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

Abstract

An information processing device according to one aspect of the present invention comprises an estimation unit and an optimization unit. The estimation unit estimates a parameter value included in a prediction model that calculates a predicted value of social welfare, the prediction model having been modeled on the basis of data including past user behavior, the status of the user, and an evaluation value with regard to the past user behavior, a time change function representing a change in user interests over time, an influence vector representing the degree of influence of the status of the user on the social welfare, and behavior choices of the user. The optimization unit calculates a predicted value of the social welfare from the prediction model into which the parameter value has been substituted, and extracts a behavior choice at which the predicted value is maximized.

Description

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

[0001] The present invention relates to an information processing device, an information processing method, and an information processing program.

[0002] Humans are often faced with a large number of behavioral choices. These choices include the next product to purchase, the next movie to watch, or the exercise to do for health. With limited financial resources and time, it can be difficult to independently identify choices that align with one's interests. Therefore, technology has been developed that can recommend optimal behavioral choices that align with human interests by estimating the similarity between the direction of human interests and behavioral choices.

[0003] Wei Lu, Stratis Ioannidis, Smriti Bhagat, Laks VS Lakshmanan, “Optimal Recommendations under Attraction, Aversion, and Social Influence”, KDD'14: Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data mining, pp.811-820 (2014)

[0004] Because human interests change over time, technology has been developed that uses past history to estimate social welfare, which is the degree of similarity between a person's interests and their behavioral options, and then recommends the best behavioral options for that person (see, for example, Non-Patent Document 1). However, people do not decide their actions based solely on their own interests; their behavior also changes depending on the natural and social environments surrounding them, such as the day of the week, weather, current location, and whether or not they have work. However, no technology has been developed that can estimate social welfare taking into account a person's own situation.

[0005] The present invention has been made in light of the above circumstances, and its purpose is to provide an information processing device that can estimate social welfare taking into account human interests and situations.

[0006] An information processing device according to one aspect of the present invention predicts social welfare, which represents the degree of similarity between a user's interest direction and behavioral options, and includes an estimation unit and an optimization unit. The estimation unit estimates parameter values ​​included in a prediction model that calculates a predicted value of social welfare, the prediction model being modeled based on data including past user behavior, the user's situation, and evaluation values ​​for the past user behavior, a time change function that represents changes in the user's interests over time, an influence vector that represents the degree of influence the user's situation has on the social welfare, and the user's behavioral options. The optimization unit calculates the predicted value of social welfare from the prediction model into which the parameter values ​​have been substituted, and extracts the behavioral option that maximizes the predicted value.

[0007] According to this invention, it is possible to provide an information processing device that can estimate social welfare taking into account human interests and situations, i.e., the user's interests and situation. By taking into account not only the user's interests but also the situation surrounding the user, it becomes possible to recommend optimal behavioral options that are in line with the user's interests and situation.

[0008] Fig. 1 is a block diagram showing the functional configuration of an information processing device according to an embodiment; Fig. 2 is a diagram showing the functional configuration of an estimation unit in the information processing device according to an embodiment; Fig. 3 is a diagram showing the functional configuration of an optimization unit in the information processing device according to an embodiment; Fig. 4 is a flowchart showing processing executed by the information processing device according to an embodiment; and Fig. 5 is a block diagram showing the hardware configuration of the information processing device according to an embodiment.

[0009] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In the following description, components having the same function and configuration will be given the same reference numerals. In this specification, social welfare represents the similarity between the direction of a person's (user's) interests and their behavioral options, in other words, the degree to which a behavior satisfies the user's desires, or the degree to which a behavior satisfies the user's desires.

[0010] 1. Functional Configuration of the Embodiment The functional configuration of an information processing device 1 according to the embodiment will be described with reference to Fig. 1. Fig. 1 is a block diagram showing the functional configuration of the information processing device 1.

[0011] The information processing device 1 includes a control unit 10 , a data storage unit 20 , a program storage unit 30 , an input / output unit 40 , and a transmission / reception unit 50 .

