JP6599534B1 - Information processing apparatus, information processing method, and program - Google Patents

Abstract

[Object] To find out a user's true intention (deep needs) while suppressing costs. An information processing apparatus includes: an acquisition unit that acquires opinion information from a user; a determination unit that determines content of the opinion information; and a question or answer to the user based on a determination result by the determination unit. It has a dialogue control unit 30 that makes a response, and a decision unit 40 that decides the user's deep needs candidates. The acquisition unit 10 acquires a plurality of opinion information from the user when the user answers the response from the dialogue control unit 30. The determination unit 40 calculates an evaluation value related to the theme desired by the user and an evaluation value related to the emotion of the user based on the plurality of opinion information from the user, and based on the calculation result, Determine deep needs candidates. [Selection] Figure 1

Description

  The present invention relates to an information processing apparatus, an information processing method, and a program for acquiring opinion information from a user.

  Conventionally, attempts have been made to effectively collect user information and questionnaire responses through natural dialogue. In Patent Document 1, a user terminal is provided that is connected to a content management system via a communication network and outputs conversation content including character display and inputs information for dialogue with the character. Conversation content creation unit for creating conversation content so as to include information for collecting profiles and information for collecting questionnaire responses, and conversation content that is created and transmitted to the user terminal A communication unit that receives input information input from the user terminal in response to the user profile, a user profile DB that stores a user profile included in the received input information for each user, and a questionnaire included in the received input information It has been disclosed that it has a totaling DB that counts the answers.

JP 2003-196462 A

  However, in the conventional mode, the true intention (deep needs) of the user cannot be heard. The present invention provides an information processing apparatus and the like that can hear a user's true intention (deep needs).

An information processing apparatus according to the present invention includes:
An acquisition unit for acquiring opinion information from the user;
A determination unit for determining the content of the opinion information;
Based on a determination result by the determination unit, a dialogue control unit that performs at least a question or a response to the user;
A determination unit for determining the deep needs candidates of the user;
With
When the user answers the question or response from the dialog control unit, the acquisition unit acquires a plurality of opinion information from the user,
The determination unit calculates an evaluation value related to the theme desired by the user and an evaluation value related to the emotion of the user based on the plurality of opinion information from the user, and based on the calculation result, the depth of the user Needs candidates may be determined.

In the information processing apparatus according to the present invention,
The emotion used when the determination unit determines the deep needs candidate may be a concern emotion or an expectation emotion.

In the information processing apparatus according to the present invention,
The determination unit classifies the user's emotion based on the plurality of opinion information from the user, and determines the deep needs candidate when the user's emotion is determined to be emotions other than concern and expectation. Instead, the deep need candidate may be determined when it is determined that the user's emotion is a concern or expectation.

In the information processing apparatus according to the present invention,
The determining unit calculates an evaluation value related to a plurality of themes or an evaluation value related to a plurality of emotions, and determines one or more content categories as deep needs candidates based on a theme and emotions for which the evaluation value is equal to or greater than a threshold value. May be.

In the information processing apparatus according to the present invention,
The determination unit may calculate an evaluation value relating to a plurality of themes or an evaluation value relating to a plurality of emotions based on the accumulated opinion information every time opinion information is input from the user.

In the information processing apparatus according to the present invention,
A limited number of times is provided for the number of pieces of opinion information acquired from the user,
When the limited number of opinion information is acquired from the user, the determination unit may determine the most likely content category as a deep needs candidate.

In the information processing apparatus according to the present invention,
The dialogue control unit makes a response associated with the deep needs candidate to the user,
Information about whether or not the acquisition unit classifies the content category may be acquired from the user.

In the information processing apparatus according to the present invention,
When information indicating that the deep needs candidate is different from the user's needs is input from the user, the acquisition unit may restart acquisition of opinion information from the user.

In the information processing apparatus according to the present invention,
The number of pieces of opinion information acquired again by the acquisition unit may be smaller than the number of pieces of opinion information acquired by the acquisition unit until the deep need candidate is first determined by the determination unit.

In the information processing apparatus according to the present invention,
The opinion information is opinion information about a local public entity,
The acquisition unit acquires opinion information about local public bodies,
The determination unit may determine a deep need candidate of a user who is a user of a local public entity.

