Information processing device, information presenting method, and information presenting program

Abstract

An information processing device (1), in order to enable the presentation of useful information when recommending a recommendation, comprises: a generation unit (11) that generates questions or hypotheses according to the recommendation; and a presentation unit (12) that presents answers to the questions generated by the generation unit (11) or the results of verifying the hypotheses generated by the generation unit (11).

Description

Information processing device, information presentation method, and information presentation program

[0001] Related to technology for recommending products and services.

[0002] Technologies for recommending products and services have been known for some time. For example, Patent Document 1 listed below discloses a product recommendation device that generates customer groups by purchase reason tendency based on product purchase history information and product factors that indicate the reasons for purchasing the products, and determines recommended products to be recommended to each customer depending on which customer group by purchase reason tendency the customer belongs to.

[0003] Japanese Patent Application Laid-Open No. 2015-201090

[0004] The product recommendation device described in Patent Document 1 can only determine recommended products, but cannot present information useful when recommending the determined recommended products, and in this respect there is room for improvement. One aspect of the present invention has been made in consideration of the above-mentioned viewpoint, and one example of its purpose is to provide an information processing device or the like that can present information useful when recommending a recommendation target.

[0005] An information processing device according to one aspect of the present invention includes a generation means for generating a question or hypothesis according to a recommendation target determined to be recommended to a subject, and a presentation means for presenting an answer to the question generated by the generation means or a verification result of the hypothesis generated by the generation means.

[0006] An information presentation method according to one aspect of the present invention includes at least one processor generating a question or hypothesis corresponding to a recommendation target determined to be recommended to a subject, and presenting an answer to the generated question or a verification result of the generated hypothesis.

[0007] An information presentation program according to one aspect of the present invention causes a computer to function as a generation means for generating questions or hypotheses corresponding to recommendation targets determined to be recommended to a subject, and a presentation means for presenting answers to the questions generated by the generation means or verification results of the hypotheses generated by the generation means.

[0008] According to one aspect of the present invention, it is possible to present information that is useful when recommending a recommendation target.

[0009] FIG. 1 is a block diagram showing the configuration of an information processing device according to exemplary embodiment 1 of the present invention. FIG. 2 is a flow diagram showing the flow of an information presentation method according to exemplary embodiment 1 of the present invention. FIG. 3 is a diagram showing an overview of an information presentation method according to exemplary embodiment 2 of the present invention. FIG. 4 is a block diagram showing the configuration of an information processing device according to exemplary embodiment 2 of the present invention. FIG. 5 is a diagram showing an example of generation of a prediction model and an example of generation of a recommendation reason. FIG. 6 is a flow diagram showing the flow of an information presentation method according to exemplary embodiment 2 of the present invention. FIG. 7 is a diagram showing an example of a computer that executes instructions of a program that is software that realizes each function of each device according to each exemplary embodiment of the present invention.

[0010] [First Exemplary Embodiment] A first exemplary embodiment of the present invention will be described in detail with reference to the drawings. This exemplary embodiment is a basic form of the exemplary embodiments described below.

[0011] (Configuration of Information Processing Device 1) The configuration of the information processing device 1 according to this exemplary embodiment will be described with reference to Fig. 1. Fig. 1 is a block diagram showing the configuration of the information processing device 1. As shown in Fig. 1, the information processing device 1 includes a generation unit (generation means) 11 and a presentation unit (presentation means) 12.

[0012] The generation unit 11 generates a question or a hypothesis according to the recommendation target determined to be recommended to the subject. Note that the generation unit 11 may generate both a question and a hypothesis.

[0013] The presentation unit 12 presents an answer to a question generated by the generation unit 11 or a verification result of a hypothesis generated by the generation unit 11. Note that if the generation unit 11 generates both a question and a hypothesis, the presentation unit 12 may present both the answer to the question and the verification result of the hypothesis. Furthermore, the generation of the answer to the question and the verification of the hypothesis may be performed by the information processing device 1 or may be performed by another information processing device.

[0014] As described above, the information processing device 1 according to this exemplary embodiment includes the generation unit 11 that generates questions or hypotheses according to a recommendation target determined to be recommended to a subject, and the presentation unit 12 that presents answers to the questions generated by the generation unit 11 or verification results of the hypotheses generated by the generation unit 11. Therefore, the information processing device 1 according to this exemplary embodiment has the effect of being able to present information that is useful when recommending a recommendation target.

[0015] (Information Presentation Program) The functions of the above-described information processing device 1 can also be realized by a program. The information presentation program according to this exemplary embodiment causes a computer to function as the generation unit 11 and the presentation unit 12. This information presentation program has the effect of making it possible to present information that is useful when recommending a recommendation target.

[0016] (Flow of Information Presentation Method) The flow of the information presentation method according to this exemplary embodiment will be described with reference to Fig. 2. Fig. 2 is a flow diagram showing the flow of the information presentation method. Note that the execution entity of each step in this information presentation method may be a processor provided in the information processing device 1, or a processor provided in another device, or each step may be executed by a processor provided in a different device.

[0017] In S11, at least one processor generates a question or a hypothesis according to the recommendation target determined to be recommended to the subject. Note that in S11, both a question and a hypothesis may be generated.

