Message optimization device, message optimization method, and program

Abstract

To make it possible to further enhance the presentation effectiveness of messages including benefits.SOLUTION: A message optimization device of an embodiment is configured to: store a plurality of candidate messages each having feature quantities formed by combining elements representing perspectives on benefits and elements representing ways of conveying the benefits; when a target user to whom a message is to be presented is specified in this state, acquire characteristic data of a target user, input the acquired data to a score predictor, and acquire a message evaluation score corresponding to the characteristic data of the target user; select a candidate message for which a value calculated from the feature quantities of the candidate message and the message evaluation score satisfies a predetermined condition from among the plurality of candidate messages, and present the selected candidate message to the target user.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] One aspect of the present invention relates to a message optimization device, method, and program for generating a message to encourage a person to take a predetermined action, for example. [Background technology]

[0002] Presenting messages is one way to encourage people to change their behavior in areas such as health, congestion relief, and disaster prevention. Messages encouraging behavior change generally include information about the benefits that can be obtained from the recommended behavior. For example, a message encouraging health-related behavior change might include information such as "quitting smoking halves the risk of lung cancer." Including information about benefits in messages can be expected to have a certain effect in encouraging people to change their behavior.

[0003] Meanwhile, techniques for effectively communicating benefits include, for example, a method that focuses on psychological findings such as cognitive dissonance (see, for example, Non-Patent Document 1) and a method that incorporates nudges from behavioral economics, such as MINDSPACE (see, for example, Non-Patent Documents 2 or 3). Of these, the method described in Non-Patent Document 2 in particular enhances the effectiveness of message presentation by selecting the way in which benefits are communicated according to a person's personality traits. [Prior art documents] [Non-patent literature]

[0004] [Non-Patent Document 1] Tae Sato, Ryosuke Aoki, and Munenori Koyasu, "Message Presentation Methods for Cognitive Dissonance to Induce Behavioral Change," Multimedia, Distributed, Collaborative and Mobile Symposium, DICOMO 2019. [Non-patent document 2] T. Hamatani, W. Yamada, and K. Ochiai, "Consideration of methods for presenting social messages to change health-related behavior," Information Processing Society of Japan Research Report, Vol. 2021-DPS-187 No. 5, Vol. 2021-MBL-99 No. 5, Vol. 2021-ITS-85 No. 5, May 27, 2021. [Non-patent document 3] Hirofumi Matsumoto, "Promoting stair use using nudges among young women: Are environmental conservation messages effective?", Physical Education Research, Vol. 67, 319-327, 2022. Summary of the Invention [Problem to be solved by the invention]

[0005] However, the technology described in the above non-patent document selects how to convey benefits depending on a person's personality traits, and depending on the benefits, the effectiveness of presenting the message may be limited.

[0006] The present invention has been made in light of the above circumstances, and aims to provide a technique that makes it possible to further enhance the effectiveness of presenting a message that includes benefits. [Means for solving the problem]

[0007] In order to solve the above problem, one aspect of a message optimization device or message optimization method according to the present invention, when presenting a message containing information about benefits to a user to encourage the user to change their behavior, first stores multiple candidate messages having features that combine elements that represent the perspective of the benefits and elements that represent the manner of conveying the benefits. In this state, when a target user is designated, characteristic data of the target user is acquired and input to a score predictor, and a message evaluation score corresponding to the target user's characteristic data is obtained from the score predictor. Then, from the multiple candidate messages, a candidate message is selected whose value calculated from the feature values ​​and the message evaluation score satisfies a predetermined condition, and the selected candidate message is presented to the target user.