An information processing method according to the present invention includes:
A step of acquiring opinion information from the user by the acquisition unit;
A step of determining the content of the opinion information by a determination unit;
Based on the determination result by the determination unit, the dialog control unit makes a question or a response to the user;
A step of determining deep user needs of the user by a determination unit;
With
When the user answers the question or response from the dialog control unit, the acquisition unit acquires a plurality of opinion information from the user,
The determination unit calculates an evaluation value related to the theme desired by the user and an evaluation value related to the emotion of the user based on the plurality of opinion information from the user, and based on the calculation result, the depth of the user Needs candidates may be determined.

The program according to the present invention is:
A program for installing in an information processing apparatus,
The information processing device installed with the program
An acquisition function to acquire opinion information from the user;
A determination function for determining the content of the opinion information;
Based on the determination result by the determination function, a question response function for performing a question or a response to the user;
A decision function for determining candidate deep needs of the user;
With
By the user answering the question or answer from the question answering function, the obtaining function obtains a plurality of opinion information from the user,
The determination function calculates an evaluation value related to the theme desired by the user and an evaluation value related to the emotion of the user based on the plurality of opinion information from the user, and based on the calculation result, the depth of the user Needs candidates may be determined.

  In the present invention, the acquisition unit acquires a plurality of opinion information from the user by the user answering the question or the response from the dialogue control unit, and the determination unit determines the opinion from the user based on the plurality of opinion information. In the case of adopting the aspect classified into the content categories, it is possible to more surely acquire the original needs (deep needs) of the user, not the superficial needs.

FIG. 1 is a schematic diagram showing a configuration of an information processing apparatus according to an embodiment of the present invention. FIG. 2 is a diagram showing the contents of categories that can be used in the embodiment of the present invention. FIG. 3 is a diagram for explaining the flow in the first embodiment of the present invention. FIG. 4 is a diagram for explaining a second embodiment of the present invention. FIG. 5 is a diagram for explaining a third embodiment of the present invention. FIG. 6 is a diagram showing a screen following FIG. FIG. 7 is a diagram showing a screen following FIG. FIG. 8 is a diagram showing a screen following FIG. FIG. 9 is a diagram showing a screen following FIG. FIG. 10 is a diagram showing an example of information flow in the information processing apparatus that can be used in the embodiment of the present invention.

Embodiment << Configuration and Method >>
Embodiments of an information processing apparatus and an information processing method according to the present invention will be described below. In the present embodiment, a program that allows the computer to execute the information processing method of the present embodiment by being installed in a computer such as a personal computer, and a recording medium that records the program are also provided.

  The information processing apparatus according to the present embodiment may be installed in any place, and may be a cloud. The information processing apparatus according to the present embodiment may be composed of a single device or a plurality of devices. Further, when an information processing apparatus is configured from a plurality of devices, each device does not need to be provided in the same space such as the same room, and may be provided in different rooms, different buildings, different areas, and the like.

  The information processing apparatus of this embodiment is generated by installing a program, for example. This program may be distributed by e-mail, may be obtained by logging in after accessing a predetermined URL, or may be recorded on a recording medium. The program according to the present embodiment is used to generate the information processing apparatus described below, and the recording medium according to the present embodiment is used to record the program. Further, the information processing method of the present embodiment is performed by an information processing apparatus in which the program is installed.

  As shown in FIG. 1, the information processing apparatus according to the present embodiment is based on an acquisition unit 10 that acquires opinion information from a user, a determination unit 20 that determines the content of opinion information, and a determination result by the determination unit 20. The dialogue control unit 30 that makes a question or a reply to the user, and the determination unit 40 that determines a user's deep needs candidates may be included. The acquisition unit 10 may acquire a plurality of pieces of opinion information from the user by replying to the response from the dialogue control unit 30. The determination unit 40 may determine a user's deep need candidates based on a plurality of pieces of opinion information. The determination unit 40 stores not only the target user's opinion information but also other information (for example, regional knowledge data as shown in FIG. 10, resident opinion data of residents who are other users, etc.) in the storage unit 60 in advance. The user's deep needs candidates may be determined using external information).

  A communication unit 50 that can communicate with the user terminal 200 may be provided. The acquisition unit 10 may acquire a plurality of opinion information by the communication unit 50 communicating with the user terminal 200 a plurality of times. The information processing apparatus may include a storage unit 60 that stores various information.

  Each time opinion information is input from the user, the determination unit 40 may perform classification related to the user's deep needs candidates based on the accumulated opinion information. In this case, every time opinion information is input from the user, an evaluation value corresponding to a certain theme category or each theme category may be calculated by the determination unit 40. And this evaluation value may be memorize | stored in the memory | storage part 60 with the process until opinion information is input.