[0012] The control unit 10 includes, as processing functions for implementing the embodiment, an estimation unit 11 and an optimization unit 12. The control unit 10 comprehensively controls the data storage unit 20, the program storage unit 30, the input / output unit 40, and the transmission / reception unit 50.

[0013] The estimation unit 11 estimates parameter values ​​included in a model or function for predicting a user's social welfare (hereinafter referred to as a prediction model or prediction function). The prediction model is modeled or generated based on experimental data 111, a time change function 112 of the user's interests, an influence vector 113 of the user's situation, and behavioral options 114, all of which are shown in FIG. 2 and will be described later in detail. The prediction model, the experimental data 111, the time change function 112 of the user's interests, the influence vector 113 of the user's situation, and the behavioral options 114 will be described later in detail.

[0014] The optimization unit 12 calculates a predicted value of the user's social welfare from the prediction model into which the parameter values ​​estimated by the estimation unit 11 are substituted. Furthermore, the optimization unit 12 extracts, from the calculated predicted values, a behavioral option that maximizes the predicted value.

[0015] The data storage unit 20 stores data or information necessary for implementing the embodiment and data or information generated in the course of executing various processes. That is, the data storage unit 20 stores data or information related to experimental data 111, a time-varying function of user interest 112, an influence vector of the user's situation 113, and behavioral options 114.

[0016] The program storage unit 30 stores programs necessary for executing various controls and processes according to the embodiment. That is, the program storage unit 30 stores information processing programs, control programs, and other application programs for executing processes in the information processing device 1 according to the embodiment.

[0017] The input / output unit 40 transmits data or information input by the user to the control unit 10. The input / output unit 40 also outputs data or information received from the control unit 10 to the user. For example, the input / output unit 40 outputs the action options extracted by the optimization unit 12.

[0018] The transmitting / receiving unit 50 can transmit and receive data, information, and an information processing program according to the embodiment to and from a server device included in the communication network NW. The transmitting / receiving unit 50 may receive, for example, data or information related to experimental data 111, a time-varying function 112 of a user's interest, an influence vector 113 of a user's situation, and behavioral options 114 from a server device included in the communication network NW. The transmitting / receiving unit 50 may also transmit behavioral options extracted by the optimization unit 12 to a server device included in the communication network NW.

[0019] Next, a description will be given of the functional configuration of the estimation unit 11 in the information processing device 1. FIG.

[0020] The estimation unit 11 includes a parameter estimation unit 110. The parameter estimation unit 110 receives as input experimental data 111, a time-varying function of the user's interest 112, an influence vector of the user's situation 113, and data or information regarding behavioral options 114.

[0021] The parameter estimation unit 110 estimates parameter values ​​included in a social welfare prediction model modeled based on experimental data 111, a time-varying function of user interests 112, an influence vector of the user's situation 113, and behavioral options 114.

[0022] The experimental data 111 includes past user actions, past user situations, and evaluation values ​​for the past user actions. Specifically, the experimental data 111 includes a user ID (denoted as i, where i=1, 2, ..., n), an action ID (denoted as j, where j=1, 2, ..., m), a vector representing the user situation, an evaluation value for the action (denoted as r), and a time (denoted as t) when the action was evaluated and the evaluation value was assigned.

[0023] A past user's behavior is an action that the user performed in the past, such as a product that the user purchased, a movie that the user watched, or an exercise that the user performed. A past user's situation is the user's situation when the past user's behavior was performed. A user's situation includes the natural and social environment surrounding the user, such as the day of the week, the weather, the current location, and whether or not the user has work.

[0024] The vector representing the user's situation is C ｉ It is denoted as (t) and is a k-dimensional vector (k is an integer equal to or greater than 1). In the k-dimensional vector, the component corresponding to the user's situation is set to 1, and the other components are set to 0.

