JP2018049587A - Estimation device, estimation method and estimation program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To easily estimate a user's interest affected by a causal relation.SOLUTION: An estimation device comprises: an identification section which identifies a group of a predetermined phenomenon, a result having a causal relation therewith and a polar character of the result from conceptual information which associates phenomena, results having causal relations therewith and polar characters to be indexes of interest affected by the results on a user; and an estimation section which estimates the result caused by the predetermined phenomenon and the interest affected by the result on the user on the basis of the group identified by the identification section.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an estimation device, an estimation method, and an estimation program.

  2. Description of the Related Art Conventionally, a technique for performing inference based on pre-formulated conditions and input conditions, such as an expert system, is known. As an example of such a technique, a technique for estimating a causal relationship is known.

JP 2009-059323 A

  However, in the above-described conventional technology, it may be difficult to estimate the interest caused by the causal relationship.

  For example, it is considered that the interest generated by the causal relationship changes in various ways depending on the user, the state of the user, and the like. As a result, even if a certain cause-and-effect relationship can be estimated, it is not possible to easily estimate what kind of interest is caused to the user by the cause-and-effect relationship.

  The present application has been made in view of the above, and an object thereof is to easily estimate a user's interest caused by a causal relationship.

  The estimation apparatus according to the present application includes a predetermined phenomenon from the concept information that associates a phenomenon, a result having a causal relationship with the phenomenon, and a polarity that is an index of interest that the result has on the user. Based on a combination of a result having a causal relationship with a predetermined phenomenon and the polarity of the result, a combination specified by the specific unit, a result when a predetermined phenomenon occurs, and the result And an estimation unit for estimating an interest on the user.

  According to one aspect of the embodiment, it is possible to easily estimate the user's interest caused by the causal relationship.

Drawing 1 is a figure for explaining an example of an operation effect which an estimating device concerning an embodiment exhibits. FIG. 2 is a diagram illustrating an example of a functional configuration included in the estimation apparatus according to the embodiment. FIG. 3 is a diagram illustrating an example of information registered in the structure information database according to the embodiment. FIG. 4 is a diagram illustrating an example of information registered in the polarity pair dictionary database according to the embodiment. FIG. 5 is a diagram illustrating an example of information registered in the semantic structure database according to the embodiment. FIG. 6 is a flowchart for explaining an example of the flow of generation processing executed by the estimation apparatus according to the embodiment. FIG. 7 is a flowchart for explaining an example of the flow of estimation processing executed by the estimation apparatus according to the embodiment.

  Hereinafter, an embodiment (hereinafter referred to as “embodiment”) for carrying out an estimation apparatus, an estimation method, and an estimation program according to the present application will be described in detail with reference to the drawings. In addition, the estimation apparatus, the estimation method, and the estimation program according to the present application are not limited by this embodiment. Further, in the following embodiments, the same parts and processes are denoted by the same reference numerals, and redundant description is omitted.

  Moreover, in the following description, as an example of the estimation process executed by the estimation apparatus 10, when a predetermined phenomenon occurs, a result caused by the predetermined phenomenon, that is, a result having a causal relationship with the predetermined phenomenon is estimated. In addition, an example of processing for estimating the interest that the result has on the user U01 will be described, but the embodiment is not limited thereto. For example, when an arbitrary phenomenon occurs at an arbitrary timing, the estimation apparatus 10 may estimate a result having a causal relationship with the phenomenon and an interest of the result on the user U01.

[1. (Concept of estimation processing)
First, the concept of the estimation process executed by the estimation apparatus 10 will be described with reference to FIG. Drawing 1 is a figure for explaining an example of an operation effect which an estimating device concerning an embodiment exhibits. For example, the estimation device 10 is realized by one or a plurality of information processing devices such as a server device or a cloud system, and via a network N (for example, see FIG. 2) such as a mobile communication network or a wireless local area network (LAN). The information processing apparatus can communicate with the terminal apparatus 200 used by the predetermined information distribution server 100 and the user U01.

  The information distribution server 100 is realized by one or a plurality of information processing apparatuses such as a server apparatus and a cloud system, and distributes various types of information via the network N. For example, the information distribution server 100 may distribute arbitrary information such as train route information and delay information, weather information, information related to finance such as stocks and exchanges, and news. In the following description, it is assumed that the information distribution server 100 is a server device that distributes train delay information.

  The terminal device 200 is an information processing device such as a mobile terminal such as a smartphone, a tablet terminal or a PDA (Personal Digital Assistant), or a notebook PC (Personal Computer). For example, when receiving a result of an estimation process (to be described as “estimation result” hereinafter) from the estimation device 10, the terminal device 200 receives the estimated result as a predetermined UI (User Interface). The content of the estimation result is provided to the user U01.

  When a predetermined phenomenon occurs, the estimation device 10 executes an estimation process for estimating a result caused by the phenomenon and an interest of the estimated result on the user U01. For example, the estimating apparatus 10 stores in advance a semantic structure database in which a phenomenon is linked to a result that can be generated by the phenomenon. Then, when a certain phenomenon occurs, the estimation device 10 estimates a pair of the phenomenon and the result having a causal relationship by tracing a result associated with the phenomenon, and the pair of the estimated phenomenon and the result. Estimates the interest that the user has on U01.

  However, the combination of the phenomenon and the result, that is, the interest generated by the causal relationship is considered to change variously according to the user, the state of the user, and the like. As a result, even if a certain cause-and-effect relationship can be estimated, it is not possible to easily estimate what kind of interest is caused to the user by the cause-and-effect relationship.

  Moreover, in the field of deep learning such as deep learning, the possibility of automatically extracting features from data without formulating conditions by humans is shown. However, it is unclear what kind of features the model generated by deep learning is learning, that is, what conditions are being formulated. In terms of difficulty in understanding, there is a limit to flexibility such as manual improvement.

