Intervention device

Abstract

In the present invention, a calculation unit (151) calculates the degree of agreement of users with respect to an account on the basis of the content of reactions of the users with respect to a posting from the account in an SNS. A prediction unit (152) predicts the degree of spread of the influence of information sharing by the account on the basis of the degree of agreement and information indicating the relationship between the users in the SNS. A selection unit (153) selects an intervention target user from among the users of the SNS on the basis of the degree of spread of the influence.

Description

intervention device

[0001] The present invention relates to interventional devices.

[0002] BACKGROUND ART Conventionally, social networking services (SNSs) have been utilized as communication tools between companies, organizations, or groups and users. For example, SNSs are utilized for marketing activities, news dissemination, and the like.

[0003] On the other hand, there are often cases where negative opinions (anti-comments, negative campaigns, etc.) arise in the process of spreading information in response to posts from specific accounts on SNS, and opinions undesirable for the source of such information spread. To address this problem, a technique is known for selecting users who should post information first in order to suppress the generation and spread of negative opinions (see, for example, Non-Patent Document 1).

[0004] W. Chen et al. “Influence Maximization in Social Networks When Negative Opinions May Emerge and Propagate,” SDM, 2011. [online], [Retrieved October 19, 2023], Internet (https: / / epubs.siam.org / doi / 10.1137 / 1.9781611972818.33)

[0005] However, conventional techniques have the problem that they cannot prevent the negative impact caused by negative posts against specific accounts on SNS.

[0006] For example, in SNS, advertisements and news are generally transmitted from specific accounts operated by companies, etc. On the other hand, the technology described in Non-Patent Document 1 is not applicable when the source of information is a specific account of a company, etc., because the information is transmitted by arbitrarily selected users.

[0007] In order to solve the above-mentioned problems and achieve the objectives, the intervention device is characterized by having: a calculation unit that calculates a user's degree of agreement with an account on an SNS based on the content of the user's reaction to a post from the account; a prediction unit that predicts a degree of influence of information disseminated by the account on the SNS based on information indicating relationships between users and the degree of agreement; and a selection unit that selects a user to be intervened from among users of the SNS based on the degree of influence.

[0008] According to the present invention, it is possible to prevent the adverse effects caused by negative posts to specific accounts on SNS.

[0009] FIG. 1 is a diagram showing an example of the configuration of an intervention device according to the first embodiment. FIG. 2 is a diagram showing an example of a reaction history. FIG. 3 is a diagram explaining influence diffusion. FIG. 4 is a diagram explaining influence diffusion. FIG. 5 is a diagram showing an algorithm of MIOA-greedy for Interv-IM. FIG. 6 is a diagram showing an algorithm of MIOA-top-k for Interv-IM. FIG. 7 is a flowchart showing the processing flow of the intervention device according to the first embodiment. FIG. 8 is a diagram explaining the effect of the first embodiment. FIG. 9 is a diagram showing an example of a computer that executes an intervention program.

[0010] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of an interventional device according to the present invention will be described in detail below with reference to the accompanying drawings. However, the present invention is not limited to the embodiment described below.

[0011] In an embodiment, a user of an SNS may be an individual (natural person), or may be a company, institution, or organization. Each user has an account on the SNS and performs information sharing actions through the account. In an SNS, a user is identified by the account. An information sharing action is to transmit data such as text or images to a server or the like via the Internet or the like, thereby making the data viewable by other users. In addition, such an act of transmitting data on an SNS may be referred to as a "post."

[0012] Users transmit information by posting advertisements, news, opinions, etc. through their accounts. Information is often transmitted not only by individuals but also by accounts of news media, companies, etc.

[0013] Users may also react to posted information. Reactions are expressed by expressing opinions (posting), reposting, rating, etc. on posts for information dissemination. Reposting is making a further post quoting information such as posted text. Ratings are made by operations that indicate a positive or negative opinion of the post (for example, pressing the like button, pressing the dislike button, or selecting the number of stars).

