Information processing method, information processing apparatus, and information processing program

Abstract

This invention provides an information processing method that enables the mathematical visualization of behavioral preferences. [Solution] An information processing method performed by an information processing device 10 equipped with an acquisition means 32, a calculation means 34, and an arrangement means 36, wherein the acquisition means 32 is a step of acquiring answers to a plurality of pairs of questions, the calculation means 34 is a step of quantifying the answers and calculating an action preference vector based on an axis vector associated with each question, and the arrangement means 36 is a step of arranging the action preference vectors in a multidimensional space, and the pairs of questions have multiple stages of choices.

Description

Technical Field

[0001] The present invention relates to, for example, an information processing method, an information processing apparatus, and an information processing program that enable mathematical visualization of action preferences.

Background Art

[0002] For example, Patent Document 1 (such as paragraph 0010) described later discloses an invention related to a personality diagnosis system 1, in which survey result data investigating the relevance to egograms and Big Five personality diagnostics at the molecular, cellular, system, and cognitive levels in the neuroscience field is input to a computer, and personality analysis results are output as data.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of Invention

Problems to be Solved by the Invention

[0004] Incidentally, Patent Document 1 (such as paragraph 0010) states that more accurate personality analysis results, which cannot be sufficiently obtained from the personality analysis results by egograms and Big Five personality diagnostics, can be obtained. However, it is desirable that various analysis methods exist not only for methods based on grasping personality, but also for the tendency of action preferences related to a subject to be evaluated.

[0005] An object of the present invention is to provide an information processing method, an information processing apparatus, and an information processing program that enable mathematical visualization of action preferences by a novel method.

Means for Solving the Problems

[0006] (A) To achieve the above object, an information processing method according to the present invention comprises: An information processing method performed by an information processing device equipped with acquisition means, calculation means, and arrangement means, The acquisition means includes the step of acquiring answers to a plurality of pairs of questions, The calculation means includes the steps of quantifying the answers and calculating an action preference vector based on the axis vectors associated with each question, The arrangement means includes the step of arranging the action preference vectors in a multidimensional space, The aforementioned question-and-answer method is an information processing method having multiple levels of choices. (B) In order to achieve the above objective, the information processing apparatus according to the present invention A means of obtaining answers to multiple pairs of questions, A calculation means that quantifies the aforementioned answers and calculates an action preference vector based on the axis vectors associated with each question, The system includes arrangement means for arranging the aforementioned behavior preference vectors in a multidimensional space, The aforementioned question is an information processing device having multiple levels of choices. (C) In order to achieve the above objective, the information processing program according to the present invention is This is an information processing program that causes a computer to execute the information processing method described in (A) above. [Effects of the Invention]

[0007] According to the above invention, it is possible to provide an information processing method, an information processing device, and an information processing program that enable the mathematical visualization of behavioral preferences by a novel method. [Brief explanation of the drawing]

[0008] [Figure 1] (a) is a flowchart illustrating an information processing method according to one embodiment of the present invention, and (b) is a block diagram illustrating the functional configuration of the information processing device. [Figure 2] (a) is a schematic diagram showing the hardware configuration of the information processing device, and (b) is a schematic diagram showing the hardware configuration of the server device. [Figure 3] This is an explanatory diagram showing an example of a 6-question diagnostic form. [Figure 4] This is a chart showing an example of a conversion table. [Figure 5] This is a chart showing "comparison category", "vector direction", and "processing meaning" for each question. [Figure 6] This is a chart showing each "step" performed for time-series integration and the content of each "step". [Figure 7] This is a flowchart showing the flow of question generation. [Figure 8] This is an explanatory diagram showing generalization of the number of vertices related to question generation. [Figure 9] This is an explanatory diagram structurally showing explanatory matters according to the embodiment. [Description of Embodiments]

[0009] <Introduction> Hereinafter, one embodiment of the present invention will be described. The description of the present embodiment is composed of the following sections. <Introduction> <Overview of Embodiment> <Overview of Hardware Configuration Related to Information Processing Apparatus 10> <6-Question Diagnosis (Multiple Target Questions)> <Description of Conversion Table> <Correspondence Between Questions and Spatial Axes (Example: 4-Category Structure)> <Spatial Structure and Grid (Point) Distribution> <<Overview of Spatial Structure>> <<Principles for Setting Weight Coefficients>> <Time-Series Integration> <Main Characteristic Processes According to the Present Embodiment> <Specific Example of Question Generation> <Generalization of Number of Vertices k> <Main Advantages of the Invention According to the Present Embodiment> <Inventions Extractable from the Embodiment> <Others> Further, FIG. 9 shows the correspondence relationship of the matters described in the present embodiment.

[0010] <Overview of Embodiments> The information processing method according to this embodiment converts selection data (data representing the selection result) into a vector space for the purpose of evaluating behavioral preferences. Conventionally, evaluation of behavioral preferences has mainly relied on experience and semantic interpretation, which has presented challenges in terms of reproducibility. The information processing method and information processing device according to this embodiment directly measure the behavioral selection itself and represent it as a reproducible structure.

[0011] As shown in Figure 1(a), the information processing method according to this embodiment performs the following steps in order: input (S(step) 1), transformation (S2), synthesis (S3), output (S4), and expansion (S5), thereby performing a mathematical visualization of behavioral preferences.