[2. Regarding estimation processing executed by estimation apparatus according to embodiment]
Therefore, the estimation device 10 executes the following estimation process. For example, the estimating apparatus 10 associates a phenomenon, a result having a causal relationship with the phenomenon, and a semantic structure database 33 (that is, “concept information” that associates the polarity with which the result is an index of interest to the user. )). In addition, the estimation device 10 identifies a set of a predetermined phenomenon, a result having a causal relationship with the predetermined phenomenon, and a polarity of the result from the semantic structure database 33. Then, the estimation device 10 estimates the result when a predetermined phenomenon occurs and the interest that the result has on the user based on the specified set.

  Hereinafter, an example of a functional configuration and an effect of the estimation device 10 that realizes the above-described estimation process will be described with reference to the drawings. In the following description, as an example of the estimation process executed by the estimation apparatus 10, a state in which a certain event has occurred with respect to a certain subject is described as a “phenomenon”, and a result generated by the certain phenomenon is described as a “result”. In the following description, “subject”, “event”, and “result” may be collectively referred to as “element”.

[2-1. Example of functional configuration)
FIG. 2 is a diagram illustrating an example of a functional configuration included in the estimation apparatus according to the embodiment. As illustrated in FIG. 2, the estimation device 10 includes a communication unit 20, a storage unit 30, and a control unit 40. The communication unit 20 is realized by, for example, a NIC (Network Interface Card). The communication unit 20 is connected to the network N by wire or wireless, and transmits and receives questions and responses to and from the terminal device 200.

  The storage unit 30 is realized by, for example, a semiconductor memory device such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk. The storage unit 30 includes a structure information database 31, a polarity pair dictionary database 32, and a semantic structure database 33 that are various data for executing the estimation process. Hereinafter, an example of information registered in the structure information database 31, the polarity pair dictionary database 32, and the semantic structure database 33 will be described with reference to FIGS.

  In the structure information database 31, structure information indicating a concept structure between concepts (hereinafter referred to as “subjects”) that are main subjects of the phenomenon is registered. As a more specific example, the thesaurus data that systematically summarizes concepts having a hierarchical structure is registered in the structure information database 31. For example, in the structure information database 31, a word indicating the subject (may be a term composed of one or a plurality of words; the same applies hereinafter) is registered. For example, FIG. 3 is a diagram illustrating an example of information registered in the structure information database according to the embodiment. As shown in FIG. 3, information having items such as “first hierarchy”, “second hierarchy”, and “third hierarchy” is registered in the structure information database 31.

  Here, the word belonging to the “first hierarchy” is a word indicating the subject belonging to the highest hierarchy among the words registered in the structure information database 31, that is, the most abstract concept. The word belonging to the “second hierarchy” is a word indicating a concept that is a lower level of the concept indicated by the associated word belonging to the “first hierarchy”. The word belonging to the “third hierarchy” is a word indicating a concept belonging to a lower level of the concept indicated by the associated word belonging to the “second hierarchy”. In the example shown in FIG. 3, an example from “first layer” to “third layer” is described, but the thesaurus data having an arbitrary number of layers is registered in the structure information database 31. Good.

  For example, in the example illustrated in FIG. 3, the first hierarchy “public transportation”, the second hierarchy “train”, and the third hierarchy “route # 1” are registered. Such information includes a concept indicated by the word “train” as a subordinate concept of the concept indicated by the word “public transportation”, and “route # 1” as a subordinate concept of the concept indicated by the word “train”. This means that there is a concept indicated by the word.

  In the polarity pair dictionary database 32, information on a polarity pair dictionary (that is, “pair information”) in which events or results having a concept of a pair are associated with each other is registered. For example, in the polar pair dictionary database 32, pair information in which a word indicating a certain event or result and a word indicating a concept opposite to the event or result are registered.

  For example, FIG. 4 is a diagram illustrating an example of information registered in the polarity pair dictionary database according to the embodiment. As shown in FIG. 4, “first element” and “second element” are registered in the polarity pair dictionary database 32 in association with each other. Here, the “first element” and the “second element” are words indicating opposite concepts, and are words indicating a concept that is a phenomenon event or a concept that is a result.

  For example, in the example shown in FIG. 4, a first element “delay” and a second element “on-time” are registered in the polarity pair dictionary database 32 in association with each other. Such information indicates that the first element “delay” and the second element “on-time” are words having a concept of being paired (opposite).

  In the semantic structure database 33, conceptual information in which a phenomenon, a result having a causal relationship with the phenomenon, and a polarity as an index of interest that the result has on the user is registered. More specifically, in the semantic structure database 33, the subject of the phenomenon and the event related to the subject are linked. In the semantic structure database 33, each event is associated with a result having a causal relationship with the event. In the semantic structure database 33, the polarity indicating the interest of each result on the user is registered in association with the result, and the polarity indicating the interest of each event on the user corresponds to the event. It is registered with it.

  For example, FIG. 5 is a diagram illustrating an example of information registered in the semantic structure database according to the embodiment. In the example illustrated in FIG. 5, an example in which the conceptual relevance of elements registered in the semantic structure database 33 is realized by an RDB (Relational Data Base) is described.

  As shown in FIG. 5, information having items such as “element ID (Identifier)”, “element”, “attribute”, “polarity”, and “structure” is registered in the semantic structure database 33. Here, the “element ID” is an identifier for identifying an element (subject, event, and result) registered in the semantic structure database 33. The “element” is a word registered as an element in the semantic structure database 33. The “attribute” is information indicating whether the associated element is a subject, an event, or a result.

  The “polarity” is a value that serves as an index of the interest (profit or disadvantage) that the associated element has on the user U01. For example, the larger the polarity value, the more the associated element indicates the concept of giving a benefit to the user U01. The smaller the polarity value, the more the associated element is associated with the user U01. The concept of giving disadvantages shall be shown. When the value of the polarity is zero, the concept that the associated element does not give a profit or a disadvantage to the user U01 is shown.

  The “structure” is an element ID of another element having a connection with the associated element. That is, the “structure” is information indicating a causal relationship between a subject and an event related to the subject and a phenomenon and a result by linking each element.