[0014] Furthermore, reactions to posted information may be indirect. For example, other users may further react to a user's reaction to information posted by a company account.

[0015] [Configuration of First Embodiment] The configuration of an interventional device according to the first embodiment will be described with reference to Fig. 1. Fig. 1 is a diagram showing an example of the configuration of the interventional device according to the first embodiment.

[0016] The intervention device 10 selects candidate users who need intervention based on a history of user responses to messages from a specific account on the SNS. The intervention device 10 can output information indicating a list of the selected users. The intervention device 10 also performs intervention for the selected users.

[0017] As shown in FIG. 1 , the intervention device 10 includes a communication unit 11 , an input unit 12 , an output unit 13 , a storage unit 14 , and a control unit 15 .

[0018] The communication unit 11 performs data communication with other devices via a network, and is, for example, a network interface card (NIC).

[0019] The input unit 12 receives input of data. For example, the input unit 12 is an interface connected to input devices such as a mouse and a keyboard.

[0020] The output unit 13 outputs data. The output unit 13 is an interface connected to an output device such as a display and a speaker.

[0021] The storage unit 14 is a storage device such as a hard disk drive (HDD), a solid state drive (SSD), or an optical disk. The storage unit 14 may also be a rewritable semiconductor memory such as a random access memory (RAM), a flash memory, or a non-volatile static random access memory (NVSRAM). The storage unit 14 stores an operating system (OS) and various programs executed by the intervention device 10.

[0022] The storage unit 14 stores a reaction history 141. The reaction history 141 is a history of a user's reaction to a message from a specific account on the SNS. For example, the reaction history 141 is data in a table format as shown in Fig. 2. Fig. 2 is a diagram showing an example of the reaction history.

[0023] As shown in FIG. 2 , the reaction history 141 is information that associates a "respondent," a "responder," and "reaction content." The "respondent" is information that identifies the user (or account) that sent the information. The "responder" is information that identifies the user (or account) that showed a reaction to the information sent by the "respondent." The "reaction content" is information that indicates whether the reaction of the "responder" was positive or negative. The reaction history 141 may be created in advance by a device different from the intervention device 10 or manually, or may be created by the intervention device 10.

[0024] For example, a positive response is when a responder quotes a text posted by a respondent and reposts it with a positive comment such as, "I agree with this post." Alternatively, for example, a positive response may be when a responder quotes a text posted by a respondent and reposts it without a comment.

[0025] For example, a negative response is when the responder quotes the text posted by the respondent and reposts it with a negative comment such as, "I disagree with this post."

[0026] Whether a repost is positive or negative may be determined based on whether the comments added at the time of reposting contain specific keywords such as "for" or "against," or may be determined by a machine learning model for natural language analysis, or may be determined manually.

[0027] For example, the first line of the reaction history 141 in Fig. 2 indicates that the responder user 201 gave a positive reaction to a post by the respondent user 203. Also, for example, the second line of the reaction history 141 in Fig. 2 indicates that the responder user 201 gave a negative reaction to a post by the respondent user 203. In this way, because responders react on a post-by-post basis, they may give both positive and negative reactions to posts by the same respondent.

[0028] The control unit 15 controls the entire intervention device 10. The control unit 15 is, for example, an electronic circuit such as a central processing unit (CPU), a micro processing unit (MPU), or a graphics processing unit (GPU), or an integrated circuit such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA). The control unit 15 also has an internal memory for storing programs defining various processing procedures and control data, and executes each process using the internal memory. The control unit 15 also functions as various processing units by running various programs. For example, the control unit 15 has a calculation unit 151, a prediction unit 152, a selection unit 153, and an intervention unit 154.

[0029] The calculation unit 151 calculates the degree of agreement of a user with an account (hereinafter referred to as "agreement degree") based on the content of the user's reaction to a post from a specific account on SNS. Note that the specific account and user here correspond to the respondent and the responder, respectively, in the reaction history 141. The specific account and the post may also be referred to as the sender and the post, respectively.