[0025] The time change function 112 of the user's interest is a function that represents the time change of the interest of the user i at the time (time step) t. This time change function is expressed as u ｉ Let (t). ｉ (t) is a real vector of any d-dimensional (d is an integer greater than or equal to 1) and has any number of parameters θ ｉ Parameter θ ｉ The value of is estimated by the parameter estimation unit 110 of the estimation unit 11.

[0026] The influence vector 113 of the user situation represents the degree of influence that the situation of user i has on social welfare. The influence vector representing the influence of the situation of user i is defined as Bｉ ・C ｉ (t), where matrix B ｉ is the response coefficient for the situation of user i and has dimensions k × d. ｉ is a quantity (parameter value) estimated by the parameter estimation unit 110 of the estimation unit 11. ｉ The components of may vary from situation to situation or from user to user, and in a particular situation, ｉ The component may not depend on the user. ｉ As described above, (t) is a vector representing the user's situation and is included in the experimental data 111 .

[0027] The behavioral options 114 are behavioral options for the user i. The behavioral options recommended for the user i are denoted by v ｊ The action options are: ｊ is an arbitrary d-dimensional real vector.

[0028] The processing executed by the parameter estimation unit 110 will be described below.

[0029] The prediction model of user i's social welfare is F ｉ Let's say.

[0030] Then, F ｉ Is F ｉ =<u ｉ (t)+B ｉ ・C ｉ (t),ν ｊ > and can be expressed in the form of an inner product. ｉ As mentioned above, is a model for predicting the social welfare of user i, that is, a model or function for calculating the predicted value of social welfare of user i. The advantage of formulating social welfare in this way is that it is possible to obtain a predicted value of social welfare that takes into account changes in a person's interests as well as the circumstances surrounding that person, and to recommend options for action based on the predicted value.

[0031] A function u that represents the change in the interest of user i over time ｉ Parameter θ included in (t) ｉ, and the response coefficient B for the situation of user i ｉ In order to estimate, first, the parameter u ｉ ,ν ｊ ,λ,μ,are estimated using cross-validation.

[0032]

[0033] Here, n is the number of users, m is the number of behavioral options, and E ⊂ [n] × [m] holds. ｉｊ represents the dataset.

[0034] Next, the estimated parameters u ｉ ,ν ｊ ,λ,μ,u ｉ ０ =u ｉ The time-varying function u that minimizes the following equation (2) is ｉ Estimate (t).

[0035]

[0036] Here, T represents the prediction period. The parameter θ that minimizes the error defined by the following equation (3) is ｉ ,B ｉ ,κ is estimated by cross-validation and gradient descent.

[0037]

[0038] Then, the estimated parameters θ ｉ ,B ｉ , κ are output to the optimization unit 12 .

[0039] Next, we will explain the functional configuration of the optimization unit 12 in the information processing device 1. Fig. 3 is a diagram showing the functional configuration of the optimization unit 12 in the information processing device 1. The optimization unit 12 extracts, for each user, behavioral options that maximize social welfare on average during the prediction target period T.

[0040] The optimization unit 12 includes a social welfare prediction unit 120. The social welfare prediction unit 120 receives, excluding the experimental data 111, data or information related to a time-varying function 112 of the user's interest, an influence vector 113 of the user's situation, and behavioral options 114, as well as parameter values ​​estimated by the parameter estimation unit 110 (i.e., parameters θ ｉ ,B ｉ , κ value) 115 is input.

[0041] The social welfare prediction unit 120 predicts the social welfare of the user in the prediction period T. That is, it calculates the predicted value of the user's social welfare in the prediction period T. Next, the social welfare prediction unit 120 extracts the behavioral option that maximizes the calculated predicted value of social welfare on average.

[0042] The processing executed by the social welfare prediction unit 120 will be described in detail below.