  For example, in the example shown in FIG. 5, element ID “1”, element “train”, attribute “subject”, polarity “0”, and structure “2” are registered in association with each other. Such information indicates that the element indicated by the element ID “1” is “train” and the attribute is “subject”. Such information also indicates that the polarity of the element “train” is “0” and does not give interest to the user U01. Such information indicates that the element “train” is associated with the element indicated by the element ID “2”. As will be described later, the estimating apparatus 10 estimates each result registered in the semantic structure database 33 based on the value of the item “structure”, thereby estimating a result having a causal relationship with a certain phenomenon. It is possible.

  Returning to FIG. 2, the description will be continued. The control unit 40 is stored in a storage device inside the estimation device 10 by, for example, a CPU (Central Processing Unit), an MPU (Micro Processing Unit), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or the like. The various programs are executed by using a storage area such as a RAM as a work area. In the example illustrated in FIG. 2, the control unit 40 includes a collection unit 41, a determination unit 42, a grant unit 43, a structuring unit 44, an acquisition unit 45, a specification unit 46, and an estimation unit 47 (hereinafter collectively referred to as each processing unit). 41 to 47.).

  The connection relationship between the processing units 41 to 47 included in the control unit 40 is not limited to the connection relationship illustrated in FIG. 2, and may be another connection relationship. Each of the processing units 41 to 47 realizes and executes functions and actions (for example, FIG. 1) of generation processing and estimation processing as described below, but these are functional units arranged for explanation. It does not matter whether it matches the actual hardware elements or software modules. That is, the estimation device 10 may realize and execute the generation process and the estimation process in arbitrary functional units as long as the functions and actions of the following generation process and estimation process can be realized and executed.

  In the following description, it is assumed that a generation process for generating the semantic structure database 33 is performed by each of the processing units 41 to 44 prior to the estimation process. In the following description, it is assumed that the estimation process is executed by the processing units 45 to 47 using the semantic structure database 33 generated by the generation process.

[2-2. Example of effects in the generation process]
Hereinafter, the contents of the generation process executed and realized by each of the processing units 41 to 44 will be described using the flowchart shown in FIG. FIG. 6 is a flowchart for explaining an example of the flow of generation processing executed by the estimation apparatus according to the embodiment.

  The collection unit 41 collects abstract concepts including phenomena and results (step S101). For example, the collection unit 41 generates a phenomenon (ie, “event”) and a result having a causal relationship with the phenomenon (ie, “action”) from any text content such as news, blogs, and microblog articles on the Internet. ) And collect. More specifically, the collection unit 41 extracts a set of a word as a subject and a word indicating an event related to the subject as a word set indicating a phenomenon from arbitrary text content. Further, the collection unit 41 extracts a word having a causal relationship with the phenomenon indicated by the extracted word set, that is, a word indicating a result resulting from the occurrence of the phenomenon, as a word indicating the result.

  For example, the collection unit 41 analyzes text content on the Internet by a character analysis technique such as morphological analysis. Then, the collection unit 41 collects three words indicating a subject, an event, and a result, that is, a word composed of the subject and the event, and a result generated by the phenomenon. For example, as shown in FIG. 1A, the collection unit 41 includes a combination of a phenomenon consisting of a subject “train” and an event “delay” and a result “delay” resulting from the phenomenon, a subject “company”, and A combination of a phenomenon consisting of an event “being late” and a result “angry” resulting from the phenomenon, a phenomenon consisting of the subject “alarm” and an event “setting”, and a result resulting from the phenomenon “sleeping” Collect a pair.

  Further, the determination unit 42 determines whether or not the action of the user U01 affects the result (step S102). For example, the determination unit 42 analyzes a triple of words indicating the phenomenon collected by the collection unit 41 and the result when the phenomenon occurs, and the personal action of the user U01 can eliminate the causal relationship between the phenomenon and the result. It is determined whether or not. For example, as shown in FIG. 1B, the determination unit 42 selects the word train “train”, “delay”, and “delay” collected by the collection unit 41, and the selected triple is It is determined whether or not the causal relationship shown can be resolved by individual power. Here, the train delay is considered to be a phenomenon that cannot be operated by personal power. Therefore, the determination unit 42 determines that the causal relationship indicated by the triples “train”, “delay”, and “delay” cannot be resolved by individual power.

  Similarly, the determination unit 42 selects the triple “company”, “late”, and “get angry”, and determines whether or not the causal relationship indicated by the selected triple can be resolved by individual power. Here, an event such as being late for the company is considered to be a phenomenon that can be operated by individual power. Therefore, the determination unit 42 determines that the causal relationship indicated by the triples “company”, “being late”, and “get angry” can be resolved by individual power.

  Further, the determination unit 42 selects the triples “alarm”, “setting”, and “sleeping”, and determines whether or not the causal relationship indicated by the selected triples can be resolved by individual power. Here, an event such as setting an alarm is considered to be a phenomenon that can be operated by individual power. Therefore, the determination unit 42 determines that the causal relationship indicated by the triples “wake-up”, “setting”, and “sleeping” can be resolved by individual power.

  In addition, the determination part 42 should just implement | achieve the determination mentioned above based on arbitrary learning techniques, such as deep learning and an associative memory model, for example. For example, the determination unit 42 uses a model that learns in advance a combination of a subject and an event that can be operated by an individual's power, a combination of a subject and an event that cannot be operated by an individual's power, and the like. What is necessary is just to determine whether the causal relationship which this word shows can be eliminated by an individual's power.

  The assigning unit 43 assigns upstream information according to the determination result by the determining unit 42 (step S103). Here, the run-up information is information indicating whether or not a certain result can be traced back to the phenomenon or event that caused the result. For example, when the determination unit 42 determines that a certain causal relationship can be resolved by individual power, the granting unit 43 indicates an event or phenomenon from a word indicating the result among three words indicating the causal relationship. The upstream information that the word can be upstream is given.

  Further, for example, when the determination unit 42 determines that a certain causal relationship cannot be resolved by the individual's power, the granting unit 43 determines an event or phenomenon from a word indicating the result among three words indicating the causal relationship. Ascending information indicating that the ascending word cannot be ascended is given. That is, the granting unit 43 can eliminate the cause of the causal relationship by setting information indicating whether or not the element indicating the phenomenon can be traced from the element indicating the result to the triplet word. Set the indicator.