  The acquisition unit 10 may acquire opinion information input from a user terminal 200 such as a smartphone or a PC possessed by the user. The determination unit 20 may have an artificial intelligence function and machine learn a determination result for opinion information. Further, the determination unit 40 may also have an artificial intelligence function, and machine learning may be performed to determine deep needs candidates from a plurality of pieces of opinion information. Access from the user terminal 200 to the information processing apparatus may be performed by the user terminal 200 such as a smartphone reading a QR code (registered trademark).

  When the determination unit 20 and the determination unit 40 have an artificial intelligence function, the determination unit 20 and the determination unit 40 use machine learning technology so that the probability that a target event will occur is most likely from past performance data. Employment variables (elements) to be used and their coefficients (weights) may be determined. Various models can be used for machine learning. For example, a tree model may be used, or other models may be used.

  The determination unit 40 classifies the user's desired theme and the user's emotion based on a plurality of opinion information from the user, and determines the user's deep needs candidates based on the theme and emotion. Also good. The determination unit 40 may determine a deep need candidate of the user from the theme category and the emotion category whose evaluation value is equal to or greater than the threshold value based on the opinion information. The evaluation value in the present embodiment is a concept including a probability, and the probability can be used as the evaluation value.

  The determination unit 40 may determine a user's emotion based on opinion information acquired and accumulated from a user who is a target of exchange (referred to as “target user”), and may classify the emotion into a predetermined emotion category. . The determination unit 40 may classify opinions from the user into a predetermined theme category based on the opinion information acquired and accumulated from the target user. The determination unit 40 may determine the content category based on the determined emotion category and theme category. The content category determined in this way may be a deep need candidate.

  The theme category may include, for example, a basic philosophy category related to a basic philosophy, a basic policy category related to basic policy, an administrative management category related to administrative management, a process category related to processes, a countermeasure category related to countermeasures, and the like. Each category may include multiple classifications. The theme category may have a medium classification and / or a small classification (see FIG. 2).

  The sentiment category may include a concern category for concerns, an expectation category for expectations, a confirmation category to be confirmed, a request claim proposal category for requests / claims / suggestions, a question category for questions, etc. Good. As a result of examination by the inventors, it has been confirmed that there is a very high possibility that the user needs deep needs in the case of the concern category or the expectation category. For this reason, the determination unit 40 determines deep needs candidates only when the highest emotion category is a concern category or an expected category and the evaluation value is equal to or greater than a threshold, and the other emotion categories are the highest. In some cases, even if the evaluation value is equal to or higher than the threshold value, the deep needs need not be determined. For example, when there is a request to create a hospital, in the case of “concern”, it is possible that access to medical care is currently bad, while in the case of “expectation”, new employment is It can be thought that it wants. In the present embodiment, “concern” is an emotion including some concern / anxiety, and “expectation” is an emotion including some expectation / request.

  As described above, the determination unit 40 may classify opinion information from the user using both the theme category and the emotion category. As shown in FIG. 2, the subcategory and emotion category of the theme category are a table, and the determination unit 40 determines the content category from the theme category and the emotion category, and the response content set in the content category is determined. Based on this, a reply to the target user may be made by the dialogue control unit 30.

  As shown in FIG. 2, the basic philosophy category may include a category related to a future image and a constraint condition as a middle classification, and each category in the middle classification may include a plurality of minor classification categories. For each category in the small classification, the response content for each of the case where the emotion category is the concern category and the expectation category is stored in the storage unit 60, and the response content is returned by the dialog control unit 30. May be.

  As shown in Fig. 2, the basic policy categories have categories related to human resource development / activity, economy, industry, environment / energy, culture, disaster prevention / safety, and land use / transport as the middle categories. May have a plurality of minor categories. For each category in the small classification, the response content for each of the case where the emotion category is the concern category and the expectation category is stored in the storage unit 60, and the response content is returned by the dialog control unit 30. May be.

  As shown in FIG. 2, the administrative management category may have a category relating to finance, public property, administrative services, and public dependence as a middle classification. Further, each category in the medium classification may have a plurality of small classification categories. For each category in the small classification, the response content for each of the case where the emotion category is the concern category and the expectation category is stored in the storage unit 60, and the response content is returned by the dialog control unit 30. May be.

  As shown in FIG. 2, the process category has categories relating to determination time, determination method, examination method, information provision / explanation, and participation / participation as middle classification, and each category in the middle classification is a category of a plurality of minor classifications. You may have. For each category in the small classification, the response content for each of the case where the emotion category is the concern category and the expectation category is stored in the storage unit 60, and the response content is returned by the dialog control unit 30. May be.