[0008] According to one aspect of the present invention, for example, a candidate message that is appropriate in terms of both the benefits and the way it is conveyed is selected and sent depending on the characteristics of the target user, thereby making it possible to present a message that is more effective in encouraging behavioral change to the user. [Effects of the Invention]

[0009] That is, according to one aspect of the present invention, it is possible to provide a technology that can further enhance the effectiveness of presenting a message containing benefits. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram showing an example of the configuration of a message delivery system including a message optimization device according to an embodiment of the present invention. [Figure 2] FIG. 2 is a block diagram showing an example of a hardware configuration of a message optimization device according to an embodiment of the present invention. [Figure 3] FIG. 3 is a block diagram showing an example of the software configuration of a message optimization device according to an embodiment of the present invention. [Figure 4] FIG. 4 is a flowchart showing an example of the procedure and content of the score predictor estimation process executed by the control unit of the message optimization device shown in FIG. [Figure 5] FIG. 5 is a flowchart showing an example of the processing procedure and processing content of the message optimization processing executed by the control unit of the message optimization device shown in FIG. [Figure 6] FIG. 6 is a flowchart showing a specific example of the message optimization process shown in FIG. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0012] [One embodiment] (overview) When presenting a message to a user that includes benefits of a recommended action, factors that affect the effectiveness of the message include "how the benefits are communicated" and "the perspective of the benefits."

[0013] There are four ways to communicate the benefits: (1) Messenger: Emphasize that the recommendation comes from an authority or important person. (2) Incentives: Emphasize the disadvantages of not taking action (3) Social norms: emphasize that many other people are doing it (4) Simplicity: Emphasize the ease and simplicity of the recommended action.

[0014] On the other hand, the "perspective of benefits" can be considered to be of three types, depending on the entity that will enjoy the benefits. (1) Self-help: The benefits you gain through your own recommended actions (2) Mutual aid: Benefits that members of a nearby community receive through one's recommended actions (3) Public assistance: Benefits gained by the government or local government through one's own recommended actions.

[0015] For example, the benefits of stockpiling in preparation for a disaster are not only "self-help," which reduces one's own risk in the event of a disaster, but also "mutual help," which allows more residents to use stockpiles at evacuation shelters, and "public help," which reduces the amount of stockpiles required by local governments. Which benefit resonates most with the message recipient is likely to depend on the recipient's personality and values.

[0016] Therefore, one embodiment of the present invention prepares multiple candidate messages that include both elements of "how to convey the benefits" and "perspective of the benefits," and selects the most appropriate candidate message from the multiple candidate messages and presents it to the user depending on the characteristics of the user to whom the message is to be presented.

[0017] According to one embodiment of the present invention, the most appropriate message is selected and presented to the user, taking into consideration both the "way of communicating the benefits" and the "perspective of the benefits" depending on the characteristics of the user to whom the message is presented. As a result, it is expected that an increase in users who receive the message and actively engage in the recommended behavior can be expected.

[0018] (Configuration example) (1) System FIG. 1 is a diagram showing an example of the configuration of a message delivery system including a message optimization device CS according to an embodiment of the present invention.

[0019] The system according to one embodiment includes a message optimization device CS. The message optimization device CS optimizes a plurality of user terminals UT1 to UT2 used by users to whom messages are to be delivered. N and the management database DB via the network NW.

[0020] The management database (DB) collects and manages characteristic information about each of multiple users who may be the target of messages. The characteristic information includes, for example, demographic attributes (age, gender, family structure, etc.), personality traits (Big Five, etc.), and psychological traits (CD-RISC, self-efficacy, etc.).

[0021] The management database DB also stores and manages multiple candidate messages created in advance by a system administrator or the like according to the type of recommended action. The candidate messages take into consideration both how to convey the benefits and the perspective of the benefits.

[0022] Furthermore, the management database DB collects and manages the feature quantities and user evaluation scores for each of a plurality of messages that have been presented to the user in the past.

[0023] User terminals UT1 to UT NThe user terminals UT1 to UT2 are, for example, smartphones used by users, and receive messages from the message optimization device CS that encourage recommended actions, for example, for health or disaster prevention. N The device is not limited to a smartphone, but may also be a personal computer or a tablet terminal.

[0024] The network NW comprises a wide area network with the Internet at its core, and an access network for accessing this wide area network. The access network may be, for example, a public communication network using wired or wireless connections, or a local area network (LAN) using wired or wireless connections, but is not limited to these.

[0025] (2) Message Optimization Device CS 2 and 3 are block diagrams showing an example of the hardware configuration and software configuration of the message optimization device CS, respectively.

[0026] The message optimization device CS is provided in, for example, a server computer located on the Web or cloud, and includes a control unit 1 that uses a hardware processor such as a central processing unit (CPU). A storage unit having a program storage unit 2 and a data storage unit 3, and a communication interface (hereinafter, interface will be referred to as I / F) unit 4 are connected to the control unit 1 via a bus 5.