[0012] In the input (S1), the selection results for the paired questions (pair comparison) are obtained. In transformation (S2), vectorization is performed based on direction and intensity. In synthesis (S3), spatial synthesis of weighted vectors is performed. In the output (S4), the spatial position (e.g., grid) is determined. In the extension (S5), narrative structure generation is performed through time-series integration.

[0013] Furthermore, the algorithm for the information processing method has the following structure (processes A1 to A6).

[0014] The processing content of process A1 is response acquisition and / or normalization. In this process A1, the selection result is decomposed into direction and intensity. The "response acquisition" process in process A1 is performed in input (S1). The "normalization" process in process A1 is performed in transformation (S2). Furthermore, the process of decomposing the selection result into direction and intensity corresponds to "normalization". Figure 9 structurally illustrates these relationships.

[0015] The process of step A2 is the generation of a direction vector. The characteristics of this process A2 can be expressed by the following formula. vi=f(ri,di) Here, the "f" in f(ri,di) indicates that the characteristics of the process are generalized as a function. Process A2 is performed in the transformation (S2), as shown in Figure 9.

[0016] The processing content of process A3 is a decomposition of contributions to the category space (also called "action category"). The characteristics of this process A3 can be expressed by the following formula. vi=Σαi,k·ek Here, the number of vertices k has the same meaning as the number of vertices k (the full-width character k) described later. As will be described later, the number of vertices k can be an integer of 3 or more (Figure 8). Step A3 is performed in synthesis (S3), as shown in Figure 9.

[0017] The process in step A4 is weighted synthesis (weighted vector synthesis). The characteristics of this process A4 can be expressed by the following formula. Vg = Σw(g,i)·vi Here, w(g,i) is a domain-dependent function. Furthermore, w(g,i) is the weight coefficient w(g,i) of the grid (or points) described later. As shown in Figure 9, step A4 is performed after step A3 in the synthesis (S3).

[0018] The process of step A5 is position determination. The characteristics of this process A5 can be expressed by the following formula. g*=argmax||Vg|| In step A5, the maximum norm region is selected. Process A5 is performed at output (S4), as shown in Figure 9.

[0019] The processing content of process A6 is time-series integration. The characteristics of this process A6 can be expressed by the following formula. S = {g*t | t∈T} In process A6, the results obtained in processes A4 (weighted synthesis) and A5 (location determination) at different times (for example, every few years, or at any arbitrary time) are sequentially arranged and time-series-oriented to produce an output narrative structure. Step A6 is performed in extension (S5), as shown in Figure 9.

[0020] <Overview of the hardware configuration related to the information processing device 10> Figure 2(a) shows an overview of the information processing device 10 used to execute the information processing method of this embodiment. The information processing device 10 according to this embodiment is composed of a server device (management server) 12, a communication network 14, a user terminal device 16, an input device 18, and a display device 20, etc. The server device 12 and the user terminal device 16 are connected to each other so as to be able to communicate with each other via the communication network 14 which constitutes the cloud.

[0021] The database 13 is also connected to the communication network 14. The database 13 may be integrated into the server device 12. Furthermore, "user" refers to a user of the information processing device 10, but specific examples of "user" will be described later.

[0022] Server device 12 has server application software (information processing program) installed for providing information that mathematically visualizes behavioral preferences. User terminal device 16 has user application software (information processing program) installed for inputting personal information and providing information that mathematically visualizes behavioral preferences.

[0023] Furthermore, it is also possible that the server device 12 and the user terminal device 16 do not install server application software or user application software, but instead request the execution of various application software (information processing programs) from higher-level computer equipment in the cloud via the communication network 14. The server device 12 can be used as a higher-level computer equipment for the user terminal device 16.

[0024] Furthermore, the server device 12 may be composed of server equipment distributed across multiple locations. General-purpose computer equipment with typical hardware configurations can be used as server equipment. For example, an administrator terminal device (not shown) may be connected to the server device 12 via the communication network 14, or directly without using the communication network 14.

[0025] As user terminal devices 16 (and administrator terminal devices), general personal computers (hereinafter referred to as "PCs"), notebook PCs, smartphones, or tablet devices can be used. In the example in Figure 2(a), a desktop PC is shown as the user terminal device 16. Although only one user terminal device 16 is shown in the example in Figure 2(a), there may be multiple user terminal devices 16.

[0026] The input device 18 is, for example, an operation input means such as a keyboard, mouse, or touch panel. The input device 18 may be integrated with the user terminal device 16. The input device 18 may encompass multiple input devices such as a keyboard, mouse, and touch panel.

[0027] The display device 20 is composed of display means such as a general display device. The display device 20 may be integrated with the user terminal device 16. Furthermore, the display device 20 may also have an integrated touch panel.

[0028] Examples of communication networks 14 include the Internet, LAN, WAN, public telephone lines, base stations, mobile communication networks, and those interconnected via gateways, etc. (including so-called clouds).

[0029] Figure 2(b) shows a schematic representation of the hardware configuration of the server device 12. While Figure 2(b) shows the server device (management server) 12, a PC or similar device with a similar hardware configuration can also be used for the user terminal device 16. The hardware configuration of the server device 12 will be described below, and the hardware configuration of the user terminal device 16 will not be explained.