[0043] The social welfare prediction unit 120 calculates a predicted value of social welfare using a prediction model generated based on a time change function 112 of the user's interests, an influence vector 113 of the user's situation, behavioral options 114, and parameter values ​​115. That is, the social welfare prediction unit 120 calculates a predicted value of social welfare using a time change function u of the user i's interests in the prediction target period T. ｉ (t) and the influence vector B of the user's situation ｉ ・C ｉ (t) and the action options ν ｊ By calculating the time-varying function u ｉ (t) and the influence vector B ｉ ・C ｉ (t) and the action options ν ｊ The social welfare prediction unit 120 calculates the predicted value of the social welfare of the user i by calculating all combinations of ν and ν. Then, the social welfare prediction unit 120 calculates the behavioral option v that maximizes the predicted value of the social welfare of the user i on average during the prediction target period T. ｊ Extract.

[0044] Thereafter, the behavioral options v extracted by the social welfare prediction unit 120 are ｊ will be output.

[0045] 2. Operation of the embodiment The operation of the information processing device 1 according to the embodiment will be described with reference to Fig. 4. Fig. 4 is a flowchart showing processing executed by the information processing device 1. The processing of the flowchart shown in Fig. 4 is controlled by a control unit (e.g., processor) 10 provided in the information processing device 1, which includes an estimation unit 11 and an optimization unit 12.

[0046] First, the estimation unit 11 acquires, for example, experimental data 111 and a time-varying function u of interest of a user i from the data storage unit 20. ｉ (t), the influence vector B of user i’s situation ｉ ・C ｉ (t), and the action options ν ｊ (S1). Here, the experimental data, the time-varying function of the user's interest, the influence vector of the user's situation, and the behavioral options are acquired from the data storage unit 20, but this is not limitative. They may be acquired from a server device included in the communication network NW via the transmitting / receiving unit 50, or may be input via the input / output unit 40.

[0047] Next, the estimation unit 11 calculates the time-varying function u of the interest of the user i based on the acquired experimental data 111. ｉ (t), the influence vector B of user i’s situation ｉ ・C ｉ (t), and the action options ν ｊ In a prediction model that calculates a predicted value of social welfare of user i based on the above, parameter values ​​included in the prediction model are estimated (S2). The parameter values ​​are a time-varying function u ｉ Parameter θ included in (t) ｉ and the influence vector B of the situation of user i ｉ ・C ｉ Response coefficient B for the situation of user i included in (t) ｉ Includes:

[0048] Next, the optimization unit 12 calculates the predicted value of social welfare of user i using the parameter values ​​estimated in step S2 and the prediction model for calculating the predicted value of social welfare, i.e., using the prediction model into which the parameter values ​​have been substituted (S3). In more detail, the optimization unit 12 calculates the predicted value of social welfare of user i using the time-varying function u ｉ (t), the influence vector B of user i’s situation ｉ ・C ｉ (t), the choice of action ν ｊ , and a prediction model F based on the estimated parameter values ｉ =<u ｉ (t)+B ｉ ・C ｉ (t),ν ｊ > Calculate the predicted value of social welfare of user i in the prediction target period T.

[0049] Furthermore, the optimization unit 12 selects an action option v that maximizes the predicted social welfare value on average from the calculated predicted social welfare values. ｊ The optimization unit 12 extracts the extracted action options v ｊ is transmitted to the input / output unit 40.

[0050] Thereafter, the input / output unit 40 receives the action options v ｊ is output (S5).

[0051] 3. Hardware Configuration of the Embodiment Next, an example of the hardware configuration of the information processing device 1 of the embodiment will be described with reference to Fig. 5. Here, an example in which the information processing device 1 is configured by a computer will be described.

[0052] 5 is a block diagram showing the hardware configuration of the information processing device 1. The information processing device 1 (e.g., a computer) has a processor 61, a read-only memory (ROM) 62, a random access memory (RAM) 63, a storage device 64, an input / output interface 65, and a communication interface 66.