  As shown in FIG. 2, the countermeasure category has a category related to maintenance / renewal costs, facility reduction costs, and facility quantity optimization as a middle category, and each category in the middle category has a plurality of small category categories. Also good. For each category in the small classification, the response content for each of the case where the emotion category is the concern category and the expectation category is stored in the storage unit 60, and the response content is returned by the dialog control unit 30. May be.

  In the above description, the case where the emotion category is a concern category and the case where the response category is an expectation category has been described using a different response content by the dialogue control unit 30, but the present invention is not limited to this, and the emotion category is a concern category. The response contents may be the same in the case of the case and the case of the expected category. For example, in the case of the basic philosophy category, the basic policy category, the administrative management category, and the process category, the response content by the dialogue control unit 30 differs depending on whether the emotion category is a concern category or an expected category. If the emotion category is the concern category, the response content may be the same when the emotion category is the expected category.

  The determination unit 40 calculates evaluation values for a plurality of content categories, and determines one or more content categories as deep needs candidates when an evaluation value for one or more content categories is equal to or greater than a threshold value. Good. In this case, the dialog control unit 30 makes a response (for example, “response content” described above) associated with the deep needs candidate to the user, and the information about whether the acquisition unit 10 classifies the content category is the user. May be obtained from When the evaluation value is equal to or higher than the threshold value in the plurality of content categories, the determination unit 40 may determine the highest evaluation value as the deep needs candidate for the opinion information of the user, or all of the values that are equal to or higher than the threshold value May be determined as candidates for deep needs.

  When a plurality of deep need candidates are determined, the dialogue control unit 30 sends a response (for example, “response content” described above) associated with each deep need candidate to the user, and the acquisition unit 10 receives the content category. You may acquire the information about the pros and cons of classifying from the user. In this case, the dialogue control unit 30 responds to the user in order from the highest evaluation value among the plurality of deep need candidates, and the acquisition unit 10 classifies the content category into the content category. You may acquire information about the right or wrong of the user. Moreover, it is not restricted to such an aspect, The dialogue control part 30 makes the response linked | related with several deep needs candidates with respect to a user at once, and the acquisition part 10 classifies into each content category right or wrong. You may acquire the information about from a user.

  A limited number of times may be provided for the number of pieces of opinion information acquired from the user. Then, when the limited number of opinion information is acquired from the user, the determination unit 40 may determine the most likely content category (for example, the content category with the highest evaluation value) as a deep need candidate. Also in this case, information on whether or not the dialogue control unit 30 responds to the user in order from the highest evaluation value to the user and the acquisition unit 10 classifies the content category. May be obtained from the user.

  The opinion information is opinion information related to the administration of the local public entity and the acquisition unit 10 may acquire opinion information related to the administration of the local public entity. In this case, the determination unit 40 determines candidates for deep needs of users who are administrative users such as local public bodies. Local governments include prefectures and municipalities. It should be noted that the information processing apparatus of the present embodiment can be used not only in such administration but also in companies and the like.

  As shown in FIG. 10, the storage unit 60 may store information on resident opinion data, assumed themes, and regional knowledge data for each region. In the aspect shown in FIG. 10, the intention emotion and the theme are determined based on the opinion information input from the user terminal 200, the deep needs candidates are estimated using these, and as a result, the response selection is performed, and the user is selected. It is shown to be output to the terminal 200. In addition, the storage unit 60 may store an interview scenario, and the interview scenario may be appropriately corrected by input from a predetermined person such as a facilitator.

  As described above, when the information about the right or wrong of being classified into a content category is acquired from the target user, it is determined only when the target user approves it by classifying it into the content category. The classification into the content category by the unit 40 may be confirmed and confirmed as deep needs. By obtaining approval in this way, it is possible to form an agreement with the target user, and it is possible to reliably acquire deep needs. The user's deep needs are stored in the storage unit 60, and are used, for example, to be reflected in future service provision.

  When information indicating that the deep needs candidate is different from the user's original needs is input from the user, the acquisition unit 10 may restart acquisition of opinion information from the user. In this case, the number of pieces of opinion information acquired again by the acquisition unit 10 may be smaller than the number of pieces of opinion information acquired by the acquisition unit 10 until the deep need candidate is determined by the determination unit 40. As an example, the number of times of initial opinion information acquisition may be 3 to 5, and the number of times of opinion information acquired again may be 1 to 4 times.

  Each member such as the acquisition unit 10, the determination unit 20, the dialogue control unit 30, the determination unit 40, and the communication unit 50 according to the present embodiment may be realized by one or a plurality of IC chips or electronic modules, It may be realized by a circuit configuration.