[0030] The server device 12 shown in Figure 2(b) includes a control unit 21, a storage unit 22, a communication unit 23, etc., internally. The storage unit 22 and the communication unit 23 may be externally mounted. The server device 12 may also include peripheral devices such as an input device (input device 18) and a display device (display device 20), similar to the user terminal device 16.

[0031] The control unit 21, although not shown in the diagram, is composed of a CPU (Central Processing Unit), ROM (Read Only Memory), RAM (Random Access Memory), etc. The CPU of the control unit 21 loads various computer programs stored in the ROM and memory unit 22 onto the RAM and executes them. The control unit 21 may also be any processing circuit or arithmetic circuit equipped with multiple CPUs, a multi-core CPU, a GPU (Graphics Processing Unit), a microcontroller, volatile or non-volatile memory, etc.

[0032] The storage unit 22 is a non-volatile storage unit that includes semiconductor memory such as ROM or RAM, an HDD (Hard Disk Drive), or an SSD (Solid State Drive) for storing various types of information. The storage unit 22 stores operating system programs, driver programs, application programs, and data used for processing in the processor (in this case, the control unit 21).

[0033] The program stored in the storage unit 22 may be provided on a non-temporary recording medium (not shown) on which the program is recorded in a readable format. Examples of recording media include portable memory such as CD-ROM, USB (Universal Serial Bus) memory, SD (Secure Digital) card, microSD card, and CompactFlash (registered trademark). In this case, the control unit 21 reads the program from the recording medium using a reading device (not shown) and installs the read program into the storage unit 22.

[0034] The program stored in the memory unit 22 may be provided via communication through the communication unit 23. In this case, the control unit 21 obtains the program through the communication unit 23 and installs the obtained program into the memory unit 22.

[0035] The communication unit 23 is equipped with an interface circuit for communicating with other terminal devices via the communication network 14. The communication unit 23 performs wired or wireless data communication in accordance with a predetermined communication protocol. When the communication unit 23 receives information to be sent to another terminal device from the control unit 21, it transmits the input information to the other terminal device. The communication unit 23 outputs information received from other terminal devices via the communication network 14 to the control unit 21.

[0036] The server device 12 shown in Figure 2(b) communicates with the user terminal device 16 and performs authentication. When the user enters their account information (such as ID and password) on the user terminal device 16 and authentication for login (including verification of billing information, etc.) is approved, they are permitted to use the behavior preference evaluation support site (website, My Page). Login authentication may also be performed on the user terminal device 16.

[0037] The behavior preference evaluation support site can also be called, for example, "behavior preference evaluation site." The behavior preference evaluation support site consists of multiple web pages displaying content for the user. By connecting to the URL (Uniform Resource Locator) corresponding to each site page on the user terminal device 16, the corresponding page is displayed, and the displayed page can be used.

[0038] In this embodiment, the information processing device 10 can also be considered as an information processing system that mathematically visualizes behavioral preferences. Furthermore, the various processes (steps) for mathematical visualization performed using the information processing device 10 constitute the information processing method in this embodiment. Moreover, the computer programs that cause the server device 12 and the user terminal device 16 to execute processing for mathematical visualization are both information processing programs according to this embodiment.

[0039] Furthermore, if the server device 12 (or user terminal device 16) performs information processing for mathematical visualization on its own, it is also possible to consider only the server device 12 (or user terminal device 16) as the information processing device according to this embodiment.

[0040] Figure 1(b) schematically shows the functional configuration of the information processing device 10. The information processing device 10 includes components such as acquisition means 32, calculation means 34, arrangement means 36, generation means 38, estimation means 40, selection means 42, and display means 44. The acquisition means 32, calculation means 34, arrangement means 36, generation means 38, estimation means 40, and selection means 42 are configured, for example, using the control unit 21, storage unit 22, communication unit 23, etc. of the server device 12. The display means 44 is configured using the control unit 21, storage unit 22, communication unit 23, etc. of the server device 12, and the user terminal device 16 (including the display device 20).

[0041] The acquisition means 32 acquires the user's response. The acquisition means 32 acquires the responses (responses to the six radio buttons) to a plurality of paired questions (six questions in the six-question diagnosis) consisting of multiple levels of choices (in this embodiment, a pair of response indicators in the six-question diagnosis questions described later). In this embodiment, the processing by the acquisition means 32 corresponds to acquiring the response (step A1) from the input (S1, Figure 9). The acquisition means 32 is configured, for example, using the control unit 21, storage unit 22, communication unit 23, etc. of the server device 12.

[0042] The calculation means 34 quantifies the answers obtained by the acquisition means 32 and calculates an action preference vector based on the axis vector associated with each question. The action preference vector is expressed by the same formula as formula 4 (described later) related to question design. In this embodiment, the processing of the calculation means 34 corresponds to step A2 (direction vector generation) of the transformation (S2). The calculation means 34 is configured, for example, using the control unit 21, storage unit 22, communication unit 23, etc. of the server device 12.