[0053] 1 , the control unit 10 including the estimation unit 11 and the optimization unit 12 corresponds to, for example, a processor 61 and a RAM 63. The data storage unit 20 corresponds to, for example, a ROM 62, a RAM 63, and a storage device 64. The program storage unit 30 corresponds to, for example, the ROM 62 and a storage device 64. The input / output unit 40 corresponds to, for example, an input / output interface 65 (and an input / output device 73). The transmission / reception unit 50 corresponds to, for example, a communication interface 66.

[0054] The processor 61, ROM 62, RAM 63, storage device 64, input / output interface 65, and communication interface 66 are electrically connected to one another via a bus 67, and are capable of exchanging data and signals with one another.

[0055] The processor 61 is configured by a general-purpose hardware processor including, for example, a CPU (Central Processing Unit), a GPU (Graphical Processing Unit), etc. The processor 61 controls the ROM 62, the RAM 63, the storage device 64, the input / output interface 65, and the communication interface 66 as a whole.

[0056] The ROM 62 is a read-only nonvolatile memory. The ROM 62 non-temporarily stores a startup program required when starting up the information processing device 1 of the embodiment. The information processing device 1 starts up when the processor 61 executes the program in the ROM 62. The ROM 62 is, for example, configured from an EPROM (Erasable Programmable Read Only Memory), and stores various startup settings in addition to the startup program.

[0057] The RAM 63 is a volatile memory that can be written to and read from. The RAM 63 temporarily stores programs required for processing by the processor 61 and data required for executing the programs. The processor 61, for example, executes the programs in the RAM 63 to perform calculations on the data in the RAM 63 and store the calculation results in the RAM 63.

[0058] The storage device 64 is configured with a non-volatile memory such as a hard disk drive (HDD) or a solid state drive (SSD). The storage device 64 non-temporarily stores the programs executed by the processor 61 and data required for executing the programs. The processor 61 loads the programs and data in the storage device 64 into the RAM 63 and executes the programs to perform various functions.

[0059] The input / output interface 65 is connected to an input device 71 and an output device 72, and enables input of information from the input device 71 and output of information to the output device 72. The input device 71 includes, for example, a keyboard, a mouse, a touch panel, and a disk drive. The input device 71 is not limited to these, but may include any other input device. The output device 72 includes, for example, a display and a disk drive. The output device 72 is not limited to these, but may include any other output device. The input device 71 and the output device 72 may be configured as an input / output device 73 that has the functions of both the input device 71 and the output device 72.

[0060] The communication interface 66 is connected to the communication network NW and enables reception of information from the server device and transmission of information to the server device.

[0061] The input / output interface 65 and the communication interface 66 are, for example, wired or wireless interfaces. The wired interface includes a port to which a device is connected. The wireless interface includes Bluetooth (registered trademark), Wi-Fi (registered trademark), etc.

[0062] An information processing program for causing the information processing device 1 according to the embodiment to function is provided to a computer via, for example, a computer-readable recording medium 74. This recording medium 74 is called a non-transitory computer-readable recording medium. Non-transitory computer-readable recording media include, for example, disks such as flexible disks, optical disks (CD-ROM, CD-R, DVD-ROM, DVD-R, etc.), magneto-optical disks (MO, etc.), semiconductor memories, USB memories, etc.

[0063] In addition, the information processing program may be stored in a server device on the communication network NW, downloaded from the server device, and non-temporarily stored in the storage device 64.

[0064] For example, in response to an input instruction to start the information processing device 1, the processor 61 reads an information processing program from the storage device 64 into a program area of ​​the RAM 63, and also reads data necessary for executing the information processing program from the storage device 64 into a data area of ​​the RAM 63. The processor 61 calculates data in the data area in accordance with the information processing program and writes the calculation results into the data area. Through these operations, the processor 61, RAM 63, storage device 64, input / output interface 65, and communication interface 66 work together to execute at least some of the functions of the components of the information processing device 1, i.e., the estimation unit 11 and the optimization unit 12.