[0018] In S12, at least one processor presents an answer to the question generated in S11 or a verification result of the hypothesis generated in S11. Note that if both a question and a hypothesis are generated in S11, both the answer to the question and the verification result of the hypothesis may be presented in S12.

[0019] As described above, the information presentation method according to this exemplary embodiment includes at least one processor generating a question or a hypothesis according to a recommendation target determined to be recommended to a subject, and presenting an answer to the generated question or a verification result of the generated hypothesis. This information presentation method has the effect of making it possible to present information useful when recommending a recommendation target.

[0020] [Exemplary Embodiment 2] (Outline of Information Presentation Method) Fig. 3 is a diagram showing an outline of an information presentation method (hereinafter referred to as the present method) according to this exemplary embodiment. Fig. 3 shows two people, Person A and Person B, of whom Person B is a recommender who recommends a product, service, or other recommended item, and Person A is the target of the recommendation.

[0021] Person B may be, for example, a salesperson, and in this case, the success or failure of the sales, i.e., whether or not a contract for the recommended product is reached, will depend on what kind of product is recommended to Person A and how. According to this method, useful information can be presented when recommending a product to Person B, making Person B's sales activities more effective.

[0022] In this method, attribute data D1 indicating the type of person B, the target of recommendation, is first input into the prediction model 211, and the probability of a sale being reached when each candidate product is recommended to person A is predicted. In the example of Figure 3, of the candidate products A and B, the sale probability of product A is predicted to be 0.82, and the sale probability of product B is predicted to be 0.1. Note that the recommended products A and B may be goods or services, or a combination thereof. Details of the prediction model 211 will be described later.

[0023] In this example, the numerical range of the accuracy is set to 0 to 1, so it can be said that there is a high probability of a contract being concluded when product A is recommended, and a low probability of a contract being concluded when product B is recommended. Therefore, in this example, product A is determined as the product to be recommended. Note that whether or not the probability of a contract being concluded is high may be determined based on a predetermined threshold. In other words, in this method, a product whose accuracy calculated using the prediction model 211 is equal to or greater than a threshold may be determined as the product to be recommended.

[0024] Next, this method generates a recommendation reason for the determined recommended target. In the example of Figure 3, the recommendation reasons are that person A's hobby is golf and that his annual income is 7 million yen or more. The recommendation reasons are information that is useful in recommending product A. However, based on the recommendation reasons shown in Figure 3 alone, person B cannot understand the relationship between person A's characteristics, such as his hobby being golf and his annual income being 7 million yen or more, and product A. For this reason, person B may not be able to effectively promote product A to person A.

[0025] Therefore, in this method, questions and hypotheses are generated according to the recommendation target determined as described above. In the example of FIG. 3, the questions "What is golf useful for?" and "What kind of person has an annual income of 7 million yen?" and the hypothesis "Product A is related to golf" are generated. Note that this method may generate only either a question or a hypothesis. Also, only one question may be generated, or multiple questions may be generated. The same applies to hypotheses. Note that the method for generating questions and hypotheses will be described later.

[0026] This method then generates responses to the questions and hypotheses generated as described above. In the example of Figure 3, the answer "Building connections and improving health" is generated in response to the question "What is golf good for?", and the answer "He's a successful businessman" is generated in response to the question "What kind of person earns 7 million yen a year?". Furthermore, the verification result "Not directly related" is generated in response to the hypothesis "Product A is related to golf."

[0027] In this method, the answers and verification results generated as described above are presented to Person B. This information is useful when Person B recommends Product A to Person A. For example, in the example of Figure 3, Person B, upon receiving the answers and verification results, realizes that many customers with an annual income of 7 million yen or more play golf as part of their career development. Person B then comes up with the idea of ​​proposing a plan that allows for network formation in combination with Product A. In this way, the answers and verification results presented by this method are useful information that contribute to making Person B's sales activities more effective.

[0028] As described above, this method includes generating a question or hypothesis according to a recommendation target (product A in the example of FIG. 3) to be recommended to a target person (person A in the example of FIG. 3), and presenting an answer to the generated question or a verification result of the generated hypothesis. Therefore, this method has the effect of making it possible to present information that is useful when recommending a recommendation target.

[0029] In this method, the information presented to the user may also be a recommendation target. For example, some online shopping sites automatically determine and present recommended products to users of the site. On such shopping sites, the method may be used to generate questions or hypotheses corresponding to the recommended products, and present the answers or verification results along with the recommended products. This allows users to understand the reasons and background for the product recommendation, as well as various information about the product itself, and motivates them to purchase.

[0030] (Configuration of Information Processing Device 2) The configuration of the information processing device 2 according to this exemplary embodiment will be described with reference to Fig. 4. Fig. 4 is a block diagram showing the configuration of the information processing device 2. As shown in the figure, the information processing device 2 includes a control unit 20 that controls each unit of the information processing device 2, and a storage unit 21 that stores various data used by the information processing device 2. The information processing device 2 also includes an input unit 22 that accepts user input operations to the information processing device 2, an output unit 23 that allows the information processing device 2 to output data, and a communication unit 24 that allows the information processing device 2 to communicate with other devices.