[0043] The arrangement means 36 arranges the action preference vectors calculated by the calculation means 34 in a multidimensional space. In this embodiment, the processing by the arrangement means 36 corresponds to step A5 (position determination) of the output (S4, Figure 9). The arrangement means 36 is configured, for example, using the control unit 21, storage unit 22, communication unit 23, etc. of the server device 12.

[0044] The generation means 38, estimation means 40, and selection means 42 are used for generating questions (question generation, question design). Details of question generation (question design) will be described later based on Figure 7, but among these, the generation means 38 extracts question axes using categorical principal component analysis (CATPCA). In this embodiment, the processing of the generation means 38 corresponds to the CATPCA (S12) processing shown in Figure 7, which is part of the "question design" in Figures 9 and 7. The generation means 38 is configured, for example, using the control unit 21, storage unit 22, communication unit 23, etc. of the server device 12.

[0045] The estimation means 40 estimates the selection probability using a pair comparison model. In this embodiment, the processing of the estimation means 40 corresponds to the processing of the BT model (S13) shown in Figure 7, among the "question design" in Figures 9 and 7. The estimation means 40 is configured, for example, using the control unit 21, storage unit 22, communication unit 23, etc. of the server device 12.

[0046] The selection means 42 selects questions that satisfy neutrality based on the selection probability estimated by the estimation means 40 (automatic selection by neutrality index). In this embodiment, the processing of the selection means 42 corresponds to the optimal selection (S14) process shown in Figure 7 of the "question design" in Figures 9 and 7. The selection means 42 is configured using, for example, the control unit 21, storage unit 22, communication unit 23, etc. of the server device 12.

[0047] The display means 44 displays the position of the calculation result in a multidimensional space when supporting the evaluation of behavior preferences (or when evaluating behavior preferences). In this embodiment, the display processing using the display means 44 when supporting the evaluation of behavior preferences (or when evaluating behavior preferences) can be performed after step A5 (position determination) of the output (S4, Figure 9). Note that the output means for the calculation result is not limited to the display means 44, and other output means (for example, external output means such as audio output means, printing means, notification means, etc.) can also be used. In addition, the display means 44 displays the question when setting a question (question design). This question is selected as a question that satisfies neutrality based on the selection probability estimated by the estimation means 40.

[0048] <6-question diagnostic test (multiple pairs of questions)> In the input (S1) of Figure 1(a), the selection results for multiple paired questions (pair comparisons) are obtained. As for the "paired questions (pair comparisons)," it is possible to adopt a 6-question diagnostic set, as generalized in Figure 3. Note that the "6 questions" referred to here is the number of questions determined when the number of vertices k (Figure 8), which will be described later, is set to 4. The number of questions (6 questions) and the number of categories (4 categories A to D, which will be described later) in the 6-question diagnostic set exemplified here are merely examples, and the number of questions and categories will change depending on the number of vertices k.

[0049] Figure 3 shows a portion of the 6-question diagnostic form 30 used for input (S1). The 6-question diagnostic form 30 is displayed as an image on a display unit such as an LCD screen (such as the display device 20 in Figure 2(a)). For example, it can be displayed as a vertical scrollable image on the screen of a mobile information terminal such as a smartphone.

[0050] The 6-question diagnostic form 30 displays six questions, numbered (1) through (6). Furthermore, for each question (1) through (6) on the 6-question diagnostic form 30, there is a check box 66, a left-hand answer indicator box 68, and a right-hand answer indicator box 70. Note that in Figure 3, the areas of the check box 66, the left-hand answer indicator box 68, and the right-hand answer indicator box 70 are each surrounded by a dashed line, but these dashed lines are not displayed in the image presented to the user.

[0051] In check box 66, six radio buttons are displayed for each question (36 in total for all six questions). In Figure 3, the six circles located below the numbers "1" through "6" in check box 66 represent each of the radio buttons.

[0052] In the example in Figure 3, the numbers "1" through "6" are displayed above the radio buttons. These numbers "1" through "6" indicate the different answer positions. The numbers "1" through "6" are shown in the following explanation to distinguish the six radio buttons, but they may also be displayed on the image presented to the user.

[0053] In the following, when describing each radio button, we will distinguish them by the combination of the number attached to each question (here, the numbers in parentheses for (1) to (6)) and the number corresponding to the answer position "1" to "6" in check box 66.

[0054] For example, for question (1), the radio button for answer position "1" will be represented as radio button 1-1. Also, for example, for question (4), the radio button for answer position "5" in check box 66 will be represented as radio button 4-5. In this way, below, the 36 radio buttons will be distinguished as radio buttons 1-1, 1-2, 1-3, ..., 6-4, 6-5, and 6-6.

[0055] For question (1), only one of the six radio buttons 1-1 to 1-6 can be selected. When the user selects one of the radio buttons 1-1 to 1-6, a black circle, for example, will be added to the center of the selected radio button 1-1 to 1-6.

[0056] In the example in Figure 3, the second radio button from the left, 1-2, is selected. For example, regarding question (1), the string "d-1" is displayed in the left answer indicator column 68 as the indicator for the answer. Furthermore, the string "a-1" is displayed in the right answer indicator column 70 as the indicator for the answer (answer indicator).