  For example, the determination unit 42 determines that the causal relationship indicated by the triples “train”, “delay”, and “delay” cannot be resolved by individual power. In such a case, as shown in FIG. 1C, the assigning unit 43 can trace from the element “delay” to the element “delay”, but from the element “delay” to the element “delay”. Set up the upstream information that cannot be traced.

  In addition, for example, the determination unit 42 determines that the causal relationship indicated by the triples “company”, “being late”, and “being angry” can be resolved by individual power. In such a case, as shown in FIG. 1D, the granting unit 43 can trace from the element “late” to the element “get angry”, and from the element “get angry” to the element “late” Set up the run-up information to the effect that it can be traced to “Yes”.

  The granting unit 43 sets up the upstream information indicating whether or not the elements can be traced from the results to the events for all triples for which the judgment unit 42 has determined whether the causal relationship can be resolved by individual power. Will be. Here, for example, the granting unit 43 may set information indicating the extent to which the causal relationship can be resolved by individual power as the upstream information. In such a case, when the value of the run-up information is higher than a predetermined threshold, the estimation apparatus 10 executes a process of tracing the element from the result to the event on the assumption that there is a high possibility that the information can be resolved by an individual. It will be.

  Subsequently, the assigning unit 43 assigns polarity to each element (step S104). For example, the granting unit 43 determines whether each element collected as a triple by the collecting unit 41 indicates a concept that gives profit to the user U01 or a concept that gives a disadvantage. And the provision part 43 gives the positive value according to the grade of profit to an element as a polarity, when showing the concept which gives profit with respect to the user U01. On the other hand, when the imparting unit 43 indicates a concept that causes a disadvantage to the user U01, the imparting unit 43 assigns a negative value corresponding to the degree of the disadvantage to the element as a polarity. In addition, the provision part 43 sets zero as a polarity value, when the concept which an element shows has no interest with respect to the user U01.

  For example, as illustrated in (E) of FIG. 1, the granting unit 43 determines whether or not the element “delay” has a disadvantage for the user U01. For example, when the grant unit 43 determines that the element “delay” has a disadvantage for the user U01, “(−3)” is set as a polarity value according to the degree of the disadvantage. Give. Further, for example, when the granting unit 43 determines that the element “company” does not have an interest for the user U01, the granting unit 43 assigns “(0)” to the element “company” as a polarity value.

  The structuring unit 44 generates the semantic structure database 33 in which the abstract concept is structured into a graph (step S105). For example, the structuring unit 44 registers the triple words collected by the collecting unit 41 in the semantic structure database 33 as elements indicating subjects, events, and results, respectively. In addition, the structuring unit 44 can change the value of “structure” between each element so that the event element can be traced from the main element and the resulting element can be traced from the event element. sign up.

  In addition, the structuring unit 44 traces from the resulting element to the event element when the upstream information indicating that the result element can be traced to the event element is set. Register the value of “structure” between elements. In addition, the structuring unit 44 traces from the resulting element to the event element when the upstream information indicating that the result element cannot be traced to the event element is set. Register the value of “structure” between elements so that it is not possible.

  For example, the structuring unit 44 assigns the element ID of the element “delay” to the item “structure” of the element “delay” so that the element “delay” can be traced between the elements “delay”. The value is registered in the semantic structure database 33. On the other hand, the structuring unit 44 sets the element ID of the element “delay” in the item “structure” of the element “delay” so that the element “delay” cannot be traced from the element “delay” to the element “delay”. It is not associated as a value.

  In addition, for example, the structuring unit 44 sets the element ID of the element “late” to the item “elemented” so that the element can be traced from the element “late” to the element “angry”. While registering in the semantic structure database 33 as the value of “structure”, the element ID of the element “delayed” is assigned to the element “delayed” so that the element can be traced from the element “angry” to the element “delayed”. It is registered in the semantic structure database 33 as the value of the item “structure” of “angered”.

  In addition, the structuring unit 44 sets the polarity value provided by the applying unit 43 for each element. For example, the structuring unit 44 registers the element “delay” in the semantic structure database 33 in association with the polarity “−3”.

  Here, the structuring unit 44 performs graph structuring of the semantic structure by collecting elements indicating the same concept. For example, the word “late” included in the triple “train”, “delay”, “late” and the word “late” included in the triple “company”, “late”, “get angry” Are words indicating the same concept. In such a case, the structuring unit 44 registers the word “late” included in each triplet in the semantic structure database 33 as one element.

  As a result of such processing, the structuring unit 44 generates a semantic structure database 33, which is information indicating the structure of the concept indicated by each word, from the triple words as shown in FIG. Can do. More specifically, the structuring unit 44 associates a phenomenon composed of a subject and an event, a result having a causal relationship with the phenomenon, and a polarity that is an index of interest that the result has on the user. Can be generated. In such a semantic structure database 33, by tracing another element linked to the element from an arbitrary element, a phenomenon including a subject and an event related to the subject, and a result having a causal relationship with the phenomenon Etc. can be specified.

  In the following description, it is assumed that each element is linked to the semantic structure database 33 in the order of element “train”, element “delay”, element “delay”, and element “get angry”. In the following description, it is assumed that each element is linked to the semantic structure database 33 in the order of the element “alarm”, the element “do not ring”, the element “late”, and the element “get angry”.

[2-3. Example of effects in estimation process)
Next, the content of the estimation process executed and realized by each of the processing units 45 to 47 will be described using the flowchart shown in FIG. FIG. 7 is a flowchart for explaining an example of the flow of estimation processing executed by the estimation apparatus according to the embodiment.

  The acquisition unit 45 acquires current status information (step S201). For example, as illustrated in (G) of FIG. 1, the acquisition unit 45 acquires, from the information distribution server 100, a notification that the train is delayed as current information. In such a case, the acquisition unit 45 transmits the current status information to the specifying unit 46.