[0031] The control unit 20 also includes a recommendation unit (recommendation means) 201, a recommendation reason generation unit (recommendation reason generation means) 202, a generation unit (generation means) 203, a response unit (response means) 204, and a presentation unit (presentation means) 205. The storage unit 21 stores a prediction model 211 and a generation model 212.

[0032] The recommendation unit 201 determines a recommendation target to be recommended to the subject. The recommendation target may be an object, a service, or a combination thereof, as described above. It is also desirable that the recommendation target be determined according to the attributes of the subject. For example, the recommendation unit 201 may determine a recommendation target according to the attributes of the subject using a prediction model 211. The prediction model 211 will be described later in the section "Method for determining a recommendation target and method for generating a recommendation reason."

[0033] The recommendation reason generation unit 202 generates a recommendation reason for the recommended object determined by the recommendation unit 201. More precisely, the recommendation reason generation unit 202 generates information indicating the recommendation reason for the recommended object determined by the recommendation unit 201, but here the information indicating the recommendation reason will be simply referred to as the recommendation reason. A method for generating a recommendation reason will be described later in the section "Method for determining a recommended object and method for generating a recommendation reason."

[0034] The generation unit 203 generates a question or a hypothesis according to the recommendation target determined by the recommendation unit 201. More precisely, the generation unit 203 generates a question sentence, which is a sentence expressing a question in natural language, or a hypothesis sentence, which is a sentence expressing a hypothesis in natural language, but here, a hypothesis sentence will be simply referred to as a hypothesis, and a question sentence will simply be referred to as a question. A generation model 212 can also be used to generate a question or a hypothesis. Note that the question generation unit that generates a question and the hypothesis generation unit that generates a hypothesis may each be provided as separate blocks. Details of the method for generating questions and hypotheses and the generation model 212 will be described later in the section "Method for generating questions and hypotheses."

[0035] The response unit 204 generates a response to the question or hypothesis generated by the generation unit 203. Note that the answer generation unit that generates an answer to a question and the hypothesis verification unit that generates a verification result of the hypothesis may each be provided as separate blocks. As described above, the questions and hypotheses generated by the generation unit 203 are both sentences, so it is possible to generate a response to them using natural language processing techniques. Details of the method for generating an answer to a question and the method for verifying a hypothesis will be explained later in the section "Answer Generation / Hypothesis Verification Method."

[0036] The presentation unit 205 presents an answer to a question generated by the generation unit 203 or a verification result of a hypothesis generated by the generation unit 203. The presentation method may be any method that allows the subject to recognize the content to be presented. For example, if the output unit 23 is a display device, the presentation unit 205 may present the answer or the verification result by displaying and outputting it on the output unit 23. Furthermore, for example, if the output unit 23 is an audio output device, the presentation unit 205 may present the answer or the verification result by outputting it as audio on the output unit 23. Furthermore, the presentation unit 205 may output the answer or the verification result to a device external to the information processing device 2.

[0037] As described above, the information processing device 2 according to this exemplary embodiment includes the generation unit 203 that generates questions or hypotheses according to a recommendation target determined to be recommended to a subject, and the presentation unit 205 that presents answers to the questions generated by the generation unit 203 or verification results of the hypotheses generated by the generation unit 203. Therefore, the information processing device 2 according to this exemplary embodiment has the effect of being able to present information that is useful when recommending a recommendation target.

[0038] (Method of Determining a Recommendation Target and Method of Generating a Recommendation Reason) A method of determining a recommendation target by the recommendation unit 201 and a method of generating a recommendation reason by the recommendation reason generation unit 202 will be described with reference to Fig. 5. Fig. 5 is a diagram showing an example of generating a prediction model 211 and an example of generating a recommendation reason.

[0039] 5, a prediction model 211 is generated by learning using training data D2. The training data D2 is data indicating, for each of a plurality of customers, the customer's annual income and hobbies, the products that the customer was approached to purchase, and whether or not a contract for the products was concluded, in response to the customer's ID (identification information).

[0040] The training data D2 indicates the relationship between a person's attributes and the results of recommending a recommended product to a person with the attributes. Therefore, by learning using the training data D2, it is possible to generate a prediction model 211 that predicts the likelihood of a contract being reached when a specific product is recommended to a target person, based on the target person's attributes.

[0041] The prediction model 211 is information representing the relationship between explanatory variables and dependent variables. The prediction model 211 is, for example, a component for estimating the results of an estimation target by calculating a dependent variable based on the explanatory variables. The prediction model 211 is generated by executing a learning algorithm using training data, for which the dependent variable value has already been obtained, and arbitrary parameters as input. The prediction model 211 may be represented, for example, by a function c that maps an input x to a correct answer y. The prediction model 211 may also estimate the numerical value of the estimation target or the label of the estimation target. Furthermore, the prediction model 211 may output a variable describing the probability distribution of the dependent variable. The prediction model 211 may also be referred to as a "learning model," an "analysis model," an "AI (artificial intelligence) model," a "trained model," an "inference model," or a "prediction formula." The explanatory variables are variables used as inputs in the prediction model. The explanatory variables may also be referred to as "features" or "characteristics."