[0057] Here, "d-1" is a generalization of the string of characters in a question belonging to D (one of the behavior categories, described later), using the symbol "d-1". A "behavior category" (also simply called a "category") is a classification of behavior. Furthermore, D is one of the four behavior categories A through D.

[0058] Similarly, "a-1" is a generalization of the string of characters in a question belonging to A (one of the behavior categories, described later), using the symbol "a-1". Furthermore, A is one of the four behavior categories A through D.

[0059] The number of action categories A to D (4) corresponds to the number of vertices k being "4" in this embodiment. Furthermore, the number of action categories A to D (4) corresponds to the "number of vertices (number of vertex points)" of "4" related to the "quadrilateral" in Figure 8. The number of action categories is also referred to as the "number of categories."

[0060] The aforementioned "d-1" is shown in general terms in this embodiment, but specifically it is one of the actions that belong to action category D. Similarly, the aforementioned "a-1" is specifically one of the actions that belong to action category A.

[0061] In other words, each of the action categories A through D has three actions defined (a total of 3 x 4 (=12) actions), as follows: Action Category A: a-1, a-2, a-3 Action Category B: b-1, b-2, b-3 Action Category C: c-1, c-2, c-3 Action Category D: d-1, d-2, d-3

[0062] Similarly, for the other questions (2) to (6), left and right answer indicators are defined. For question (2), "a-2" and "c-1" are defined, and for question (3), "b-1" and "a-3" are defined.

[0063] Question (4) specifies "d-2" and "b-2," and question (5) specifies "b-3" and "c-2." Furthermore, question (6) specifies "c-3" and "d-3."

[0064] In question (1), the action category in the left response index column 68 is "D," and the action category in the right response index column 70 is "A." In other words, in question (1), action categories "D" and "A" form a pair (DA).

[0065] Similarly, in question (2), "A" and "C" are combined to form a pair (AC), and in question (3), "B" and "A" are combined to form a pair (BA). In question (4), "D" and "B" form a pair (DB), and in question (5), "B" and "C" form a pair (BC). And in question (6), "C" and "D" form a pair (CD).

[0066] As described above, questions (1) to (6) employ a paired comparison format of the indicators shown in the left response indicator column 68 and the right response indicator column 70. These combinations of response indicators are quantified and used to create a grid (16 grids if there are 4 categories), but the grid (points) and the creation of the grid (points) will be explained later.

[0067] For example, taking question (4) as an example, the range between the response indicators in the left response indicator column 68 and the right response indicator column 70 is divided into six levels using radio buttons 4-1 to 4-6. The user (the person being evaluated, whether themselves or someone else) considers which level between the behavior of "d-2" and the behavior of "b-2" they are.

[0068] The user selects the appropriate radio button from radio buttons 4-1 to 4-6 corresponding to question (4), displays a black circle mark, and enters their answer. If the user selects radio button 4-1, the user's evaluation of the subject will indicate that the tendency towards "d-2" is much stronger (larger, more frequent) than the tendency towards "b-2". In other words, the tendency towards "d-2" is the strongest between "d-2" and "b-2".

[0069] If the user selects radio button 4-2, the user's evaluation of the subject will indicate that the tendency towards "d-2" is considerably stronger than the tendency towards "b-2".

[0070] If the user selects radio button 4-3, the user's evaluation of the subject will indicate that the tendency towards "d-2" is slightly stronger than the tendency towards "b-2".

[0071] If the user selects radio button 4-4, the user's evaluation of the subject will indicate that the tendency towards "b-2" is slightly stronger than the tendency towards "d-2".

[0072] Furthermore, if the user selects radio buttons 4-5, the user's evaluation of the subject will show a much stronger tendency towards "b-2" than towards "d-2".

[0073] Furthermore, if the user selects radio buttons 4-6, the user's evaluation of the subject will show a much stronger tendency towards "b-2" than towards "d-2". In other words, between "d-2" and "b-2", the tendency towards "b-2" will be the strongest.

[0074] Similarly, for the other questions (1)-(3), (5), and (6), the position of the selected radio button represents the relative strength of the tendency related to the behavior indicated by the pair of response indicators.

[0075] Thus, in the 6-question diagnosis, answers of "1" to "6" are obtained for each of the questions (1) to (6). Questions (1) to (6) are structured as a pair comparison between behavior categories (A / B / C / D) (Figure 3). In other words, for each of the six evaluation areas related to each of the questions (1) to (6) (radio buttons 1-1 to 6-6 in this embodiment), one evaluation area was selected for each of the questions (1) to (6).

[0076] <Explanation of the conversion table> Figure 4 shows an example of a transformation table. The transformation table is provided to explain the process performed in the transformation (S2) shown in Figure 1. Furthermore, the transformation table defines the input values ​​for normalization (the process of converting the answer position into a coefficient) in step A1 of the transformation (S2) and for direction vector generation (vi=f(ri,di)) in step A2, and functions as an explanation of the process spanning from step A1 to step A2.

[0077] In the conversion table in the example in Figure 4, the leftmost column shows "Answer Position," "Coefficient (Example)," and "Meaning." "Answer Position" numbers "1" through "6" correspond to answer positions "1" through "6" in the aforementioned 6-question diagnosis (Figure 3).