  An example of an information processing method by the information processing apparatus according to the present embodiment will be described.

  First, opinion information is acquired from the target user by the acquisition unit 10. For example, when the target user inputs opinion information from the user terminal 200 such as a tablet or a PC, the opinion information is acquired by the acquisition unit 10 and stored in the storage unit 60.

  When the target user's opinion information is acquired by the acquisition unit 10 in this way, the determination unit 20 determines what the content of the opinion information is. When the determination unit 20 has an artificial intelligence function, the content of the opinion information may be determined by applying the content of the acquired opinion information to a model generated by machine learning. This determination result is also stored in the storage unit 60.

  When the content of the opinion information of the target user is determined by the determination unit 20, a dialogue such as a question or a response to the opinion information is performed by the dialogue control unit 30. Acquisition of the opinion information of the target user by the acquisition unit 10, determination of the content of opinion information by the determination unit 20, and a question or response by the dialogue control unit 30 are performed a plurality of times. When the determination unit 20 determines that the content of the opinion information is a question, the dialog control unit 30 answers the question, and the determination unit 20 determines that the content of the opinion information is determined by the target user. In such a case, the information reflecting the determination is stored in the storage unit 60, and an answer to the effect is given to the user terminal 200.

  In this process, the determination unit 40 repeats the acquisition of the target user's opinion information by the acquisition unit 10, the determination of the content of the opinion information by the determination unit 20, and the question or answer by the dialogue control unit 30. The content category regarding the true intention of the target user) is determined. The determination of the deep needs candidate by the determination unit 40 may be performed each time the target user's opinion information is acquired, or acquisition of the target user's opinion information after the stage where a plurality of opinion information is acquired from the target user. The determination of the deep needs may be performed by the determination unit 40 every time the determination is performed.

  Further, evaluation of the target user's opinion information accumulated by the determination unit 40 is performed, and a classification in which the evaluation value (including the probability) in the emotion category and the theme category is equal to or greater than a predetermined value is determined as a deep needs candidate. Good. In this case, a content category based on a classification in which the evaluation value in the emotion category is equal to or higher than the predetermined value and a classification in which the evaluation value in the theme category is equal to or higher than the predetermined value is determined as a deep need candidate.

  When the deep need candidate is thus determined by the determination unit 40, the response content corresponding to the deep need candidate is read from the storage unit 60 (see FIG. 2), and the dialog control unit 30 determines that the deep need candidate is self- The target user is asked whether or not it agrees with the opinion.

  If it is determined by the target user that the deep need candidate matches the opinion of the user, that fact is input from the target user, and this is acquired by the acquisition unit 20. As a result, the determination unit 40 determines that the deep need candidate is the deep need of the target user, and the storage unit 60 stores this.

  On the other hand, when the target user determines that the deep need candidate does not match his / her opinion, the determination unit 40 determines that the deep need candidate is not the target user's deep need, and the dialog control unit 30 The question or response is started again, and the collection of opinion information from the target user by the acquisition unit 10 is started again.

  Then, re-evaluation is performed on the opinion information of the target user accumulated by the determination unit 40, and a classification in which the evaluation value (including the probability) in the emotion category and the theme category is a predetermined value or more is determined as a deep needs candidate. The dialogue control unit 30 may ask whether the newly determined deep need candidates match the deep needs of the target user (see FIG. 2). Even if the target user determines that the deep needs candidate does not match his / her opinion, the opinion information acquired from the target user so far may be used as accumulated information or reset once. May be.

  The series of steps described above may be repeated until it is determined that the deep needs candidate determined by the determination unit 40 matches the opinion of the user.

"effect"
Next, the effects of the present embodiment having the above-described configuration that have not been described will be mainly described.

  The acquisition unit 10 acquires a plurality of opinion information from the user by the user answering a question or a response from the dialogue control unit 30, and the determination unit 40 determines a predetermined opinion from the user based on the plurality of opinion information. When the mode of classification into the content categories is adopted, not the superficial needs but the user's original needs (deep needs) can be acquired more reliably. That is, it is possible to form an agreement with the user by exchanging a plurality of times with information input from the user terminal 200 such as a smartphone.