[0042] The prediction model 211 may be any model capable of predicting the likelihood of a deal being concluded, and the learning algorithm for generating the prediction model is not particularly limited. For example, the learning algorithm for generating the prediction model 211 may be a random forest, a support vector machine, a naive Bayes, or a neural network.

[0043] Alternatively, the prediction model 211 may be a piecewise linear model. The piecewise linear model is constructed by setting categories so that predictions can be made using a linear model and generating a linear model for each category. For example, assume that three categories are set: Category 1, in which the attribute value of "hobby" is "golf" and the attribute value of "annual income" is equal to or greater than "700," Category 2, in which the attribute value of "annual income" is equal to or greater than "1200," and Category 3, in which the attribute value of "annual income" is equal to or less than "400." In this case, for each of these categories, a linear model is generated that predicts the likelihood of a sale being reached when product A is recommended to a target person based on the target person's attribute values.

[0044] One method for generating a piecewise linear model uses FAB inference (Factorized Asymptotic Bayesian Inference). A method for generating a piecewise linear model using FAB inference is disclosed in, for example, U.S. Patent Publication US2014 / 0222741A1.

[0045] When determining a recommendation target using a piecewise linear model, the recommendation unit 201 identifies a category of the target person from the target person's attributes and predicts the likelihood of a sale being concluded using the linear model for that category. For example, the person in attribute data D1 shown in FIG. 5 (the person whose customer ID is 2011) has an "annual income" of "720" and an "interest" of "golf." Therefore, the recommendation unit 201 identifies that the target person falls into the above-mentioned category 1, and predicts the likelihood of a sale being concluded when product A is recommended to the target person using the linear model corresponding to category 1, based on the attribute value (e.g., the value of "annual income") indicated in the attribute data D1.

[0046] As described above, the recommendation unit 201 can predict the likelihood that each of the recommendation candidates will result in a contract using the prediction model 211. Then, the recommendation unit 201 can determine the recommendation target based on the prediction results. For example, the recommendation unit 201 may determine the candidate with the highest likelihood of resulting in a contract as the recommendation target, or may determine the candidate with a likelihood of resulting in a contract that is equal to or greater than a predetermined threshold as the recommendation target.

[0047] Next, a method for generating a recommendation reason will be described. When the prediction model 211 is a piecewise linear model, the recommendation reason generation unit 202 may use the conditions defined for the category to which the target person belongs as the recommendation reason. For example, a person with a customer ID of 2011 indicated in the attribute data D1 belongs to category 1 (the category for hobby = golf and annual income > 700) among categories 1 to 3 of the above-mentioned piecewise linear model. Therefore, the recommendation reason generation unit 202 may use the fact that the person's hobby is golf and that their annual income is 7 million yen or more as the recommendation reason for recommending product A to the person.

[0048] Graph G1 shown in FIG. 5 shows the number of successful transactions for individuals belonging to each of categories 1 to 3 of the piecewise linear model described above. In graph G1, categories with a high number of successful transactions are placed at the top, and categories with a high number of unsuccessful transactions are placed at the bottom. The individual with customer ID 2011 shown in attribute data D1 belongs to category 1 (hobby = golf and annual income > 700), which has the highest number of successful transactions in graph G1. Therefore, the fact that this individual's hobby is golf and that their annual income is 7 million yen or more can be considered a valid reason (or basis) for recommending product A to this individual.

[0049] The above-described methods for determining recommended items and generating recommendation reasons are merely examples. For example, a prediction model generated by learning the relationship between a person's attributes and recommended items to be recommended to a person with those attributes may be used to determine recommended items. In this case, the output of the prediction model becomes the recommended items.

[0050] Furthermore, the format of the input data input to the information processing device 2 to determine the recommendation target is not particularly limited. For example, the input data can be attribute data D1 in a table format as shown in Fig. 5, or data in other data formats such as images and audio can also be used as the input data. The input data can be used to determine the recommendation target after converting the format as necessary.

[0051] The prediction algorithm is not particularly limited, and the probability of a deal being concluded may be predicted using predetermined rules, such as an attribution model. It is also possible to determine a recommendation target without using a prediction model generated by machine learning. For example, if product or service purchase history information is available, the purchase history information may be used to classify customers into multiple groups in advance according to their purchasing tendencies and customer attributes. In this case, the recommendation unit 201 may classify target individuals into one of the groups and determine a recommendation target according to the purchasing tendencies of the group (for example, a high purchase frequency or a high total purchase amount within the group).

[0052] Furthermore, any method may be appropriately adopted as a method for generating a recommendation reason according to a method for determining a recommended target. For example, the recommendation reason generation unit 202 may use, as a recommendation reason, an attribute that has a relatively strong correlation with the successful purchase of a product among the attributes of the target person used to determine the recommended target. For example, assume that a common attribute among many people who have successfully purchased a specific product in past cases is that their hobby is watching videos. In this case, if the target person's attributes include watching videos as a hobby, the recommendation reason generation unit 202 may use, as a reason for recommending the product, the hobby being watching videos, or the high success rate of purchases of the product among people whose hobby is watching videos.

