first graph includes a node indicating an element constituting the first kind of specification and a second kind of edge connecting the second kind of node,
The processor comprising:
Accepting input of a plurality of propositions,
In generating the second graph,
Determining an entity among the extracted elements as the first seed node,
Wherein when determining that the plurality of propositions that have accepted the input include the first kind proposition, the relation of the first kind proposition is determined to be the second kind node,
And connects the node indicating the elements constituting the first type specification and the determined second kind node to the second type edge. The method according to claim 1,
Wherein the context information indicates the number of occurrences of relations appearing after the relation in the document for each relation indicated by the proposition of the document,
The processor comprising:
Selects a relation from the specific relation on the basis of the occurrence count of the specific relation indicated by the context information,
And outputs information indicating the selected relation. The method according to claim 1,
Wherein the context information indicates a distance on the document to a relation appearing after the relation in the document for each relation indicated by the proposition of the document,
The processor comprising:
Selects a relation from the specific relation on the basis of a relation represented by the second graph included in the first graph and a distance between the relation indicated by the context information and the specific relation,
And outputs information indicating the selected relation. The method according to claim 1,
Wherein the processor outputs information indicating the predetermined relation when a predetermined relation is included in the specific relation. A method for supporting an idea of a user by an idea support device,
The idea support apparatus comprises:
Graph information indicating a structure of a first graph including a node representing an element constituting a proposition included in a document defining knowledge and a first kind edge representing a relation between the elements indicated by the proposition;
Retaining context information indicating a sequence of appearance of a relation in the document,
The method comprises:
Wherein the idea support apparatus comprises:
We accept input of proposition,
Extracts a relation between an element included in the proposition in which the input is accepted and the element indicated by the proposition in which the input is received,
Generating a second graph including a node indicating the extracted element and including a relation between the extracted elements as the first kind edge,
Refers to the graph information to specify a relation appearing after the relation in the document by referring to the context information when it is determined that the relation represented by the second graph is included in the first graph,
And outputting information indicating the specific relation. The method according to claim 6,
The document includes a plurality of propositions,
Wherein the node of the first graph comprises:
A first kind node indicating an entity included in the plurality of propositions,
And a second kind node indicating a relation of a first kind proposition, in which all elements and relations are contained in one element of another proposition in the plurality of propositions, as a proposition contained in the plurality of propositions,
Wherein the first graph includes a node indicating an element constituting the first kind of specification and a second kind of edge connecting the second kind of node,
The method comprises:
Wherein the idea support apparatus comprises:
Accepting input of a plurality of propositions,
In generating the second graph,
Determining an entity among the extracted elements as the first seed node,
Wherein when determining that the plurality of propositions that have accepted the input include the first kind proposition, the relation of the first kind proposition is determined to be the second kind node,
Connecting a node representing the element constituting the first type specification and the determined second kind node to the second type edge. The method according to claim 6,
Wherein the context information indicates the number of occurrences of relations appearing after the relation in the document for each relation indicated by the proposition of the document,
The method comprises:
Wherein the idea support apparatus comprises:
Selects a relation from the specific relation on the basis of the occurrence count of the specific relation indicated by the context information,
And outputting information indicating the selected relation. The method according to claim 6,
Wherein the context information indicates a distance on the document to a relation appearing after the relation in the document for each relation indicated by the proposition of the document,
The method comprises:
Wherein the idea support apparatus comprises:
Selects a relation from the specific relation on the basis of a relation represented by the second graph included in the first graph and a distance between the relation indicated by the context information and the specific relation,
And outputting information indicating the selected relation. The method according to claim 6,
Wherein the idea support apparatus outputs information indicating the predetermined relation when a predetermined relation is included in the specific relation.

Images