  In the past, in order to acquire such a user's original needs (deep needs), it was necessary for an experienced person in charge to ask the user a plurality of questions, which incurred costs including labor costs. In this regard, by adopting the aspect of the present embodiment, the information processing apparatus can acquire the user's original needs (deep needs) by itself, and thus the cost can be greatly reduced. In particular, when a briefing session is held at a local government, only residents of a specific category gather at the briefing session, and the opinions of the residents of a category that does not gather at the briefing session (for example, young people) are not fully heard. There was a thing. As a result, troubles may have occurred with residents at the time when a certain measure was introduced, but according to this embodiment, it is possible to obtain opinion information of residents through the web, etc. Acquire the deep needs of a wide range of residents.

  In general, residents' opinions on local governments are often criticisms and requests for local governments, but by adopting this embodiment, the reasons for superficial criticism and requests will be deepened. Therefore, it is possible to acquire the deep needs of users such as residents.

  When the determination unit 40 classifies the user's desired theme and the user's emotion based on a plurality of opinion information from the user, and determines the deep needs of the user based on the theme and emotion It is possible to determine deep needs using both the theme and emotion perspectives, and to obtain deep needs efficiently and accurately. In addition, according to what the inventors have confirmed, categorizing using emotions in this way is very useful when capturing deep needs.

  When the decision unit 40 performs classification related to the deep needs of the user every time the opinion information is input from the user, it is possible to acquire information on the process leading to the deep needs and to analyze the process of obtaining the deep needs. Is beneficial.

  A mode in which the determination unit 40 calculates evaluation values for a plurality of content categories and determines one or more content categories as deep needs candidates when the evaluation value for one or more content categories is equal to or greater than a threshold value. When adopted, deep needs candidates can be determined based on objective evaluation values and with a certain degree of accuracy.

  When a mode in which a limited number of times is provided for the number of opinion information acquired from the user is adopted, it is possible to efficiently determine deep needs candidates without requiring the user to answer a very large number of times. Thus, the burden on the user can be reduced, and as a result, the questioner can be expected to acquire information from many users.

  When the dialog control unit 30 makes a response associated with the deep needs candidate to the user, and the acquisition unit 10 acquires information about the right or wrong of classification into the content category, the determination unit 40 determines It is possible to confirm whether there is a candidate for deep needs. In particular, when the determination unit 40 has an artificial intelligence function, machine learning can be performed to improve the accuracy of determining the deep needs candidates for the next and subsequent times.

  When information indicating that the deep needs candidate is different from the original needs of the user is input from the user, the acquisition unit 10 adopts a mode in which acquisition of opinion information from the user is restarted, Since the question is started again to acquire the original needs, the original needs can be acquired more reliably.

  In the case of adopting an aspect in which the number of opinion information acquired again by the acquiring unit 10 is less than the number of opinion information acquired by the acquiring unit 10 until the deep need candidate is first determined by the determining unit 40, Information from the user is acquired at a more limited number of times, and the burden on the user can be reduced.

  Questions from users of local governments (typically residents of the local government) are often common throughout the country. Therefore, building a model for one local government is very beneficial in that the model can be used by other local governments, and the introduction cost can be significantly reduced. Such a model may be generated by the determination unit 40.

  In addition, by building a common model based on information from multiple local governments, it is possible to gather a lot of information in a short time, which can significantly shorten the period until the system is introduced. Is very useful.

[Example 1]
Next, an example according to the present embodiment will be described with reference to FIG.

  The content shown on the left side of FIG. 3 (shown as “bot utterance” in FIG. 3) is a question or response issued by the dialogue control unit 30, and the content shown on the right side of FIG. 3 is input by the user. (In FIG. 3, it is indicated as “user's remark”).

  In Fig. 3, the decision was made based on the three opinions given by users: "Tax increase", "I am worried that tax will be increased if there are no financial resources" and "Because the tax increase will make life more difficult" The unit 40 uses a model created by machine learning to calculate a probability (evaluation value) for which theme or intended emotion is most applicable. In the mode shown in FIG. 3, the evaluation value is calculated for each of the theme and emotion, and it is determined that each is equal to or greater than the threshold value. As a result, regarding the theme of “expansion of tax revenue” (increased tax) with a probability of 90%, “concern” with a probability of 80% is applied to determine deep needs candidates. In FIG. 3, the evaluation value is calculated for each of the theme and emotion, but the evaluation value may be calculated only for the most likely theme and emotion and compared with the threshold value.

  In the mode shown in FIG. 3, the probability is equal to or higher than the threshold value, and the intention feeling is “concern” or “expectation”, so that it is determined as a deep need candidate. When the deep need candidate is determined in this way, the dialogue control unit 30 asks the user whether the content category determined from the theme category and the emotion category is the deep need candidate and matches the original opinion of the user. To obtain user approval. If the user's approval is obtained, the deep need candidate is determined as a deep need and stored in the storage unit 60.