[0078] The "coefficient" in the next column is the coefficient (numerical conversion function) assigned to each "response position" from "1" to "6". In the example in Figure 4, the following values ​​are assigned to each response position from "1" to "6": for example, "9", "6.5", "4.5", "3", "2", and "1". The "response position" also represents the intensity of preference. Furthermore, the "coefficient" is nonlinear, not arithmetic. This allows for a higher resolution of preference intensity. The "coefficient" is also a function (numerical conversion function) that assigns different weight values ​​according to the intensity (preference intensity). Note that the values ​​of the coefficients described above are examples of the implementation and are not limited to these.

[0079] As shown in the rightmost column, the "meaning" of each coefficient is, in order: "strong preference for option 1," "preference for option 1," "somewhat preference for option 1," "somewhat preference for option 2," "preference for option 2," and "strong preference for option 2."

[0080] The relative strengths of "strong preference," "preference," and "somewhat preference" in the above definitions are as follows: "Strong preference" > "Preference" > "Somewhat preference"

[0081] In other words, a "strong preference for the first option" is more likely to favor the "first option" than a "preference for the first option." Furthermore, a "preference for the first option" is more likely to favor the "first option" than a "somewhat preference for the first option."

[0082] This relationship also applies to "somewhat preferred to the second option," "preferred to the second option," and "strongly preferred to the second option."

[0083] Furthermore, the expressions "strong preference," "preference," and "somewhat preference" only need to indicate the relative strength of each other. For example, as mentioned above, it is also possible to express it as "very strong," "quite strong," or "slightly strong." It is also possible to express it as "strong," "average," or "weak."

[0084] The aforementioned "first option" corresponds to the response indicator in one of the response indicator fields for each question (for example, the left response indicator field 68), and the "second option" corresponds to the response indicator in the other response indicator field for each question (for example, the right response indicator field 70).

[0085] Applying this to the example of question (1) in Figure 3, the "first option" corresponds to "d-1" and the "second option" corresponds to "a-1". For example, if the answer position "6" is selected, as in the example in Figure 3, the "coefficient" becomes "1", which means "a strong preference for the second option".

[0086] <Correspondence between questions and spatial axes (Example: 4-category structure)> In this embodiment, four behavior categories (A / B / C / D) are used. As mentioned above, each question (the six questions in the six-question diagnosis) is structured as a pair comparison between categories (Figure 3).

[0087] Figure 5 shows the "contrast category," "vector direction," and "processing meaning" for each question (question numbers "1" to "6").

[0088] For questions numbered "1" through "6", the "contrast categories" are, in order, "AB", "AD", "AC", "BC", "DB", and "CD". For each "contrast category", the "vector direction" is represented as "mAB", "mAD", "mAC", "mBC", "mDB", and "mCD". The "m" in "mAB" through "mCD" indicates that it is a vector, and "AB", "AD", "AC", "BC", "DB", and "CD" indicate the direction related to the vector.

[0089] For questions numbered "1" through "6", the "processing meaning" is, in order, "preference direction between categories A and B", "preference direction between categories A and D", "preference direction between categories A and C", "preference direction between categories B and C", "preference direction between categories D and B", and "preference direction between categories C and D".

[0090] This correspondence between the question and the spatial axis (Example: 4-category structure) is an explanation that corresponds to process A3 (contribution decomposition to the category (action category) space) in the synthesis (S3) described above, as shown in Figure 9.

[0091] <Spatial structure and grid (point) distribution> <<Overview of Spatial Structure>> In this embodiment, a polygonal space with k categories as vertices is used. The interior of the space is divided into M grids (regions) or points, and a composite vector is calculated for each grid or point.

[0092] In this embodiment, the value of the number of vertices k is set to "4," corresponding to the four action categories (A / B / C / D). In this case, the spatial shape related to the polygonal space becomes a quadrilateral (square), and the number of grids M becomes "16" (=4×4). The types of grids that make up the number of grids M are individual grids, edge grids, and central grids, with the number of individual grids being "4," the number of edge grids being "8," and the number of central grids being "4." As will be described in detail later (Figure 8), "grids" can also be "points."

[0093] This <<Outline of Spatial Structure>>, as shown in Figure 9, is an explanation corresponding to step A3 (contribution decomposition to the category space) in the aforementioned synthesis (S3).

[0094] <<Principles for setting weighting coefficients>> The weight coefficient w(g,i) for each grid (point) is set according to the following principle. Principle 1: The closer a grid (point) is to a vertex, the greater its contribution to the corresponding category vector (the vector shown in Figure 5). (If there are more than 4 grids (points), grids (points) other than vertices are generated.) Principle 2: When a central grid (point) is formed, the central grid (point) evenly combines the contributions of multiple categories. Principle 3: The edge grid (edge ​​points) combines the contributions of two adjacent vertices.

[0095] The specific value of w(g,i) will be determined during implementation (decided during design). Furthermore, the weight coefficient (w(g,i)) for each grid is a weight coefficient that has different values ​​depending on the grid position.

[0096] This principle for setting weighting coefficients corresponds to step A4 (weighted synthesis) in the synthesis (S3) described above, as shown in Figure 9.

[0097] <Time-series integration> In this embodiment, a process is performed to generate a narrative structure by integrating the spatial positions at multiple points in time. The narrative structure is formed by arranging the results obtained in steps A4 (weighted synthesis) and A5 (position determination) at different time points in a time series and creating a time series.