[0027] Of these, the communication I / F unit 4 uses a communication protocol defined in the network NW to communicate with the management database DB and the user terminals UT1 to UT2. N It transmits and receives information data between them.

[0028] The program storage unit 2 is a combination of a non-volatile memory that can be written to and read from at any time, such as an SSD (Solid State Drive), and a non-volatile memory such as a ROM (Read Only Memory), and stores middleware such as an OS (Operating System) as well as application programs required to execute various controls according to an embodiment. Hereinafter, the OS and each application program will be collectively referred to as the program.

[0029] The data storage unit 3 is a combination of a non-volatile memory such as an SSD that can be written to and read from at any time, and a volatile memory such as a RAM (Random Access Memory), and its storage area includes a user characteristic data storage unit 31, an evaluation score data storage unit 32, a score predictor storage unit 33, a candidate message list storage unit 34, and an optimized message storage unit 35.

[0030] The user characteristic data storage unit 31 stores data representing the characteristics of a plurality of users, which is acquired from the management database DB.

[0031] The evaluation score data storage unit 32 stores the feature amounts and evaluation scores of a plurality of messages previously presented to the user, which are acquired from the management database DB.

[0032] The score predictor storage unit 33 stores model data such as parameters relating to a score predictor created by a machine learning model, for example.

[0033] The candidate message list storage unit 34 stores a plurality of candidate messages acquired from the management database DB.

[0034] The optimized message storage unit 35 temporarily stores the optimized message selected from the plurality of candidate messages until it is sent to the user.

[0035] The control unit 1 includes the following processing functions necessary to implement one embodiment of the present invention: a user characteristic data acquisition processing unit 11, an evaluation score data acquisition processing unit 12, a score predictor estimation processing unit 13, a candidate message list acquisition processing unit 14, a target user information acquisition processing unit 15, a message optimization processing unit 16, and an optimized message transmission processing unit 17.

[0036] The processing units 11 to 17 are all realized by causing a hardware processor of the control unit 1 to execute an application program stored in the program storage unit 2. Note that some or all of the processing units 11 to 17 may be realized using hardware such as an LSI (Large Scale Integration) or an ASIC (Application Specific Integrated Circuit).

[0037] During the estimation phase of the score predictor, the user characteristic data acquisition processing unit 11 acquires characteristic data of multiple users who may be the target of message presentation from the management database DB via the communication I / F unit 4, and stores the acquired characteristic data of the multiple users in the user characteristic data storage unit 31.

[0038] In the estimation phase of the score predictor, the evaluation score data acquisition processing unit 12 acquires data indicating the features and evaluation scores of multiple messages previously presented to the user from the management database DB via the communication I / F unit 4, and stores the acquired data indicating the features and evaluation scores of the multiple messages in the evaluation score data storage unit 32.

[0039] The score predictor estimation processing unit 13 uses each user characteristic data stored in the user characteristic data storage unit 31 and the evaluation score data of the message stored in the evaluation score data storage unit 32 as learning data to train a score predictor and estimates a score predictor that predicts the evaluation score of a message from the user's characteristics. The score predictor estimation processing unit 13 stores model data such as parameters related to the estimated score predictor in the score predictor storage unit 33. An example of the score predictor estimation process will be described in detail in the operation example.

[0040] During the message optimization phase, the candidate message list acquisition processing unit 14 acquires a list of candidate messages from the management database DB via the communication I / F unit 4, and stores the acquired list of candidate messages in the candidate message list storage unit 34.

[0041] The target user information acquisition processing unit 15 receives characteristic data relating to a user designated as a target to which a message is to be presented.

[0042] The message optimization processing unit 16 uses the score predictor to select an optimal candidate message for the user to be presented based on the acquired user characteristic data of the user to be presented from among multiple candidate messages stored in the candidate message list storage unit 34. Then, the message optimization processing unit 16 stores the selected optimal candidate message in the optimized message storage unit 35. An example of the message optimization processing will be described in detail in the operation example.