  The specifying unit 46 specifies a phenomenon corresponding to the current state and a result associated with the phenomenon (step S202). More specifically, the specifying unit 46 determines whether a predetermined result is obtained from conceptual information in which a phenomenon, a result having a causal relationship with the phenomenon, and a polarity that is an index of interest that the result has on the user are associated with each other. A set of a phenomenon, a result having a causal relationship with a predetermined phenomenon, and a polarity of the result is specified. For example, the specifying unit 46 specifies a phenomenon that is a combination of a predetermined subject and a predetermined event related to the predetermined subject as the predetermined phenomenon.

  For example, when the current state information indicating the phenomenon is acquired, the specifying unit 46 sets a combination of an element indicating the concept that is the subject of the phenomenon and an element indicating the concept that is the event included in the phenomenon to a predetermined value. Identify as a phenomenon. For example, when receiving the current state information that the train is delayed, the specifying unit 46 includes an element “train” indicating the main concept in the phenomenon indicated by the current state information, and an element “delay” indicating the concept that is an event. And specify. Then, as shown in FIG. 1H, the specifying unit 46 specifies a pair of the element “train” and the element “delay” associated with the “structure” value from the semantic structure database 33 as a phenomenon.

  Subsequently, the specifying unit 46 specifies a result having a causal relationship with the specified phenomenon by tracing the linked element from the element specified as the phenomenon, that is, the element indicating the subject and the event. For example, as illustrated in (I) of FIG. 1, the specifying unit 46 specifies an element “delayed” associated with the specified element “delay” as an element indicating the result.

  Further, the identifying unit 46 determines whether or not the element “late late” indicates not only “result” but also “event” based on the attribute of the identified element “late late”. When the element “late” is “event”, the specifying unit 46 follows the element indicating the result, which is associated with the element “late”. As a result, as shown in FIG. 1J, the specifying unit 46 specifies the result “angry” that occurs when the event indicated by the element “late” occurs. As a result, the identifying unit 46 identifies the cause and effect of the causal relationship “being late” when the “train” is “delayed” and the cause (phenomenon or event) and the result of the causal relationship “being late” and “getting angry”. it can.

  The identifying unit 46 also sets the polarity value “0” assigned to the element “train”, the polarity value “−3” assigned to the element “delay”, and the polarity assigned to the element “delay”. Value “−4” and the polarity value “−4” assigned to the element “angered”. That is, the identification unit 46 identifies the event and the result from the semantic structure database 33, and the polarity that the event serves as an index of the interest of the user and the index of the interest of the result that the user exerts on the user. Are respectively specified.

  Subsequently, the estimation unit 47 estimates the causal relationship and the interest on the user based on the identification result by the identification unit 46 (step S203). That is, the estimation unit 47 estimates the result when a predetermined phenomenon occurs and the interest that the result has on the user U01 by tracing other elements from the phenomenon specified by the specifying unit 46.

  For example, the estimation unit 47 determines the result when a predetermined phenomenon occurs based on the group specified by the specification unit 46 and the interest that the result has on the user with the polarity specified by the specification unit 46. Estimate based on the value. More specifically, the estimation unit 47 specifies the interest to the user based on the polarity of the event specified by the specification unit 46 and the polarity of the result.

  A more specific example will be described. For example, the specifying unit 46 specifies the element “train”, the element “delay”, the element “delay”, and the element “angered”. In such a case, the estimation unit 47 determines from the attributes registered in the semantic structure database 33 that the subject of the phenomenon is “train”, the phenomenon event is “delayed”, and the results are “late” and “anger”. Is specified. Then, from the relationship between the specified subject, the event, and the result, the estimating unit 47 determines that the phenomenon that “the train is delayed”, the result that is caused by the phenomenon “late”, and the result that is caused by the result is “angry” Is estimated. Note that the process of estimating a phenomenon or result from such elements can be realized by an associative memory model or the like.

  The identifying unit 46 also sets the polarity value “0” assigned to the element “train”, the polarity value “−3” assigned to the element “delay”, and the polarity assigned to the element “delay”. Value “−4” and the polarity value “−4” assigned to the element “angered”. In such a case, the estimation unit 47 calculates the total value of each polarity for each set of the phenomenon and the result, and the set of the event and the result that is the result, and based on the calculated total value of the polarity, Estimate the interests to users.

  For example, the estimation unit 47 sets the polarity value “0” assigned to the element “train”, the polarity value “−3” assigned to the element “delay”, and the polarity assigned to the element “delay”. “−7”, which is the sum of the values of “−4”, is used as an index of interest that the causal relationship such as “the train is delayed because it is delayed” affects the user. For example, the estimating unit 47 adds “−8”, which is the sum of the polarity value “−4” assigned to the element “get late” and the polarity value “−4” assigned to the element “angered”. "Is an index of the interest that a causal relationship such as" being angry because it is late "affects the user.

  As a result, the estimation unit 47 can easily estimate the causal relationship between the phenomenon related to the user and the result and the interest that the causal relationship has on the user from the current state information distributed by the information distribution server 100. For example, the estimating unit 47 traces the causal relationship such as “being late because the train is delayed” or the causal relationship “being angry because it is late” from the phenomenon “the train is delayed” from the semantic structure database 33. Can be estimated. Furthermore, the estimation part 47 can estimate the index of the interest which the estimated causal relationship exerts on the user using the value of the polarity given to the event or the result. As a result, the estimation device 10 facilitates estimation of the causal relationship and the interest that the causal relationship has on the user.

  Continue the explanation. Here, the estimation unit 47 determines whether or not the sum of the polarities assigned to the phenomenon and the result is negative (step S204). That is, the estimation unit 47 determines whether or not the estimated causal relationship has a disadvantage for the user.