[0098] Figure 6 shows the various "steps" performed for time-series integration and the content of each step. As shown in Figure 6, the following steps are performed for time-series integration: "determining the position at each point in time," "series generation," "extracting the change structure," and "narrative output."

[0099] In "Position Determination at Each Time Point," g*t is calculated at each time point t. In "Series Generation," as described above for process A6 (extension (S)), a time series is constructed with S={g*t|t∈T} (to ensure continuity in the time series). In "Extraction of Change Structure," the direction of movement and / or the amount of change are calculated from the series. In "Narrative Output," the results of the processing are output as a visual display or text.

[0100] This <time series integration> is an explanation corresponding to process A6 (time series integration) in the aforementioned extension (S5), as shown in Figure 9.

[0101] <Main characteristic processing according to this embodiment> In the information processing device of this embodiment, for example, the following characteristic processes (input processing, spatial processing, position determination, time series, question design) are performed. In the input processing (corresponding to "Input (S1)" to "Transformation (S2)" in Figure 1), as shown in Figure 9, vector transformations of direction and intensity (corresponding to process A2) are performed.

[0102] In the spatial processing (corresponding to "Transformation (S2)" to "Output (S4)" in Figure 1), mapping and synthesis to a multidimensional space (corresponding to process A3) are performed, as shown in Figure 9.

[0103] In position determination (corresponding to "Synthesis S3" to "Output (S4)" in Figure 1), as shown in Figure 9, vector norm maximization (corresponding to process A5) is performed. Regarding process A5, "selection of the maximum norm region" corresponds to determining the evaluation rules (framework), and "vector norm maximization" corresponds to selecting the strongest (optimal solution) among the determined rules.

[0104] In the time series (time series generation, corresponding to "extended S5" in Figure 1), as shown in Figure 9, the generation of the series structure and the generation of the narrative (generation of the narrative structure by time series integration, corresponding to process A6) are performed.

[0105] In the question design process, automatic selection is performed using a neutrality index. In this embodiment, the question design (question generation) corresponds to the stage of creating the 6-question diagnostic form 30 (the stage prior to "Input (S1)" shown in Figure 9). This question design can be done by a human, by a computer device programmed with question design procedures, or by a generative AI model that can access the question information.

[0106] <Specific examples of question generation> Figure 7 shows the flow of question generation. In question generation, candidate questions (S11), CATPCA (Category Principal Component Analysis, S12), BT model (Bradley-Terry model, S13), and optimal selection (S14) are performed in order.

[0107] Furthermore, the probability of selecting each question position "1" to "6" in questions (1) to (6) is expressed by the following formula 1.

number

[0108] Furthermore, neutrality B is expressed by the following equation 2.

number

[0109] The response values ​​for response positions "1" through "6" are expressed as shown in formula 3 below.

number

[0110] The vector (action preference vector) is expressed as shown in equation 4 below.

number

[0111] Two-dimensional projection is represented by the following equation 5.

number

[0112] <Generalization of the number of vertices k> In the above description of the embodiment, the number of questions is set to "6" and the number of vertices (number of categories) is set to "4," but the number of questions and the number of vertices can be changed.

[0113] Figure 8 shows the points when the number of vertices k is varied in the range of 3 to 5. From left to right, Figure 8 shows a triangle (equilateral triangle), a quadrilateral (regular quadrilateral), and a pentagon (regular pentagon).

[0114] For example, in a triangle, there are 3 vertices, and the number of points placed at each vertex (vertex points) is 3. Furthermore, if we add 2 points on each of the three sides (edge ​​points, a total of 6) at equal intervals, the total number of points (total points) becomes 9 (=3+6).

[0115] In a quadrilateral, there are 4 vertices, and therefore 4 vertex points. Furthermore, if we add 2 edge points (8 in total) at equal intervals along each of the four edges, and 2 points (4 in total) at equal intervals along each of the two diagonals, the total number of points becomes 16 (=4+8+4).

[0116] In a pentagon, there are 5 vertices, and therefore 5 vertex points. Furthermore, if we add 2 edge points (10 in total) at equal intervals along each of the 5 edges, and 2 points (10 in total) at equal intervals along each of the 5 diagonals, the total number of points becomes 25 (= 5 + 10 + 10).

[0117] The number of diagonals can be calculated using k(k-3) / 2 (where k is the number of vertices), so the number of points remains constant even if the value of the number of vertices k increases. Furthermore, the sum of the number of diagonals and the number of sides (k) corresponds to the number of questions (k(k-1) / 2). This is a mathematical expression of comparing all pairs of categories from k categories. For example, a triangle can be applied to 3 questions (3-question diagnosis), a quadrilateral to 6 questions (6-question diagnosis), and a pentagon to 10 questions (10-question diagnosis).

[0118] Furthermore, if the number of points (number of vertices k) is, for example, k=4, the entire figure (quadrilateral) can be divided into 16 evenly spaced "grids". For this reason, the above-described embodiment was explained assuming M=16 "grids" in space. However, in cases where it is difficult to define a "grid", such as when k=3 or k=5, it is possible to understand the invention by replacing the "grid" in the embodiment with "points" (or "multiple points", "multidimensional points", etc.).