[0043] The optimized message transmission processing unit 17 transmits the optimized message stored in the optimized message storage unit 35 from the communication I / F unit 4 to the user terminals UT1 to UT2 owned by the users to whom the message is to be presented. N Send to.

[0044] (Example of operation) Next, an example of the operation of the message optimization device CS configured as above will be described.

[0045] Here, we will assume that the characteristic data of multiple users to whom messages may be presented, multiple candidate messages prepared in advance as candidates for presentation, and users' evaluation scores for messages presented in the past are stored in advance in the management database DB.

[0046] (1) Score predictor estimation phase FIG. 4 is a flowchart showing an example of a processing procedure and processing contents in the score predictor estimation phase executed by the control unit 1 of the message optimization device CS.

[0047] (1-1) Acquisition of user characteristic data For example, when an instruction to start execution of the estimation phase is input from a system administrator or the like and this start instruction is detected in step S10, the control unit 1 of the message optimization device CS first accesses the management database DB under the control of the user characteristic data acquisition processing unit 11 in step S11. Then, the control unit 1 acquires characteristic data of multiple users who may be targets of message presentation from the management database DB via the communication I / F unit 4, and stores the acquired characteristic data of the multiple users in the user characteristic data storage unit 31.

[0048] The user characteristic data is a numerical vector {u n}≡(u1,…,u N ) However, the way to express characteristics using numerical vectors is arbitrary. For example, if only gender is taken into consideration, a two-dimensional one-hot vector representation is used, where males are represented as (0,1) and females as (1,0). If age is also taken into consideration, a three-dimensional vector representation is used, where a 35-year-old male is represented as (0,1,35).

[0049] (1-2) Acquisition of evaluation score data Next, in step S12, the control unit 1 of the message optimization device CS accesses the management database DB under the control of the evaluation score data acquisition processing unit 12. Then, data indicating the feature amounts and evaluation scores of multiple messages previously presented to the user is acquired from the management database DB via the communication I / F unit 4, and the acquired data indicating the feature amounts and evaluation scores of the multiple messages is stored in the evaluation score data storage unit 32.

[0050] The evaluation score data is the number of messages presented to the user in the past {M n}≡(M1,…,M N ) and the feature value of each message {c j n}≡(c 1 n ,…,c Mn n ) and the user's rating score for each message {s j n}≡(s 1 n ,…,s Mn n ) and

[0051] Here, the message features {c j n} is expressed as a 12-dimensional vector by combining three types of perspectives on benefits and four types of ways of communicating them. For example, if the perspective on benefits is "self-help" and the way of communication is "messenger," it is expressed as (1,0,0,0,0,0,0,0,0,0,0,0,0), if the perspective on benefits is "mutual help" and the way of communication is "messenger," it is expressed as (0,1,0,0,0,0,0,0,0,0,0,0,0), and if the perspective on benefits is "public help" and the way of communication is "messenger," it is expressed as (0,0,1,0,0,0,0,0,0,0,0,0,0).

[0052] The evaluation score is expressed as a one-dimensional numerical value, representing the strength of the message's effectiveness in encouraging the recommended behavior. For example, it can be determined by collecting responses from users through a questionnaire or by investigating whether or not users actually performed the recommended behavior. In a questionnaire, for example, questions such as "How much did you want to perform the recommended behavior? Please answer with the appropriate number from 1 to 10" are used. In a survey on the implementation of recommended behavior, for example, "1 indicates that the behavior was performed, and 0 indicates that the behavior was not performed."

[0053] (1-3) Estimation of the score estimator Next, in step S13, the control unit 1 of the message optimization device CS, under the control of the score predictor estimation processing unit 13, executes the score predictor estimation process using the acquired user characteristic data and the message evaluation score data as follows.

[0054] In other words, the score predictor has a parameter θ, and is defined to take user characteristics u as input and output a function f(u|θ) expressed as a 12-dimensional numerical value according to these user characteristics. The output function f(u|θ) corresponds to the evaluation scores for the 12 types of message characteristics. For example, if the output function f is (0.9, 0.5, 0.3, 0.1, 0.1, 0.1, 0, 0, 0, 0, 0), this indicates that the combination of "self-help" as the merit perspective and "messenger" as the communication method has the highest evaluation score, and the combination of "mutual help" as the merit perspective and "messenger" as the communication method has the next highest evaluation score.