[0030] The calculation unit 151 calculates the degree of agreement based on the response history 141. For example, the calculation unit 151 calculates, as the degree of agreement, a ratio r+ / r between the number of times r that a responder user v reposted a post of a respondent account s and the number of times r+ that a responder user v gave a positive response during a predetermined period (e.g., the most recent month).

[0031] The prediction unit 152 predicts the extent of the influence of information posted by an account based on information indicating the relationship between users on the SNS and the degree of agreement. The prediction unit 152 predicts the extent of the influence of positive and negative influences on the posting based on an influence diffusion model described below, using the social network information and the calculated degree of agreement.

[0032] Here, social network information refers to information obtained from social network data such as friendships between users, user interests, etc. The social network data may be provided by the operator of the SNS for a fee or free of charge.

[0033] The influence diffusion model will be explained below.

[0034] Each user v agrees with the sender's post q ｖ ∈[0, 1]. The degree of agreement is calculated by the calculation unit 151. The degree of agreement differs for each user.

[0035] Additionally, a user can be in one of three states: inactive, positive, or negative. Here, inactive is a state in which the user has not yet spread (reposted) the information (posted by the sender). Positive is a state in which the user has expressed a positive opinion about the information (posted by the sender) and spread (reposted). Negative is a state in which the user has expressed a negative opinion about the information (posted by the sender) and spread (reposted). Positive and negative can be rephrased as states in which the user has reacted positively and negatively, respectively.

[0036] Consider a graph based on social network information, as shown in Figure 3. Figure 3 is a diagram illustrating the diffusion of influence. Each node in the graph shown in Figure 3 represents a user. Two nodes connected by an arrow are adjacent. For example, if users have a follow or friend relationship on an SNS, the corresponding nodes are adjacent to each other in the graph.

[0037] It is assumed that the user at the node at the end of the arrow follows the user at the node from which the arrow originates. In other words, the user at the node from which the arrow originates is followed by the user at the end of the arrow. For example, user 203 is followed by user 202. Furthermore, user 201 and user 203 follow each other.

[0038] Some SNSs have a function that notifies users who follow a user when that user posts a message. This function allows users to view many of the messages posted by the users they follow.

[0039] In the influence diffusion model, when a user v in the inactive state receives information from a neighboring user in the positive state, the probability q ｖ changes to a positive state with probability 1-q ｖOn the other hand, if a user v in the inactive state receives information from a neighbor user in the negative state, the user v will change to the negative state with a probability of 1.

[0040] The prediction unit 152 predicts the state of each user according to the rules of such an influence diffusion model. For example, the prediction unit 152 simulates the state of the user using the influence diffusion model, which is a probabilistic model. Furthermore, based on the state predicted by the prediction unit 152, a diagram such as that shown in FIG. 3 can be obtained that allows understanding of the extent of influence.

[0041] Note that receiving information from a neighboring user means that the following user posts. For example, in FIG. 3, when user 201 posts, user 202 receives the information.

[0042] 3, assume that user 203 is the sender. A thumbs-up mark indicates a positive state for user 203. A thumbs-down mark indicates a negative state for user 203.

[0043] For example, a graph in which the states of all users (nodes) with respect to the user 203 are set to inactive and the degree of agreement of each user with respect to the user 203 calculated by the calculation unit 151 is given as initial values.

[0044] The prediction unit 152 predicts the state of each user using the given initial values ​​and the influence diffusion model. Fig. 3 shows the state of each user predicted by the prediction unit 152.

[0045] 3, assuming that user 203 has posted, prediction unit 152 predicts that user 201's status with respect to user 203 will be positive. Furthermore, assuming that user 201 has posted (e.g., a repost of user 203's post), prediction unit 152 predicts that user 205's status with respect to user 203 will be negative.