[0053] (Method for generating questions and hypotheses) A ​​method for generating questions and hypotheses by the generation unit 203 will be described. Various methods can be applied as a method for generating questions and hypotheses, but for example, the generation unit 203 may generate questions and hypotheses using a generation model 212 that is generated by learning the relationship between a recommendation target and a question or hypothesis corresponding to the recommendation target. This provides the effect of being able to generate valid questions or hypotheses based on the learning results, in addition to the effect provided by the information processing device 1 according to the first exemplary embodiment.

[0054] For example, by learning using training data that associates various pieces of information about the recommended target with questions that a sales representative has from the information and hypotheses that the sales representative has come up with from the information, it is possible to generate a generative model 212 that generates questions and hypotheses from various pieces of information about the recommended target. Note that the generative model is not limited to one that has been trained using training data. The generative model may also be a model based on unsupervised learning, such as a GAN (Generative Adversarial Network).

[0055] Examples of various information related to the recommended target include attributes of the recommended target (e.g., product name, product genre, price or price range, target age, etc.), reasons for recommendation, and attributes of the target or recommender (e.g., age, gender, occupation, income, career, affiliation), etc. In addition to these, for example, the probability of a sale being concluded for the recommended target (predicted by the prediction model 211) may also be used as information related to the recommended target.

[0056] By using the subject's attributes as information about the recommended target, questions or hypotheses tailored to the subject can be generated. For example, it is possible to generate questions or hypotheses tailored to the subject's gender or age group, thereby presenting answers or hypothesis verification results tailored to the subject's gender or age group.

[0057] Furthermore, by using the attributes of the recommender as information about the recommended target, questions or hypotheses tailored to the recommender can be generated. For example, questions or hypotheses tailored to the number of years the recommender has worked as a salesperson can be generated, and the answers and hypothesis verification results can be presented. This allows the recommender to provide the target with an explanation appropriate to their number of years of service.

[0058] Furthermore, by using the probability of a sale of the recommended target or the probability that the recommended target is suitable for the target person as information about the recommended target, a question or hypothesis corresponding to the probability can be generated and an answer to the question or a verification result of the hypothesis can be presented. For example, when the probability is equal to or greater than a predetermined threshold, the generation unit 203 may generate a question or hypothesis including predetermined wording reflecting the high probability (e.g., "confidently recommended," "highly recommended," "ideal for the target person," etc.). Furthermore, for example, the generation unit 203 may generate a question or hypothesis using different generation models or different templates when the probability is equal to or greater than a predetermined threshold and when it is less than the threshold. Note that the generation of questions or hypotheses using templates will be described later.

[0059] The generation unit 203 may also generate questions and hypotheses without using the generation model 212. For example, the generation unit 203 may generate questions and hypotheses using either or both of pre-created rules and templates. For example, the generation unit 203 may use a template "What is the value of (the predetermined attribute extracted from the reason for recommendation) useful for?" to extract the word "golf," which is the value of the attribute "hobby," from the reason for recommendation in the example of FIG. 3, and generate a question such as "What is golf useful for?". Similarly, the generation unit 203 may use a template "What kind of person is someone whose (the predetermined attribute extracted from the reason for recommendation) is (the value of the attribute)?" to extract the word "7 million," which is the value of the attribute "annual income," from the reason for recommendation in the example of FIG. 3, and generate a question such as "What kind of person is someone who earns 7 million yen a year?" The same applies to hypotheses. For example, by using a template such as "(recommended item) is related to (value of a specified attribute extracted from the reason for recommendation)," the generation unit 203 can generate a hypothesis such as "product A is related to golf" from the recommended item and reason for recommendation in the example of Figure 3.

[0060] An example of a rule for generating questions and hypotheses is word substitution. For example, a rule may be defined that replaces the word "golf" with the broader concept of "sports," or that the attribute value of "annual income" is replaced with "high income" or "middle income" depending on the range of its value. By applying such rules, it becomes possible to generate more general questions and hypotheses. For example, instead of or in addition to the question "What is golf good for?", it becomes possible to generate a more general question such as "What is sports good for?"

[0061] Another example of a rule for generating questions and hypotheses is a rule for selecting a template to be used depending on the attribute used for generation, etc. For example, a rule may be established that a template "What is the use of (hobbies)?" is used to generate a question about the attribute "hobbies," and a template "What kind of person has an annual income of (annual income attribute value)?" is used to generate a question about the attribute "annual income."

[0062] When generating questions and hypotheses using rules and templates, information related to the recommended item, such as attributes of the recommended item (e.g., product name, product genre, price or price range, target age, etc.), reasons for recommendation, and attributes of the target or recommender (e.g., age, gender, occupation, income, etc.), can be used as material for the questions and hypotheses.

[0063] As described above, the generation unit 203 may generate a question or a hypothesis based on the recommendation reason generated by the recommendation reason generation unit 202. This provides the effect of being able to present information that delves deeper into the recommendation reason in addition to the effect provided by the information processing device 1 according to the first exemplary embodiment.