[0055] The parameters θ are learned from the data according to the following equation:

number

[0056] Here, L(x, y) is a loss function, and an example of the loss function is the L2 norm loss function L(x, y) = ||xy|| 2 and the L1 norm loss function L(x,y)=√(||xy|| 2 ) etc.

[0057] The score predictor estimation processing unit 13 outputs the parameter θ and the function f(u|θ) as the estimation result of the score predictor. Then, in step S14, the score predictor estimation processing unit 13 stores the output parameter θ and function f(u|θ) in the score predictor storage unit 33 as model data.

[0058] The control unit 1 of the message optimization device CS repeatedly executes the estimation process of the score predictor described above using each user data and each evaluation score data, and when it determines in step S15 that the estimation process has ended, it ends the estimation phase of the score predictor.

[0059] (2) Message optimization phase FIG. 5 is a flowchart showing an example of the processing procedure and processing contents of the message optimization phase executed by the control unit 1 of the message optimization device CS.

[0060] (2-1) Obtaining a list of candidate messages When a request to start message optimization is input from, for example, a system administrator or a user, and the control unit 1 of the message optimization device CS detects this start request in step S20, first in step S21, the control unit 1 accesses the management database DB under the control of the candidate message list acquisition processing unit 14. Then, the control unit 1 acquires a list of candidate messages from the management database DB via the communication I / F unit 4, and stores the acquired list of candidate messages in the candidate message list storage unit 34.

[0061] The candidate messages may be created and listed in advance in the message optimization device CS. In this case, the candidate message list acquisition processing unit 14 reads out the candidate message list from the data storage unit 3.

[0062] The list of candidate messages is made up of the sentences ({q k}≡(q1,…,q K) and the feature of each candidate message ({c k}≡(c1,…,c K ) where the message features {c k} is the evaluation score data {s j n} and is expressed as a 12-dimensional vector.

[0063] (2-2) Acquisition of characteristic data of target users Subsequently, in step S22, the control unit 1 of the message optimization device CS, under the control of the target user information acquisition processing unit 15, acquires user information designated as a target to which the message is to be presented.

[0064] For example, the target user information acquisition processing unit 15 may be a system administrator terminal (not shown) or a user terminal UT1 to UT N When the identification information (user ID) of the presentation target user is received from the management database DB, the management database DB is accessed based on the received user ID, and the characteristic data of the presentation target user is received from the management database DB.

[0065] The characteristic data of the user to be presented is stored in the administrator terminal or the user terminals UT1 to UT N The user characteristics may be obtained directly from the user characteristic data {u n}≡(u1,…,u N ), it is expressed as a numeric vector.

[0066] (2-3) Message optimization Next, in step S23, the control unit 1 of the message optimization device CS, under the control of the message optimization processing unit 16, executes a process of optimizing the message in accordance with the characteristics of the specified presentation target user as follows.

[0067] FIG. 6 is a flowchart showing an example of the processing procedure and processing content of the message optimization processing executed by the message optimization processing unit 16. That is, first, in step S30, the message optimization processing unit 16 inputs the acquired characteristic data of the presentation target user to the score predictor. Then, in step S31, the message optimization processing unit 16 acquires a predicted score corresponding to the characteristic of the presentation target user from the score predictor. At this time, the user characteristic is denoted by u and the predicted score is denoted by Z. u Then, the predicted score Z u teeth Z u =f(u|θ) It is expressed as:

[0068] Next, in step S32, the message optimization processor 16 selects one candidate message from the candidate message list stored in the candidate message list storage unit 34. Then, in step S33, the message optimization processor 16 calculates the feature quantity of the selected candidate message and the predicted score Z u The inner product of is calculated, and it is determined in step S34 whether the calculated inner product value is maximum.

[0069] The candidate message q* that maximizes the inner product can be found as follows:

number

[0070] If the result of the above determination is that the newly calculated inner product value is the largest among the candidate messages selected so far, the message optimization processor 16 temporarily stores the candidate message in the optimized message storage unit 35 .