[0119] <Main advantages of the invention according to this embodiment> The information processing method performed by the information processing device (information processing system) 10 according to this embodiment makes it possible to provide novel information processing, such as the mathematical visualization of behavioral preferences.

[0120] More specifically, "behavioral preference" refers to, for example, a user selecting an action that they believe the target person (the subject) is likely to choose in a given situation. For instance, while product evaluation surveys have traditionally been conducted using evaluation criteria such as whether the product is liked or disliked, and how much it is liked, conventional methods were limited to scoring and classification. In contrast, this embodiment is fundamentally different from conventional technology in that it employs a novel information processing method that transforms and arranges the selection data (data representing the selection results) into a vector space.

[0121] Furthermore, in this embodiment, not only is "behavioral preference" the object of evaluation, but "behavioral preference" is also verbalized in the form of multiple pairs of questions (question generation), and users are asked to select from them. Then, in this embodiment, the answers to the multiple pairs of questions are obtained, vectorized, placed in a multidimensional space (two-dimensional in this embodiment), and their positions are displayed in the multidimensional space.

[0122] Therefore, according to the information processing method, information processing device, and information processing program of this embodiment, it becomes possible to mathematically visualize "behavioral preferences" and to diversify individual evaluation indicators more than before.

[0123] Furthermore, since the action preference vector is calculated using the following equation 4, the adoption of nonlinear coefficients improves the resolution of preference intensity, enabling highly reproducible mathematical visualization.

number

[0124] Furthermore, since the question axes are extracted using categorical principal component analysis during question generation, the independence between question axes is mathematically guaranteed, improving measurement accuracy.

[0125] Furthermore, since the selection probability is estimated using a pairwise comparison model during question generation, the selection probability of each question is objectively estimated, ensuring the validity of the questions.

[0126] Furthermore, during question generation, questions that satisfy neutrality are selected based on selection probability. Therefore, an unbiased set of questions is constructed through automatic selection using a neutrality index, ensuring fairness in evaluation.

[0127] Furthermore, since the behavior preference vector is generated in a time series during evaluation, it becomes possible to perform evaluation using a narrative structure. In this embodiment, the "narrative structure" can be described, for example, as a structure in which the subject's behavior preferences are constructed as a series of meanings having temporal or causal relationships.

[0128] <Inventions that can be extracted from the embodiments> The following inventions can be extracted from the embodiments described so far. (1) An information processing method performed by an information processing device (such as an information processing device 10) equipped with acquisition means (such as acquisition means 32), calculation means (such as calculation means 34), and arrangement means (such as arrangement means 36), The acquisition means includes the step of acquiring answers to a plurality of pairs of questions, The calculation means includes the steps of quantifying the answers and calculating an action preference vector based on the axis vectors associated with each question, The arrangement means includes the step of arranging the action preference vectors in a multidimensional space, An information processing method in which the aforementioned question has multiple levels of choices (such as answers in answer positions "1" to "6"). (2) The action preference vector is calculated by the following formula 4 using the information processing method described in (1) above.

number

[0129] <Other> It should be noted that the embodiments described above are merely examples, and those skilled in the art will understand that various changes, substitutions, and modifications can be made thereto without departing from the spirit and scope of the present invention. [Industrial applicability]

[0130] The present invention is applicable to the evaluation of various behavioral preferences (for example, the evaluation of behavioral preferences related to human resource development, education, medical care, organizational analysis, etc.). [Explanation of symbols]

[0131] 10: Information Processing Device 12: Server device 13: Database 14: Communication Network 16: User terminal device 18: Input device 20:Display device 30:6-question diagnostic sheet 32: Acquisition means 34: Calculation method 36: Placement means 38: Generation means 40: Estimation means 42: Selection method 44:Display means

Claims

1. An information processing method performed by an information processing device equipped with acquisition means, calculation means, and arrangement means, The acquisition means includes the step of acquiring answers to a plurality of pairs of questions, The calculation means includes the steps of quantifying the answers and calculating an action preference vector based on the axis vectors associated with each question, The arrangement means includes the step of arranging the action preference vectors in a multidimensional space, The aforementioned question is an information processing method in which multiple levels of choices are selected.

2. The information processing method according to claim 1, wherein the behavior preference vector is calculated by the following formula 4. [Math 4]

3. The information processing method according to claim 1 or 2, further comprising the step of extracting question axes by categorical principal component analysis.

4. The information processing method according to claim 3, further comprising the step of estimating the selection probability using a pair comparison model.

5. The information processing method according to claim 4, further comprising the step of selecting a question that satisfies neutrality based on the aforementioned selection probability.

6. The information processing method according to claim 1 or 2, wherein the calculation means further comprises a step of generating the behavior preference vector in a time series.

7. A means of obtaining answers to multiple pairs of questions, A calculation means that quantifies the aforementioned answers and calculates an action preference vector based on the axis vectors associated with each question, The system includes arrangement means for arranging the aforementioned behavior preference vectors in a multidimensional space, The aforementioned question is an information processing device having multiple levels of choices.

8. An information processing program that causes a computer to execute the information processing method described in claim 1.

Images