[0064] Furthermore, as described above, the recommendation unit 201 may determine a recommendation target according to the attributes of a target person using the prediction model 211 generated by learning the relationship between a person's attributes and the results of recommending a recommendation target to a person with the attributes, or the relationship between a person's attributes and a recommendation target to be recommended to a person with the attributes. In this case, the generation unit 203 may generate a question or a hypothesis according to the accuracy of the prediction result of the prediction model 211. This provides an effect of being able to present information that takes into account the accuracy of the prediction result, in addition to the effect achieved by the information processing device 1 according to the first exemplary embodiment.

[0065] As described above, the generation unit 203 may generate a question or a hypothesis based on the attributes of the subject, thereby achieving the effect of being able to present information suited to the subject in addition to the effect achieved by the information processing device 1 according to the exemplary embodiment 1.

[0066] As described above, the generation unit 203 may generate a question or a hypothesis based on the attributes of a recommender who recommends a recommendation target to a subject. This provides the effect of being able to present information suited to the recommender in addition to the effect provided by the information processing device 1 according to the first exemplary embodiment.

[0067] (Answer Generation / Hypothesis Verification Method) The method of generating an answer to a question and a method of verifying a hypothesis by the response unit 204 will now be described in detail. An answer to a question can be generated, for example, by using a corpus external to the information processing device 2. A corpus is a large-scale collection of structured natural language sentences. In this case, the response unit 204 detects, from among the many question sentences included in the corpus, one that is the same as the question sentence generated by the generation unit 203 or one that is similar in content, and generates an answer sentence associated with that question sentence as an answer to the question generated by the generation unit 203. Answers can also be generated in a similar manner by using a knowledge graph instead of a corpus. A knowledge graph is a graph structure in which various pieces of knowledge are systematically linked together.

[0068] On the other hand, hypotheses can be verified by using premise sentences whose content is known to be correct and a language understanding model. A language understanding model is constructed so that, when a pair of a hypothesis sentence and a premise sentence is input, it outputs an entailment score, which is an index value indicating the degree to which the premise sentence entails the hypothesis sentence. Such a language understanding model can be constructed by learning whether or not the premise sentence entails the hypothesis sentence using pairs of premise sentences and hypothesis sentences whose entailment relationship is known as training data.

[0069] For example, the response unit 204 may perform a process of inputting the hypothesis sentence and premise sentence generated by the generation unit 203 into a language understanding model for various premise sentences, and if any of the premise sentences has an implication score greater than or equal to a threshold, determine that the hypothesis of that premise sentence is correct.

[0070] The method for determining the degree of implication is not limited to the above-described method that uses a language understanding model constructed using training data. For example, the response unit 204 may calculate the similarity between the premise sentence and the hypothesis sentence that have been vectorized by the pre-trained language model, and use the calculated similarity as an index value indicating the degree of implication.

[0071] Furthermore, any determination method can be applied to determine the degree of implication as long as it can define the relationship between the hypothesis sentence and the premise sentence. For example, existing methods such as keyword matching and TF-IDF (Inverse Document Frequency) may be used to determine the degree of implication.

[0072] (Presenting Answers and Verification Results) The presentation unit 205 may present the answers to the questions generated by the generation unit 203 or the verification results of the hypotheses generated by the generation unit 203 as they are, or may generate presentation data using the answers or the verification results and present the generated presentation data. The method for generating the presentation data is not particularly limited, and the presentation data may be generated using, for example, predetermined rules or templates. For example, if the answers or verification results are words, the presentation unit 205 may embed the words in a template to create a sentence, which is used as presentation data. For example, by using a template such as "(the value of the attribute "hobbies") is useful for (words included in the question answer)," the presentation unit 205 can present the answer "Building connections and improving health" in the example of FIG. 3 as the sentence "Golf is useful for building connections and improving health."

[0073] Furthermore, the presenting unit 205 may generate sentences according to the answers or verification results by using a sentence generation model generated by learning effective explanations and phrases used by salespeople as training data, for example. This enables even inexperienced salespeople to carry out effective sales activities.

[0074] It is desirable that the presentation unit 205 presents not only the answer or verification result but also the corresponding question and hypothesis, the recommended target, and the reason for recommendation. The presentation unit 205 may also present various information regarding the recommended target (for example, specifications of the recommended target, an image showing the appearance of the recommended target, word-of-mouth reviews about the recommended target, etc.), attributes of the target person, etc.

[0075] (Accepting Questions Regarding Presented Answers and Verification Results) The response unit 204 may accept input of questions regarding the answers or verification results presented by the presentation unit 205. The response unit 204 may then generate answers to the input questions, and the presentation unit 205 may also present the answers. This provides, in addition to the effects of the information processing device 1 according to the first exemplary embodiment, an effect of being able to interactively resolve the doubts of the person who inputs the questions (for example, the recommender or the subject) by utilizing the response unit 204, which is originally intended to generate answers to questions generated by the generation unit 203.

[0076] The question may be input via the input unit 22. For example, if the input unit 22 is a device that accepts character input, such as a keyboard, the question may be input as text. The input unit 22 may also be a device that accepts voice input, such as a speaker. In this case, the question may be input as voice, and the voice input by the information processing device 2 or an external device may be converted into character data. Of course, the question may be input by a device external to the information processing device 2, and the response unit 204 may acquire the input accepted by that device via the communication unit 24.