[0071] Then, in step S36, the message optimization processor 16 determines whether or not the selection of all candidate messages has been completed. If unselected candidate messages remain, the process returns to step S32 to select the next candidate message, and the process of steps S33 to S36 is executed again for this selected candidate message. Thereafter, the process of steps S33 to S36 is repeated for all candidate messages in the same manner.

[0072] If it is determined in step S34 that the calculated inner product value is smaller than the inner product value of the candidate message temporarily stored in the optimized message storage unit 35, the message optimization processor 16 returns to step S32.

[0073] When it is determined in step S36 above that the selection of all candidate messages has been completed, in step S37 the message optimization processing unit 16 determines that the candidate message finally stored in the optimized message storage unit 35 is the message that is optimal in terms of benefits and delivery method for the characteristics of the target user, and determines it as the message to be sent.

[0074] Note that the candidate message selected as the optimal message does not necessarily have to be one; if there are multiple candidate messages with the largest inner product value, all of these candidate messages may be selected, or multiple candidate messages with inner product values ​​greater than or equal to a threshold value may be selected.

[0075] (2-4) Sending a message Finally, in step S24, the control unit 1 of the message optimization device CS reads out the message from the optimized message storage unit 35 under the control of the optimized message transmission processing unit 17, and transmits the read message q* from the communication I / F unit 4 to the user terminals UT1 to UT2 owned by the users to be presented. N Send to.

[0076] If there are multiple messages to be sent, the messages should be sent in ascending or descending order of the dot product values. Alternatively, the messages and dot product values ​​should be sent in pairs. By sending messages in this way, the recipient user terminals UT1 to UT N In this case, the order in which messages are presented can be determined according to the dot product value.

[0077] In step S25, the control unit 1 of the message optimization device CS determines whether the message optimization and presentation process has been completed for all users designated as presentation targets. If there are still users to be presented, the process returns to step S22, acquires characteristic data of the next user, and then executes the optimized message selection and transmission process in steps S23 to S25. On the other hand, when the process for all users to be presented has been completed, the control unit 1 ends the process and returns to a standby state.

[0078] (effect) As described above, in one embodiment, the feature quantities of a message are expressed as a 12-dimensional vector that combines, for example, three types of merit perspectives and four types of communication methods. Then, a score predictor that outputs a 12-dimensional message evaluation score in response to input of user characteristics is estimated using multiple pieces of user characteristic data and the feature quantities and evaluation scores of multiple messages previously presented to users as training data. Next, when a target user is designated, the characteristic data of the user is input to the score predictor to obtain a corresponding evaluation score. The candidate message with the highest inner product value between the feature quantities and the obtained evaluation score is selected from multiple candidate messages prepared in advance and transmitted to the target user terminal.

[0079] Therefore, depending on the characteristics of the target user, a candidate message that is appropriate in terms of both the benefits and the way it is conveyed will be selected and sent, making it possible to present a message that is more effective in encouraging behavioral change to the user.

[0080] [Other embodiments] (1) In one embodiment, the message optimization device has been described as having both a processing function for the estimation phase of the score predictor and a processing function for the message optimization phase. However, the message optimization device may have only a processing function for the message optimization phase. In this case, the message optimization process can be performed by acquiring model data, such as parameters related to the estimated score predictor, from an external device in advance and storing it in the score predictor storage unit 33.

[0081] (2) In one embodiment, the characteristic data of multiple users who may be presented, the evaluation score data of multiple messages presented in the past, and the candidate message list are acquired from the management database DB and used for the score predictor estimation and message optimization process. However, if the user characteristic data, message evaluation score data, and candidate message list are stored in advance in the data storage unit 3 of the message optimization device CS, there is no need to acquire them from the management database DB.

[0082] (3) In one embodiment, the functions of the message optimization device CS are provided on a server computer located on the Web or the cloud, but the functions may also be provided on an edge computer on a LAN or a personal computer such as an administrator terminal or user terminal used by a system administrator.

[0083] (4) Furthermore, the message optimization device CS may be realized not only in the form of a centralized system such as a server computer, an edge computer, or a personal computer, but also in the form of a distributed system. For example, one or more of the user characteristic data acquisition processing unit 11, the evaluation score data acquisition processing unit 12, the score predictor estimation processing unit 13, the candidate message list acquisition processing unit 14, the target user information acquisition processing unit 15, the message optimization processing unit 16, and the optimized message transmission processing unit 17 shown in Fig. 3 may be realized by distributing and executing one or more other control devices, computers, etc.