  If the sum of the polarities is negative (step S204: Yes), the specifying unit 46 traces the connection destination event or result from the result (step S205). More specifically, the specifying unit 46 determines whether or not the predetermined phenomenon and the predetermined phenomenon from the conceptual information further associated with the upstream information indicating at least one of the presence and the degree of the influence of the user's action on the result. , A set of a result having a causal relationship with a predetermined phenomenon, information on the result, and a polarity of the result is specified. For example, when specifying a causal relationship, the specifying unit 46 refers to the value of the structure associated with the element indicating the specified result, and specifies the element indicated by the referred value. Then, when the element identified from the element indicating the result is an element of the identified causal relationship, the identifying unit 46 determines that the identified causal relationship can be resolved by individual power.

  For example, in the case illustrated in FIG. 1, the identifying unit 46 searches for an element traced from the element “late”. Here, in the case shown in FIG. 1, it is not possible to go back from the element “delay” to the element “delay”. That is, in the case shown in FIG. 1, in the element “late late”, the upstream information that the causal relationship can be resolved by the user's action is not registered. Therefore, the specifying unit 46 specifies an event or a result that can eliminate the result indicated by the element “delayed” by tracing elements other than the element “delayed” from the element “delayed”. That is, the specifying unit 46 specifies the result of the specified causal relationship and another phenomenon having the causal relationship based on the upstream information. In the example illustrated in FIG. 1, it is assumed that the specifying unit 46 has traced the element “does not sound” from the element “late”.

  Here, the estimation unit 47 determines whether or not a polarity pair exists in the traced event or result (step S206). If the polarity pair exists (step S206: Yes), Based on the result, a phenomenon in which the polarity is positive and the result are estimated (step S207). On the other hand, when the polarity pair does not exist (step S206: No), the estimation unit 47 determines whether or not the element can be further traced (step S209). When the polar pair can be traced (step S209: Yes), the step S205 is executed. Moreover, the estimation part 47 complete | finishes a process, when it cannot follow (step S209: No).

  That is, when the result specified by the specifying unit 46 is disadvantageous to the user, the estimating unit 47 uses the polarity pair dictionary database 32 that associates the events or results having the concept of a pair, The result or event for reversing the interest when the phenomenon specified by the specifying unit 46 occurs is estimated. More specifically, the estimation unit 47 searches the semantic structure database 33 for a phenomenon that corresponds to the result included in the set specified by the specifying unit 46 and is not included in the specified set. Using the pair dictionary database 32, a phenomenon in which an event included in the searched phenomenon is changed to an event having a concept of a pair is estimated.

  For example, as shown in (K) of FIG. 1, when the specifying unit 46 follows the element “does not sound” from the element “late”, the estimation unit 47 indicates that the element “does not sound” indicates that the polarity pair dictionary database 32. It is determined whether or not it is registered. When the element “does not sound” is registered in the polarity pair dictionary database 32 as shown in (L) in FIG. 1, the estimating unit 47, as shown in (M) in FIG. An element associated with “does not sound”, that is, an element “ring” having the concept of the element “does not sound” and a counter electrode is specified. Then, as shown in (N) of FIG. 1, the estimation unit 47 estimates a phenomenon that can avoid the result indicated by the element “late” by replacing the element “do not sound” with the element “ring”. .

  For example, if the information shown in FIG. 1 is registered in the semantic structure database 33, the element “alarm” and the phenomenon “alarm sound” from the element “alarm” and the element “alarm” when the element “sound” is replaced with the element “sound”. Can be estimated. Therefore, the estimation unit 47 estimates an action that can avoid a result of being delayed by using a phenomenon of sounding an alarm when a phenomenon such as a train delay occurs. For example, the estimation unit 47 estimates a phenomenon in which the alarm is sounded early by generating a variation of the phenomenon in which the alarm is sounded. And the estimation part 76 performs step S204, and determines whether the estimated phenomenon and the polarity of a result are negative.

  Further, when the sum of the polarities is zero or positive (step S204: No), the estimation unit 47 outputs the estimation result (step S208) and ends the process. For example, the estimation device 10 can generate not only a causal relationship such as “being late because the train is delayed” or “being angry because it is late” from the current information that “the train is delayed”, Estimate behaviors such as “I can avoid being late”. And the estimation part 47 transmits an estimation result to the terminal device 200 which the user U01 uses as shown to (O) in FIG.

  For example, the estimation device 10 may control the terminal device 200 according to the estimation result. For example, when the estimation device 10 estimates a phenomenon such as “ring the alarm early,” the estimation device 10 may instruct the terminal device 200 to sound the alarm at a time earlier than the set time by a predetermined time. .

[2-4. Processing using structural information database]
In the above description, the causal relationship and the result are used by tracing the elements registered in the semantic structure database 33, that is, the elements indicating the subject, the phenomenon, and the result, each having a polarity. Explained the process of estimating the interest to the person. In the above description, when the estimated causal relationship and the result are disadvantageous to the user, the result is disadvantageous by using the polarity pair dictionary database 32 in which the concepts of opposite polarities are associated. A process for estimating a phenomenon that avoids the problem has been described.

  Here, the estimating apparatus 10 may use a thesaurus registered in the structure information database 31 in order to expand the concept information registered in the semantic structure database 33. Specifically, the estimation apparatus 10 may change the subject of the identified phenomenon to another subject using the structure information indicating the conceptual structure of the subjects, and estimate the result when the subject of the phenomenon is changed. Good.

  For example, the estimation apparatus 10 searches the structural information database 31 for the main element “train” as shown in FIG. In addition, the estimation apparatus 10 identifies another concept “bus” belonging to “public transportation”, which is a superordinate concept of the main “train”. Then, the estimation device 10 may estimate a causal relationship such as “being angry because the bus is late if the bus is delayed” by replacing the element “train” with “bus” in the semantic structure database 33.

  A more specific example will be described. When the estimation apparatus 10 accepts the current state information that “the train is delayed”, not only the causal relationship such as “because the train is delayed” but “the bus is delayed” It is possible to estimate the causal relationship such as “because I am late”. The result of such estimation processing may be provided to the user U01 as it is, or may be used for controlling the terminal device 200 or the like.

[3. (Modification)
The estimation apparatus 10 according to the above-described embodiment may be implemented in various different forms other than the above-described embodiment. Therefore, in the following, another embodiment of the estimation device 10 will be described.