[0046] In this way, the prediction unit 152 predicts whether the second user's state is the first state, the second state, or the third state based on information indicating whether the first user's state is a first state (positive) that indicates a positive attitude toward the account, a second state (negative) that indicates a negative attitude toward the account, or a third state (inactive) that indicates neither a positive nor a negative attitude toward the account, and the degree of agreement of a second user who has a specific relationship with the first user on the SNS. The specific relationship may be, for example, a follow relationship. For example, the first user is followed by the second user. Furthermore, positive, negative, and inactive are examples of the first state, the second state, and the third state, respectively.

[0047] The selection unit 153 selects, as users to be targeted for intervention, users who will have an impact on the spread of influence from among the users of the SNS, based on the degree of spread of influence predicted by the prediction unit 152. The selection unit 153 selects users who are likely to increase the number of users in a positive state when the agreement level is changed by intervention.

[0048] 3, since user 201 is in a positive state, if adjacent user 205 agrees more with user 203, the likelihood that user 205 will be in a positive state increases. In this case, it is expected that users 206 and 207, who are adjacent to user 205, will also be in a positive state.

[0049] If the intervention is to increase the user's agreement, the intervention is formulated as in equation (1).

[0050]

[0051] x ｖ is the degree of increase in the agreement level due to the intervention. ε is a coefficient. In other words, the agreement level q of user v increases as a result of the intervention. ｖ is εx with 1 as the maximum value. ｖ increases by only

[0052] The selection unit 153 solves the problem of finding k users who should intervene before information is transmitted, as shown in equation (2), in order to maximize the diffusion of positive influence (the degree of increase in users in a positive state) and minimize the diffusion of negative influence (the degree of increase in users in a negative state) under the influence diffusion model. Here, φ(q) indicates the degree of influence diffusion. For example, φ(q) is the difference between the number of users predicted by the prediction unit 152 to be in a positive state and the number of users predicted to be in a negative state, or the expected value thereof. Also, for example, φ(q) is the number of users predicted by the prediction unit 152 to be in a positive state, or the expected value thereof.

[0053]

[0054] The intervention unit 154 intervenes in the selected user. Here, intervention refers to providing information, education, rewards, etc. to the user, encouraging a behavioral change that is desirable for the account of the information source in relation to the user's information sharing behavior.

[0055] The intervention of the intervention unit 154 is expected to increase the degree of agreement of the selected users with the sender. The intervention unit 154 can provide the users selected by the selection unit 153 with information to encourage a positive reaction to the account's post. For example, the intervention unit 154 sends the users selected by the selection unit 153 a message asserting the legitimacy of the information posted by the sender, using a message function of the SNS or the like. For example, the intervention unit 154 provides information on literature that serves as evidence that the post is true. Furthermore, for example, the intervention unit 154 provides information such as the risk of litigation if slander is made against the post.

[0056] 4 is a diagram illustrating the diffusion of influence. The syringe mark represents a user who has been selected by the selection unit 153 and is the target of intervention. That is, in the example of FIG. 4, the selection unit 153 has selected users 202 and 205.

[0057] For example, after the degree of agreement between user 202 and user 205 with user 201 increases due to an intervention, if the prediction unit 152 predicts the state of each user with respect to user 201, the result will be as shown in Fig. 4. Unlike Fig. 3, Fig. 4 shows that the states of not only user 202 and user 205, who are the targets of the intervention, but also user 204, user 206, and user 207 are positive.

[0058] The maximization problem of equation (2) is NP-hard, and the objective function is also #P-hard. Therefore, the selection unit 153 can solve the maximization problem of equation (2) using an approximation technique called the MIA model in graph mining. The selection unit 153 can select an intervening user by approximately solving the maximization problem of equation (2) using the MIOA-greedy for Interv-IM algorithm or the MIOA-top-k for Interv-IM algorithm.

[0059] Fig. 5 shows the algorithm of MIOA-greedy for Interv-IM. As shown in the third line of Fig. 5, the selection unit 153 selects a user whose influence spread rate increases the most when an intervention is performed (excluding users who have already been selected). The selection unit 153 repeats this process k times to obtain k candidate users for intervention.