[0084] (5) In addition, the functional configuration of the message optimization device, the processing procedure and processing content, the types of merit perspectives when defining a message, the types of methods of communication, etc. can be modified and implemented in various ways without departing from the spirit of this invention.

[0085] Although the embodiments of the present invention have been described in detail above, the above description is merely an example of the present invention in every respect. It goes without saying that various improvements and modifications can be made without departing from the scope of the present invention. In other words, when implementing the present invention, specific configurations according to the embodiments may be appropriately adopted.

[0086] In short, this invention is not limited to the above-described embodiments, and in the implementation stage, the components can be modified and embodied without departing from the spirit of the invention. Furthermore, various inventions can be formed by appropriately combining multiple components disclosed in the above-described embodiments. For example, some components may be omitted from all the components shown in the embodiments. Furthermore, components from different embodiments may be appropriately combined. [Explanation of symbols]

[0087] CS: Message Optimizer DB: Management database UT1~UT N ...User terminal 1...Control unit 2...Program memory section 3...Data storage unit 4...Communication I / F section 5. Bus 11...User characteristic data acquisition processing unit 12...Evaluation score data acquisition processing unit 13...Score predictor estimation processing unit 14...Candidate message list acquisition processing unit 15...Target user information acquisition processing unit 16...Message optimization processing unit 17...Optimized message sending processing unit 31...User characteristic data storage unit 32...Evaluation score data storage unit 33...Score predictor memory unit 34...Candidate message list storage section 35...Optimized message storage unit

Claims

1. A message optimization device that presents a message including information about benefits to a user in order to encourage the user to change their behavior, a first processing unit that stores a plurality of candidate messages each having a feature that combines an element that represents a viewpoint regarding the merit and an element that represents a manner of conveying the merit; a second processing unit including a score predictor that predicts an evaluation score related to the feature amount of the message based on the characteristic data of the user, and when a presentation target user is designated, acquires the characteristic data of the presentation target user and inputs it to the score predictor, and acquires an evaluation score corresponding to the characteristic data of the presentation target user from the score predictor; a third processing unit that selects, from the plurality of candidate messages, a candidate message whose value calculated from the feature amount and the evaluation score of the candidate message satisfies a predetermined condition; a fourth processing unit that presents the selected candidate message to the presentation target user; A message optimization device comprising:

2. 2. The message optimization device according to claim 1, wherein the third processing unit calculates an inner product value between the feature value of each candidate message and the evaluation score from among the plurality of candidate messages, and selects the candidate message for which the calculated inner product value is maximum.

3. The message optimization device according to claim 1, wherein each feature of the candidate message is represented by multi-dimensional vector data that combines elements representing multiple perspectives on the expected benefits and elements representing multiple ways of conveying them.

4. The message optimization device according to claim 3 , wherein the viewpoints are of three types: self-help, mutual help, and public help.

5. The message optimization device according to claim 1, further comprising a fifth processing unit that estimates the score predictor based on characteristic data of a plurality of the users, feature quantities of a plurality of previously presented messages that have been presented to the users in the past, and the evaluation scores by the users for the previously presented messages.

6. A message optimization method for executing, by an information processing device, a process of presenting a message including information about benefits to a user in order to encourage the user to change their behavior, comprising: a step of storing a plurality of candidate messages each having a feature value that combines an element representing a viewpoint regarding the merit and an element representing a manner of conveying the merit; a score predictor for predicting an evaluation score for the feature of the message based on the user's characteristic data, and when a presentation target user is designated, acquiring the characteristic data of the presentation target user and inputting it into the score predictor, and acquiring an evaluation score corresponding to the characteristic data of the presentation target user from the score predictor; selecting, from the plurality of candidate messages, a candidate message whose value calculated from the feature amount and the evaluation score of the candidate message satisfies a predetermined condition; presenting the selected candidate message to the presentation target user; A message optimization method comprising:

7. 6. A program for causing a processor included in the message optimization device to execute processing executed by at least one of the processing units included in the message optimization device according to claim 1.

Images