[3-1. About estimation)
For example, the estimation device 10 may estimate not only phenomena and results but also user behavior. For example, when it is estimated that the specified causal relationship is disadvantageous for the user, the estimating apparatus 10 may estimate the user's behavior so that the result included in the causal relationship can be eliminated. Good. For example, the estimation device 10 registers an event or result related to the user's behavior in the semantic structure database 33, and avoids such a disadvantage when any phenomenon that adversely affects the user occurs. An event to be performed may be estimated, and a user's behavior for avoiding a disadvantage may be estimated from the estimated event.

[3-2. About the run-up information)
Whether the estimation apparatus 10 described above can trace an item “structure” indicating another element that can be traced from a certain element in association with each element, so that it can be traced from a certain cause to an event that causes the cause. It was possible to determine whether or not. However, the embodiment is not limited to this. For example, the estimation device 10 may store information indicating whether or not it is possible to go up to a causal event from the result.

  Further, for example, the estimation device 10 may store upstream information indicating the extent of room that is affected by the actions of the parties. In such a case, for example, the estimation apparatus 10 may estimate how much effort the user can avoid, based on the degree indicated by the upstream information. In addition, the estimation device 10 may prompt the user to avoid a disadvantage.

[3-3. Use of estimation results)
Here, the estimation result by the estimation apparatus 10 is not only provided to the user U01, but can be used for arbitrary processing. For example, the estimation result by the estimation device 10 may be used when providing an arbitrary service to the user U01 such as IoT (Internet of Things) or PIM (Personal Information Manager). For example, when the estimating apparatus 10 estimates a causal relationship such as “being late because the train is delayed” and estimating an action such as “raise the alarm early” in order to avoid disadvantages, based on the estimation result Thus, it may be used for processing such as quickly sounding the alarm of the user U01 or turning on the electricity of the house.

[3-4. Other embodiments]
In addition, the said embodiment is only an illustration and this invention includes what is illustrated below and other embodiment other than that. For example, the functional configuration, data structure, processing order and contents shown in the flowchart in the present application are merely examples, and the presence / absence of each element, the order of arrangement and processing execution, and specific contents can be changed as appropriate. . For example, the generation process and the estimation process described above can be realized as an apparatus, a method, and a program in a cloud system in addition to the estimation apparatus 10 as exemplified in the above embodiment.

  Moreover, the structure which implement | achieves each process part 41-47 which comprises the estimation apparatus 10 by a further independent apparatus is also common. Similarly, the configuration of the present invention is flexible, such as by realizing each means shown in the above embodiment by calling an external platform or the like with an API (application program interface) or network computing (so-called cloud). Can be changed. Furthermore, each element such as means relating to the present invention may be realized by other information processing mechanisms such as a physical electronic circuit as well as a computer control unit.

  The estimation device 10 may be realized by a front-end server that transmits and receives information to and from the terminal device 200 and a back-end server that executes generation processing and estimation processing. For example, the back-end server always executes the generation process and the estimation process described above. When the back-end server estimates a causal relationship that is disadvantageous to the user or an event or action that avoids the disadvantageous result, the back-end server notifies the front-end server of the estimation result. In such a case, the front-end server controls the terminal device 200 used by the user U01 and various devices related to IoT based on the estimation result.

[4. effect〕
As described above, the estimation apparatus 10 determines a predetermined phenomenon from conceptual information in which a phenomenon, a result having a causal relationship with the phenomenon, and a polarity that is an index of interest that the result has on the user are associated with each other. And a set of a result having a causal relationship with a predetermined phenomenon and a polarity of the result. Then, the estimation device 10 estimates the result when a predetermined phenomenon occurs and the interest that the result has on the user based on the specified set. For this reason, the estimation apparatus 10 can easily estimate the user's interest caused by the causal relationship.

  In addition, the estimation device 10 identifies a phenomenon that is a combination of a predetermined subject and a predetermined event related to the predetermined subject as the predetermined phenomenon. For this reason, the estimation apparatus 10 can easily estimate the causal relationship between a phenomenon that is a combination of various subjects and events and a result caused by the phenomenon.

  Further, the estimation apparatus 10 can obtain a predetermined phenomenon, the polarity of the event included in the phenomenon, the predetermined phenomenon, and the causality from the concept information further associating the polarity that is an index of interest that the event has on the user. A set of related results and result polarity is identified. And the estimation apparatus 10 estimates the interest which it has with respect to a user based on the polarity of the event contained in the specified group, and the polarity of a result. For this reason, the estimation apparatus 10 can easily estimate the interest that a certain phenomenon has on the user and the interest that a certain result has on the user.

  The estimation device 10 uses the polarity pair dictionary database 32 in which events having the concept of a pair or the results are associated with each other when the result estimated based on the specified set is disadvantageous to the user. Thus, a result or event for reversing the interest when a predetermined phenomenon occurs is estimated. For example, the estimation apparatus 10 searches the phenomenon corresponding to the result included in the specified set and is not included in the specified set from the concept information, and searches using the polarity pair dictionary database 32. A phenomenon in which an event included in the phenomenon is changed to an event having a concept of a pair is estimated. For this reason, for example, when an event or result that causes a disadvantage to the user occurs, the estimation device 10 can estimate a phenomenon that avoids or reduces the event or result.

  Moreover, the estimation apparatus 10 changes the subject of the identified phenomenon to another subject using the structure information database 31 indicating the conceptual structure of the subject, and estimates the result when the subject of the phenomenon is changed. For this reason, the estimation apparatus 10 extends the concept information registered in the semantic structure database 33 and enables estimation of various causal relationships.

  In addition, the estimation device 10 identifies another phenomenon having a causal relationship with the identified result based on the upstream information indicating at least one of the presence or the degree of the influence of the user's action on the result. . For this reason, the estimation apparatus 10 can implement | achieve a user's action estimation which avoids the event and result which become disadvantageous.