[0060] Figure 6 shows the algorithm of MIOA-top-k for Interv-IM. As shown in the second row of Figure 6, a combination of k users that will have the greatest increase in the degree of influence when an intervention is performed is identified. Through this process, the selection unit 153 can obtain k users who are candidates for intervention.

[0061] [Processing of the First Embodiment] Fig. 7 is a flowchart showing the processing flow of the intervention device according to the first embodiment. As shown in Fig. 7, first, the intervention device 10 calculates the degree of agreement of each user with respect to a call from a specific account (step S101). Next, the intervention device 10 predicts the extent of influence based on the social network information and the degree of agreement (step S102).

[0062] Next, the intervention device 10 selects a user for which the influence is to be varied (step S103), and then intervenes with the selected user (step S104).

[0063] [Effects of the First Embodiment] As described above, the calculation unit 151 calculates the degree of agreement of a user with an account based on the content of the user's reaction to a post from the account on the SNS. The prediction unit 152 predicts the extent of the influence of information disseminated by the account based on information indicating the relationship between users on the SNS and the degree of agreement. The selection unit 153 selects a user to be an intervention target from among users of the SNS based on the extent of the influence.

[0064] The intervention device 10 according to the first embodiment can efficiently select a user as an intervention target from among all users of the SNS, thereby preventing adverse effects caused by negative posts to a specific account on the SNS.

[0065] In particular, it will prevent posts by anti-users that disrupt business on corporate accounts (negative campaigns), the spread of negative effects caused by such posts, and the occurrence of so-called flame wars caused by misinterpreting and quoting or spreading news articles posted by news media.

[0066] Furthermore, as shown in Fig. 8, the first embodiment exhibits higher performance than the prior art in improving the extent of influence, i.e., in increasing the number of positive users (or decreasing the number of negative users). Fig. 8 is a diagram for explaining the effect of the first embodiment.

[0067] The horizontal axis of Fig. 8 represents the number of intervening users (nodes), i.e., the value of k. The vertical axis of Fig. 8 represents the extent of influence. MIOA-greedy and MIOA-top-k for Interv-IM are algorithms of the first embodiment. Distance, degree, and random are prior art.

[0068] Furthermore, the intervention unit 154 provides information for encouraging a positive reaction to the post of the account to the user selected by the selection unit 153. This makes it possible to automatically perform intervention with the selected user.

[0069] The prediction unit 152 predicts whether the state of the second user is the first state, the second state, or the third state, based on information indicating whether the state of the first user is a first state in which the first user is positive about the account, a second state in which the first user is negative about the account, or a third state in which the first user is neither positive nor negative about the account, and the degree of agreement of a second user who has a specific relationship with the first user on the SNS. In this way, the intervention device 10 can efficiently select users by simulating the degree of influence spread using the influence diffusion model.

[0070] [System Configuration, etc.] The components of each device shown in the figure are conceptual functional units and do not necessarily have to be physically configured as shown. In other words, the specific form of distribution and integration of each device is not limited to that shown in the figure, and all or part of the devices can be functionally or physically distributed or integrated in any unit depending on various loads, usage conditions, etc. Furthermore, all or any part of the processing functions performed by each device can be realized by a CPU (Central Processing Unit) and a program analyzed and executed by the CPU, or can be realized as hardware using wired logic. The program may be executed not only by the CPU but also by other processors such as a GPU.

[0071] Furthermore, among the processes described in this embodiment, all or part of the processes described as being performed automatically can be performed manually, or all or part of the processes described as being performed manually can be performed automatically using a known method.In addition, the information including the processing procedures, control procedures, specific names, various data and parameters shown in the above documents and drawings can be changed as desired unless otherwise specified.