[0077] (Flow of Information Presentation Method) The flow of the information presentation method executed by the information processing device 2 will be described with reference to Fig. 6. Fig. 6 is a flow diagram showing the flow of the information presentation method according to this exemplary embodiment.

[0078] In S21, the recommendation unit 201 acquires attribute data of the target person. The attribute data is used to determine a recommendation target and indicates the attributes of the target person. Subsequently, in S22, the recommendation unit 201 determines a recommendation target based on the attribute data acquired in S21. As described in the section "Method for determining a recommendation target / method for generating a recommendation reason," the recommendation unit 201 may determine a recommendation target based on the output value of the prediction model 211. Note that the presentation unit 205 may present a recommendation target when the recommendation target is determined in S22.

[0079] In S23, the recommendation reason generation unit 202 generates a recommendation reason for the recommendation target determined in S22. The method for generating the recommendation reason is as described in the section "Method for determining a recommendation target / Method for generating a recommendation reason." Note that when the recommendation reason is generated in S23, the presentation unit 205 may present the recommendation reason.

[0080] In S24, the generation unit 203 generates a question or a hypothesis according to the recommendation target determined in S22. As described in the section "Method for generating a question or a hypothesis," the generation unit 203 may generate a question or a hypothesis using the generation model 212.

[0081] In S25, the response unit 204 generates an answer to the question generated in S24 or a verification result for the hypothesis generated in S24. The method for generating the answer and the verification result is as described in the section "Answer Generation / Hypothesis Verification Method."

[0082] In S26, the presentation unit 205 presents the answer or verification result generated in S25. At this time, it is preferable that the presentation unit 205 also presents a question or hypothesis corresponding to the answer or verification result, in addition to the answer or verification result generated in S25. Furthermore, if the recommendation target and the recommendation reason have not been presented in S22 and S23, the presentation unit 205 may also present the recommendation target and the recommendation reason in S26.

[0083] In S27, the response unit 204 determines whether a question regarding the question presented in S26 or the verification result has been input. If the determination in S27 is YES, the process returns to S25, where the response unit 204 generates an answer to the input question, and in the subsequent S26, the presentation unit 205 presents the answer. On the other hand, if the determination in S27 is NO, the process in FIG. 6 ends.

[0084] As described above, the information presentation method according to this exemplary embodiment includes generating a question or a hypothesis according to a recommendation target determined to be recommended to a subject (S24), and presenting an answer to the generated question or a verification result of the generated hypothesis (S26). Thus, this information presentation method has the effect of making it possible to present information that is useful when recommending a recommendation target.

[0085] [Modification] The entity that executes each process described in the exemplary embodiment above is arbitrary and is not limited to the above example. In other words, an information presentation system having the same functions as the information processing device 2 can be constructed using multiple devices that can communicate with each other. For example, by distributing each block shown in FIG. 4 among multiple devices, an information presentation system having the same functions as the information processing device 2 can be constructed. Furthermore, each process in the flowchart of FIG. 6 can be shared and executed by multiple processors.

[0086] [Example of Software Implementation] Some or all of the functions of the information processing devices 1 and 2 may be implemented by hardware such as an integrated circuit (IC chip), or by software.

[0087] In the latter case, the information processing devices 1 and 2 are realized, for example, by a computer that executes instructions of a program, which is software that realizes each function. An example of such a computer (hereinafter referred to as computer C) is shown in FIG. 7. The computer C includes at least one processor C1 and at least one memory C2. The memory C2 stores a program (information presentation program) P for operating the computer C as the information processing device 1 or 2. In the computer C, the processor C1 reads and executes the program P from the memory C2, thereby realizing each function of the information processing device 1 or 2.

[0088] The processor C1 may be, for example, a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), a micro processing unit (MPU), a floating point number processing unit (FPU), a physics processing unit (PPU), a tensor processing unit (TPU), a quantum processor, a microcontroller, or a combination thereof. The memory C2 may be, for example, a flash memory, a hard disk drive (HDD), a solid state drive (SSD), or a combination thereof.

[0089] The computer C may further include a RAM (Random Access Memory) for expanding the program P during execution and for temporarily storing various data. The computer C may also include a communication interface for transmitting and receiving data to and from other devices. The computer C may also include an input / output interface for connecting input / output devices such as a keyboard, a mouse, a display, and a printer.

[0090] The program P can also be recorded on a non-transitory, tangible recording medium M that can be read by the computer C. Such a recording medium M can be, for example, a tape, a disk, a card, a semiconductor memory, or a programmable logic circuit. The computer C can acquire the program P via such a recording medium M. The program P can also be transmitted via a transmission medium. Such a transmission medium can be, for example, a communication network or broadcast waves. The computer C can also acquire the program P via such a transmission medium.

[0091] [Additional Note 1] The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the claims. For example, embodiments obtained by appropriately combining the technical means disclosed in the above-described embodiments are also included in the technical scope of the present invention.

[0092] [Additional Note 2] Part or all of the above-described embodiment can also be described as follows: However, the present invention is not limited to the following described aspects.

[0093] (Supplementary Note 1) An information processing device comprising: a generation means for generating a question or hypothesis according to a recommendation target determined to be recommended to a subject; and a presentation means for presenting an answer to the question generated by the generation means or a verification result of the hypothesis generated by the generation means.