[Example 2]
Another example will be described with reference to FIG.

  In the present embodiment, with respect to a plurality of deep needs candidates that are equal to or higher than a certain evaluation value (probability), the user's approval is obtained and the storage unit 60 stores them.

  In the mode shown in FIG. 4, “financial” tax revenue expansion (tax increase) in “administrative management” that is equal to or higher than the threshold (first threshold) in the theme category, and “future image” richness in “basic philosophy” In both cases, the content category determined by applying “concern” that is equal to or higher than the threshold (third threshold) in the emotion category matches the user's original opinion as a deep needs candidate Asking questions and getting user approval.

  In addition, the storage unit 60 may store the result of the theme category that is not approved by the user because the threshold (first threshold) is not reached. When this mode is adopted, the threshold value for obtaining user approval is set as the first threshold value, and the threshold value for the theme category that is not obtained by the user but is stored in the storage unit 60 is set as the second threshold value (first threshold value). > The second threshold), and the theme category less than the second threshold may not be stored in the storage unit 60. As an example, the first threshold is any value from 70% to 80%, the second threshold is any value from 45% to 55%, and the third threshold is any value from 70% to 80%. May be the value.

  The storage of the theme categories that are not approved by the user in this way in the storage unit 60 can be used as a reference at the time of opinion aggregation (for example, there were opinions such as XX and △△ regarding basic policies). Can be tabulated) and is valuable as reference information. When only the information above the second threshold is stored in the storage unit 60, it is advantageous in that information with low relevance can be excluded.

[Example 3]
Still another example will be described with reference to FIGS.

  In the aspect of the third embodiment, a deep need candidate is determined using only a predetermined theme category. In the embodiment shown in FIG. 5, two types of confirmations, ie, administrative administration theme and basic philosophy theme confirmation, are performed, and the classification results are stored in the storage unit 60 for other themes (basic policy theme, process theme and countermeasure theme) It is a mode in which confirmation is not performed. The administrative management theme and the basic philosophy theme are used only as an example. The deep needs candidates may be determined using only other themes, and the number of themes used when the deep needs candidates are determined. Is not limited to two.

  As shown in FIG. 5 and FIG. 6, the determination unit 40 determines the theme category and the emotion category from accumulated opinions such as “I am worried about money”, “I am not good at using money” and “Up on tax burden”. To do. In the mode shown in FIG. 6, the content category determined by applying the concern category in the sub-category “Fiscal Finance” in the medium classification “Finance” in the administrative management theme is set as a deep needs candidate, and the user As to whether it is consistent with the original opinion (shown as “confirmation 1” in FIG. 6).

  When the answer to the confirmation sentence shown in the lower part of FIG. 6 is “No”, as shown in FIG. 7, the dialogue control unit 30 further outputs a question, and the acquisition unit 10 further acquires the opinion information. Start. At this time, the determination unit 40 may perform the evaluation by adding the opinion information acquired in addition to the already accumulated opinion information. In FIG. 7, the content category determined by applying the concern category in the sub-category “constraint condition” economic / financial stability, developability, sustainability in the middle class “constraint condition” in the basic philosophy theme As a deep needs candidate, a question is made as to whether it matches the original opinion of the user (shown as “confirmation sentence 2” in FIG. 7).

  If the answer to the confirmation sentence 2 is also “No”, the dialogue control unit 30 further outputs a question, and the acquisition unit 10 further starts acquisition of opinion information. Then, after a number of exchanges, in FIG. 8, the content determined by applying the concern category in the sub-category “Finance” tax revenue expansion (tax increase) in the medium classification “Finance” in the administrative management theme. The category is a candidate for deeper needs, and a question is asked as to whether it matches the original opinion of the user (shown as “confirmation sentence 3” in FIG. 8).

  Even when the user answers “Yes”, the dialogue control unit 30 may additionally obtain approval from the user for other categories. Even when the user responds “Yes” to the confirmation sentence 3, in FIG. 9, the user is also confirmed for other deep needs candidates. This other deep need candidate may be a deep need candidate having the second highest evaluation value (probability) after the first deep need candidate. In FIG. 9, the content category determined by applying the concern category in the sub-category “future image) richness” in the middle classification “future image” in the basic philosophy theme is used as a deep needs candidate, and the user's original A question is asked as to whether it matches the opinion (shown as “confirmation 4” in FIG. 9). When the user answers “Yes” and there are no other deep need candidates to be confirmed (for example, deep needs candidates having an evaluation value equal to or higher than the first threshold), as shown in FIG. 30 may ask another question, and if there is no further question, the exchange with the user may be terminated.