[0072] [Program] In one embodiment, the intervention device 10 can be implemented by installing an intervention program that executes the above-described processes as package software or online software on a desired computer. For example, by executing the above-described intervention program on an information processing device, the information processing device can function as the intervention device 10. The information processing device referred to here includes desktop and notebook personal computers. In addition, the information processing device also includes mobile communication terminals such as smartphones, mobile phones, and PHS (Personal Handyphone System), as well as slate terminals such as PDAs (Personal Digital Assistants).

[0073] The intervention device 10 may also be implemented as a server device that provides services related to the above-described processes to a terminal device used by a user as a client. For example, the server device may be implemented as a server device that receives historical information as input and outputs candidates for users who are candidates for intervention. In this case, the server device may be implemented as a web server or as a cloud that provides services related to the above-described processes through outsourcing.

[0074] 9 is a diagram showing an example of a computer that executes an intervention program. The computer 1000 includes, for example, a memory 1010 and a CPU 1020. The computer 1000 also includes a hard disk drive interface 1030, a disk drive interface 1040, a serial port interface 1050, a video adapter 1060, and a network interface 1070. These components are connected by a bus 1080.

[0075] The memory 1010 includes a read-only memory (ROM) 1011 and a random access memory (RAM) 1012. The ROM 1011 stores a boot program such as a basic input / output system (BIOS). The hard disk drive interface 1030 is connected to a hard disk drive 1090. The disk drive interface 1040 is connected to a disk drive 1100. A removable storage medium such as a magnetic disk or optical disk is inserted into the disk drive 1100. The serial port interface 1050 is connected to a mouse 1110 and a keyboard 1120, for example. The video adapter 1060 is connected to a display 1130, for example.

[0076] The hard disk drive 1090 stores, for example, an OS 1091, an application program 1092, a program module 1093, and program data 1094. That is, the program that defines each process of the intervention device 10 is implemented as a program module 1093 in which computer-executable code is written. The program module 1093 is stored, for example, in the hard disk drive 1090. For example, the program module 1093 for executing processes similar to those of the functional configuration of the intervention device 10 is stored in the hard disk drive 1090. Note that the hard disk drive 1090 may be replaced with an SSD (Solid State Drive).

[0077] Furthermore, setting data used in the processing of the above-described embodiment is stored as program data 1094, for example, in the memory 1010 or the hard disk drive 1090. Then, the CPU 1020 reads the program module 1093 or the program data 1094 stored in the memory 1010 or the hard disk drive 1090 into the RAM 1012 as necessary, and executes the processing of the above-described embodiment.

[0078] The program module 1093 and program data 1094 may not necessarily be stored in the hard disk drive 1090, but may also be stored in a removable storage medium and read by the CPU 1020 via the disk drive 1100 or the like. Alternatively, the program module 1093 and program data 1094 may be stored in another computer connected via a network (such as a local area network (LAN) or a wide area network (WAN)). The program module 1093 and program data 1094 may then be read by the CPU 1020 from the other computer via the network interface 1070.

[0079] REFERENCE SIGNS LIST 10 intervention device 11 communication unit 12 input unit 13 output unit 14 storage unit 15 control unit 141 reaction history 151 calculation unit 152 prediction unit 153 selection unit 154 intervention unit

Claims

1. An intervention device comprising: a calculation unit that calculates a user's degree of agreement with an account on a SNS based on the content of the user's reaction to a post made by the account; a prediction unit that predicts the extent of the influence of information disseminated by the account on the SNS based on information indicating the relationship between users and the degree of agreement; and a selection unit that selects a user to be intervened from among users of the SNS based on the extent of the influence.

2. The intervention device according to claim 1, further comprising an intervention unit that provides information to the users selected by the selection unit to encourage a positive reaction to the posts of the account.

3. The intervention device described in claim 1, characterized in that the prediction unit predicts whether the state of the second user is the first state, the second state, or the third state based on information indicating whether the state of the first user is a first state that is positive toward the account, a second state that is negative toward the account, or a third state that is neither positive nor negative toward the account, and the degree of agreement of a second user who has a specific relationship with the first user in the SNS.

Images