[0065] 4. Effects of the Embodiments, etc. According to the embodiments of the present invention, it is possible to provide an information processing device that can estimate social welfare taking into account human interests and situations, that is, the user's interests and situation.

[0066] In the configuration of the embodiment, a predicted value of a user's social welfare is calculated using a prediction model (or a prediction function) based on experimental data 111, a time change function of the user's interests 112, an influence vector of the user's situation 113, and behavioral options 114. The influence vector of the user's situation 113 represents the degree of influence that the user's situation has on social welfare.

[0067] In this way, by using the influence vector 113 of the user's situation as a variable in the prediction model for calculating social welfare, it is possible to calculate a predicted value of social welfare that takes into account the user's situation, i.e., the situation surrounding the user. Furthermore, from the calculated predicted value of social welfare, options of behavior that maximize the predicted value of social welfare, i.e., multiple behavioral targets, are extracted. The extracted options of behavior are then output.

[0068] As described above, the information processing device 1 according to the embodiment can predict the social welfare of a user (i.e., the degree of satisfaction of needs) taking into consideration not only the user's interests but also the circumstances surrounding the user. This makes it possible to recommend to the user the most suitable behavioral options in line with the user's interests and circumstances.

[0069] The functional blocks described in the above embodiments can be realized as either hardware or computer software, or a combination of both. It is not necessary for the functional blocks to be distinguished as in the above examples. For example, some functions may be performed by functional blocks other than the illustrated functional blocks. Furthermore, the illustrated functional blocks may be further divided into smaller functional sub-blocks. Furthermore, the order of processing in the flowcharts described in the above embodiments can be changed as much as possible.

[0070] It should be noted that the present invention is not limited to the above-described embodiment, and can be practiced in various modified forms without departing from the spirit and scope of the present invention.

[0071] 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.

[0072] DESCRIPTION OF SYMBOLS 1... Information processing device 10... Control unit 11... Estimation unit 12... Optimization unit 20... Data storage unit 30... Program storage unit 40... Input / output unit 50... Transmitting / receiving unit 61... Processor 62... ROM 63... RAM 64... Storage device 65... Input / output interface 66... ​​Communication interface 67... Bus 71... Input device 72... Output device 73... Input / output device 74... Recording medium 110... Parameter estimation unit 111... Experimental data 112... Time change function of user's interest 113... Influence vector of user's situation 114... Action options 115... Parameter value 120... Social welfare prediction unit 121... Action options that maximize social welfare

Claims

1. An information processing device that predicts social welfare, which represents the similarity between the direction of a user's interest and behavioral options, comprising: an estimation unit that estimates parameter values ​​included in a prediction model that calculates a predicted value of social welfare, the prediction model being modeled based on data including past user behavior, the user's situation, and an evaluation value for the past user behavior, a time change function that represents changes in the user's interest over time, an influence vector that represents the degree of influence that the user's situation has on the social welfare, and the user's behavioral options; and an optimization unit that calculates the predicted value of social welfare from the prediction model into which the parameter values ​​have been substituted, and extracts the behavioral option that maximizes the predicted value.

2. The information processing device according to claim 1, wherein the parameter values ​​estimated by the estimation unit include a parameter value included in the time-varying function and a response coefficient for the user's situation included in the influence vector.

3. An information processing method for predicting social welfare, which represents the similarity between the direction of a user's interest and behavioral options, comprising: estimating parameter values ​​included in a prediction model that calculates a predicted value of social welfare, the prediction model being modeled based on data including past user behavior, the user's situation, and an evaluation value for the past user behavior, a time change function that represents changes in the user's interests over time, an influence vector that represents the degree of influence that the user's situation has on the social welfare, and the user's behavioral options; and calculating the predicted value of social welfare from the prediction model into which the parameter values ​​have been substituted, and extracting the behavioral option that maximizes the predicted value.

4. An information processing program that causes a processor to execute the processing of each unit provided in the information processing device according to claim 1.

Images