  In the third embodiment, the emotion category remains the concern category in FIGS. 5 to 9. However, the present invention is not limited to this, and the emotion category is initially a deep needs candidate by repeatedly asking questions. For example, the determination unit 40 may determine that the emotion category is changed from the “concerned category” to the “expected category”.

  The description of the above-described embodiment and the disclosure of the drawings are merely examples for explaining the invention described in the claims, and are described in the claims by the description of the above-described embodiments or the disclosure of the drawings. The invention made is not limited.

DESCRIPTION OF SYMBOLS 10 Acquisition part 20 Determination part 30 Dialog control part 40 Determination part 50 Communication part

Claims (12)

An acquisition unit for acquiring opinion information from the user;
A determination unit for determining the content of the opinion information;
Based on a determination result by the determination unit, a dialogue control unit that performs at least a question or a response to the user;
A determination unit for determining the deep needs candidates of the user;
With
When the user answers the question or response from the dialog control unit, the acquisition unit acquires a plurality of opinion information from the user,
The determination unit calculates an evaluation value related to the theme desired by the user and an evaluation value related to the emotion of the user based on the plurality of opinion information from the user, and based on the calculation result, the depth of the user An information processing apparatus that determines a need candidate.   The information processing apparatus according to claim 1, wherein the emotion used when the determination unit determines the deep needs candidate is an emotion related to concern or an emotion related to expectation.   The determination unit classifies the user's emotion based on the plurality of opinion information from the user, and determines the deep needs candidate when the user's emotion is determined to be emotions other than concern and expectation. The information processing apparatus according to claim 1, wherein the deep need candidate is determined when it is determined that the emotion of the user is a concern or an expectation.   The determining unit calculates an evaluation value related to a plurality of themes or an evaluation value related to a plurality of emotions, and determines one or more content categories as deep needs candidates based on the theme and emotions where the evaluation values are equal to or greater than a threshold value. The information processing apparatus according to any one of claims 1 to 3.   5. The determination unit according to claim 1, wherein each time opinion information is input from the user, the determination unit calculates an evaluation value related to a plurality of themes or an evaluation value related to a plurality of emotions based on the accumulated opinion information. The information processing apparatus according to claim 1. A limited number of times is provided for the number of pieces of opinion information acquired from the user,
The information processing apparatus according to any one of claims 1 to 5, wherein when the opinion information for the limited number of times is acquired from the user, the determination unit determines the most likely content category as a deep need candidate. . The dialogue control unit makes a response associated with the deep needs candidate to the user,
The information processing apparatus according to claim 4 , wherein the acquisition unit acquires information about whether or not to classify the user's opinion into the content category from the user.   The information processing apparatus according to claim 7, wherein when the user inputs information indicating that the deep need candidate is different from the user's needs, the acquisition unit restarts acquisition of opinion information from the user. .   The number of times of opinion information acquired again by the acquiring unit is smaller than the number of times of opinion information acquired by the acquiring unit until the deep need candidate is first determined by the determining unit. Information processing device. The opinion information is opinion information about a local public entity,
The acquisition unit acquires opinion information about local public bodies,
The information processing apparatus according to any one of claims 1 to 9, wherein the determination unit determines a deep need candidate of a user who is a user of a local public entity. A step of acquiring opinion information from the user by the acquisition unit;
A step of determining the content of the opinion information by a determination unit;
Based on the determination result by the determination unit, the dialog control unit makes a question or a response to the user;
A step of determining deep user needs of the user by a determination unit;
With
When the user answers the question or response from the dialog control unit, the acquisition unit acquires a plurality of opinion information from the user,
The determination unit calculates an evaluation value related to a theme desired by the user and an evaluation value related to the emotion of the user based on the plurality of opinion information from the user, and based on the calculation result, An information processing method for determining deep needs. A program for installing in an information processing apparatus,
The information processing device installed with the program
An acquisition function to acquire opinion information from the user;
A determination function for determining the content of the opinion information;
Based on the determination result by the determination function, a question response function for performing a question or a response to the user;
A decision function for determining candidate deep needs of the user;
With
By the user answering the question or answer from the question answering function, the obtaining function obtains a plurality of opinion information from the user,
The determination function calculates an evaluation value related to the theme desired by the user and an evaluation value related to the emotion of the user based on the plurality of opinion information from the user, and based on the calculation result, A program that determines deep needs.