[0094] (Supplementary Note 2) The information processing device according to Supplementary Note 1, further comprising: a recommendation reason generating means for generating a recommendation reason for the recommended object, wherein the generating means generates the question or the hypothesis based on the recommendation reason.

[0095] (Supplementary Note 3) An information processing device according to Supplementary Note 1 or 2, comprising a recommendation means for determining a recommendation target according to the attributes of a subject person using a prediction model generated by learning the relationship between the attributes of the subject person and the result of recommending a recommendation target to the subject person with the attributes, or the relationship between the attributes of the subject person and the recommendation target to be recommended to the subject person with the attributes, and the generation means for generating the question or the hypothesis based on the accuracy of the prediction result of the prediction model.

[0096] (Supplementary Note 4) The information processing device according to any one of Supplementary Notes 1 to 3, wherein the generating means generates the question or the hypothesis based on an attribute of the subject.

[0097] (Supplementary Note 5) The information processing device according to any one of Supplementary Notes 1 to 4, wherein the generating means generates the question or the hypothesis based on an attribute of a recommender who recommends the recommendation target to the subject.

[0098] (Supplementary Note 6) The information processing device according to any one of Supplementary Notes 1 to 5, wherein the generation means generates the question or the hypothesis using a generation model generated by learning a relationship between a recommendation target and a question or a hypothesis corresponding to the recommendation target.

[0099] (Supplementary Note 7) An information processing device described in any of Supplementary Notes 1 to 6, comprising a response means for generating an answer to the question generated by the generation means or a verification result of the hypothesis generated by the generation means, wherein when a question is input regarding the answer or the verification result presented by the presentation means, the response means generates an answer to the input question, and the presentation means presents the answer to the input question generated by the response means.

[0100] (Appendix 8) An information presentation method including: at least one processor generating a question or hypothesis according to a recommendation target determined to be recommended to a subject; and presenting an answer to the generated question or a verification result of the generated hypothesis.

[0101] (Appendix 9) An information presentation program that causes a computer to function as a generation means that generates questions or hypotheses according to recommendation targets determined to be recommended to a subject, and a presentation means that presents answers to the questions generated by the generation means or verification results of the hypotheses generated by the generation means.

[0102] [Additional Note 3] Some or all of the above-described embodiments can also be expressed as follows: An information processing device including at least one processor, the processor executing a process of generating a question or a hypothesis according to a recommendation target determined to be recommended to a subject, and a process of presenting an answer to the generated question or a verification result of the generated hypothesis.

[0103] The information processing device may further include a memory that stores an information presentation program for causing the processor to execute the process of generating a question or hypothesis and the process of presenting an answer or verification result. The information presentation program may be recorded on a computer-readable, non-transitory, tangible recording medium.

[0104] REFERENCE SIGNS LIST 1, 2 Information processing device 201 Recommendation unit (recommendation means) 202 Recommendation reason generation unit (recommendation reason generation means) 11, 203 Generation unit (generation means) 204 Response unit (response means) 12, 205 Presentation unit (presentation means) 211 Prediction model 212 Generation model

Claims

1. A generating means for generating a question or hypothesis according to a recommendation target determined as a target to be recommended to the subject; a presentation means for presenting an answer to a question generated by the generation means, or a verification result of a hypothesis generated by the generation means.

2. a recommendation reason generating means for generating a recommendation reason for the recommended object, The information processing apparatus according to claim 1 , wherein the generating means generates the question or the hypothesis based on the reason for recommendation.

3. a recommendation means for determining a recommendation target according to an attribute of a target person by using a prediction model generated by learning a relationship between an attribute of a person and a result of recommending a recommendation target to the person having the attribute, or a relationship between an attribute of a person and a recommendation target to be recommended to the person having the attribute, The information processing apparatus according to claim 1 , wherein the generating means generates the question or the hypothesis according to a degree of accuracy of a prediction result of the prediction model.

4. The information processing apparatus according to claim 1 , wherein the generating means generates the question or the hypothesis based on an attribute of the subject.

5. The information processing apparatus according to claim 1 , wherein the generating means generates the question or the hypothesis based on an attribute of a recommender who recommends the recommendation target to the subject.

6. The information processing device according to claim 1 , wherein the generating means generates the question or the hypothesis using a generation model generated by learning a relationship between a recommendation target and a question or a hypothesis corresponding to the recommendation target.

7. a response means for generating an answer to the question generated by the generation means or a verification result of the hypothesis generated by the generation means, the response means, when a question regarding the answer or the verification result presented by the presentation means is input, generates an answer to the input question; The information processing apparatus according to claim 1 , wherein the presenting means presents an answer to the input question, the answer being generated by the response means.

8. At least one processor generating a question or hypothesis according to the recommendation object determined to be recommended to the subject; presenting an answer to the generated question or a verification result of the generated hypothesis.

9. Computer, A generating means for generating a question or hypothesis according to a recommendation object determined to be recommended to the subject; and an information presentation program that causes the information presentation program to function as a presentation means that presents an answer to a question generated by the generation means, or a verification result of a hypothesis generated by the generation means.