  As described above, some of the embodiments of the present application have been described in detail with reference to the drawings. However, these are merely examples, and various modifications, including the aspects described in the disclosure section of the invention, based on the knowledge of those skilled in the art, It is possible to implement the present invention in other forms with improvements.

  Moreover, the above-mentioned “section (module, unit)” can be read as “means”, “circuit”, and the like. For example, the estimation unit can be read as estimation means or an estimation circuit.

DESCRIPTION OF SYMBOLS 10 Estimation apparatus 20 Communication part 30 Storage part 31 Structure information database 32 Polarity pair dictionary database 33 Semantic structure database 40 Control part 41 Collection part 42 Judgment part 43 Assignment part 44 Structure part 45 Acquisition part 46 Identification part 47 Estimation part 100 Information distribution Server 200 terminal device

Claims (12)

From a conceptual information that associates a phenomenon, a result that has a causal relationship with the phenomenon, and a polarity that is an index of interest that the result has on the user, the predetermined phenomenon and the causal relationship with the predetermined phenomenon A specific part that identifies a set of the result to have and the polarity of the result;
An estimation apparatus comprising: an estimation unit configured to estimate a result when a predetermined phenomenon occurs based on a group identified by the identification unit and an interest of the result on the user. The estimation device according to claim 1, wherein the specifying unit specifies a phenomenon that is a combination of a predetermined subject and a predetermined event related to the predetermined subject as the predetermined phenomenon. The specific unit is configured to obtain the predetermined phenomenon, the polarity of the event included in the phenomenon, and the predetermined phenomenon from the concept information in which the polarity that is an index of the interest that the event has on the user is further associated. Identify the pair of the causal result and the polarity of the result,
The estimation unit according to claim 2, wherein the estimation unit estimates an interest on the user based on a polarity of an event included in a set identified by the identification unit and a polarity of a result. apparatus. When the result estimated based on the set specified by the specifying unit has a disadvantage to the user, the estimation unit uses paired information that associates an event having a concept or a result with each other. The estimation device according to claim 2, wherein a result or an event for reversing the interest when the predetermined phenomenon occurs is estimated. The estimation unit searches the concept information for a phenomenon corresponding to a result included in the set specified by the specifying unit, and is not included in the set specified by the specifying unit. The estimation apparatus according to claim 4, wherein an estimation is performed by estimating a phenomenon in which an event included in the searched phenomenon is changed to an event having a concept of a pair. The estimation unit changes the subject of the phenomenon specified by the specifying unit to another subject using the structural information indicating the conceptual structure of the subjects, and estimates the result when the subject of the phenomenon is changed. The estimation apparatus according to any one of claims 2 to 5, characterized in that The specifying unit specifies another phenomenon having a causal relationship with the specified result based on the upstream information indicating at least one of the presence or the degree of the influence of the user's action on the result. ,
The estimation apparatus according to claim 1, wherein: An estimation method executed by an estimation device,
From a conceptual information that associates a phenomenon, a result that has a causal relationship with the phenomenon, and a polarity that is an index of interest that the result has on the user, the predetermined phenomenon and the causal relationship with the predetermined phenomenon A specific step of identifying a set of the result to have and the polarity of the result;
An estimation method comprising: an estimation step of estimating a result when a predetermined phenomenon occurs based on the set identified in the identification step and an interest of the result on the user. From a conceptual information that associates a phenomenon, a result that has a causal relationship with the phenomenon, and a polarity that is an index of interest that the result has on the user, the predetermined phenomenon and the causal relationship with the predetermined phenomenon A specific procedure for identifying a set of results to have and the polarity of the results;
An estimation program for causing a computer to execute a result when a predetermined phenomenon occurs based on the group specified in the specific procedure and an estimation procedure for estimating an interest of the result on the user. From the concept information that correlates the phenomenon composed of the subject and the event related to the subject, the result that has a causal relationship with the phenomenon, and the polarity that is an index of interest that the result has on the user. A specifying unit for specifying a set of a phenomenon, a first result having a causal relationship with the predetermined phenomenon, and a polarity of the first result;
An estimation unit that estimates a result when a predetermined phenomenon occurs based on a set specified by the specifying unit and an interest of the result on the user;
In the case where the first result is also an element indicating an event, the specifying unit determines, from the concept information, the predetermined phenomenon, the first result, the polarity of the first result, the first result, Identifying a set of the second result associated with the result of 1 and the polarity of the second result,
The estimation unit estimates an interest on the user based on the polarity of the first result and the polarity of the second result;
An estimation apparatus characterized by that. An estimation method executed by an estimation device,
From the concept information that correlates the phenomenon composed of the subject and the event related to the subject, the result that has a causal relationship with the phenomenon, and the polarity that is an index of interest that the result has on the user. A specific step of identifying a set of a phenomenon, a first result having a causal relationship with the predetermined phenomenon, and a polarity of the first result;
Based on the set specified in the specifying step, including an estimation step for estimating a result when a predetermined phenomenon occurs and an interest of the result on the user,
In the specific step, when the first result is also an element indicating an event, from the concept information, the predetermined phenomenon, the first result, the polarity of the first result, and the first result Identifying a set of the second result associated with the result of 1 and the polarity of the second result,
The estimation step estimates an interest on the user based on the polarity of the first result and the polarity of the second result.
An estimation method characterized by that. From the concept information that correlates the phenomenon composed of the subject and the event related to the subject, the result that has a causal relationship with the phenomenon, and the polarity that is an index of interest that the result has on the user. A specific procedure for identifying a set of a phenomenon, a first result having a causal relationship with the predetermined phenomenon, and a polarity of the first result;
Based on the set specified in the specific procedure, let the computer execute a result when a predetermined phenomenon occurs and an estimation procedure for estimating the interest that the result has on the user,
In the specific procedure, when the first result is also an element indicating an event, from the concept information, the predetermined phenomenon, the first result, the polarity of the first result, and the first result Identifying a set of the second result associated with the result of 1 and the polarity of the second result,
The estimation procedure estimates an interest on the user based on the polarity of the first result and the polarity of the second result.
An estimation program characterized by that.

Images