Information processing device, information processing method, and program

Abstract

To support the determination of styling policies in fashion and improve user satisfaction.SOLUTION: A server 10 of an embodiment is equipped with a user information acquisition unit 11, a user information conversion unit 12, a slot generation unit 13, an interpretability improvement unit 15, and a styling policy determination unit 16. The user information acquisition unit 11 acquires user information. The user information conversion unit 12 and the slot generation unit 13 convert the user information into a format that includes parameters related to fashion styling policies and generate structured slots from the user information. When detecting inadequacy or ambiguity in the slot information, the interpretability improvement unit 15 improves the interpretability of the slot information through dialogue with a dialogue unit 14. The styling policy determination unit 16 proposes styling policies based on the slot information with improved interpretability, and determines the styling policies while updating the slot information through dialogue with a user.SELECTED DRAWING: Figure 3

Description

[Technical Field]

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

[0002] In recent years, services have been provided in which stylists coordinate outfits over the Internet. When providing this type of service, it is necessary to present outfits that match the user's "preferences" and "what suits" the user. Conventionally, stylists coordinate outfits after receiving request information such as the situation in which the user will wear the outfit. In particular, when providing services over the Internet, it is difficult for users and stylists to communicate directly and reach an agreement on their views, and it is not always possible to coordinate clothing that satisfies users, especially those who use the service infrequently. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2018-106593 Summary of the Invention [Problem to be solved by the invention]

[0004] In the conventional services described above, clothes were coordinated based on registered information such as the situation in which the user would wear the clothes, but if the information was insufficient or vague, a sufficient agreement could not be reached between the user and the stylist, which tended to reduce the user's satisfaction with the outfits coordinated by the stylist.

[0005] In order to solve the above-mentioned problems, the present invention aims to ultimately improve styling satisfaction in fashion styling services by reaching a consensus with users' requests. [Means for solving the problem]

[0006] In order to solve the above problems, the information processing device according to the present invention comprises: An information processing device for supporting decisions on fashion styling policies, information acquisition means for acquiring information about a user as user information; an information conversion means for converting the user information into information in a format including a plurality of parameters related to fashion styling principles; a data set generating means for generating a data set in which the user information is structured based on the user information whose format has been converted; an interpretability improvement means for detecting insufficiencies and ambiguities in the data set using a dialogue system and updating the data set through dialogue with the dialogue system to improve the interpretability of the user information; a styling policy determination means for proposing a fashion styling policy to the user using the dialogue system based on the dataset with improved interpretability, and determining the styling policy while updating the dataset based on the user's response to the proposal; Equipped with. [Effects of the Invention]

[0007] According to the present invention, based on a data set that structures subjective information such as the user's "preferences" and "what suits" as well as constraints, it is possible to resolve insufficient or ambiguous information through dialogue and determine a styling policy that will provide high user satisfaction.

[0008] In addition, in a preferred embodiment of the present invention, natural language processing technology is used to enable dialogue in the natural language that users normally use, enabling more intuitive and flexible information gathering and interpretation.Furthermore, the use of generative AI makes it possible to generate sophisticated dialogue that understands the context and to resolve the ambiguity specific to the fashion industry, which requires complex judgments.

[0009] Furthermore, by including information on the user's "preferences," "what suits me," and restrictions, the system can suggest appropriate styling guidelines that the user may not even be aware of. This is particularly effective in eliminating the inadequacy and ambiguity of "preferences" and "what suits me." By quantifying subjective evaluations and converting them into classifications such as fashion taste and color, the accuracy of processing through dialogue with the user is improved, and specialized fashion elements can be taken into account.

[0010] Furthermore, interpretability can be improved stepwise and efficiently by individually determining the insufficiency and ambiguity of the dataset and generating appropriate dialogue for each.Furthermore, by determining the type of user response (agreement, disagreement, question, request) and generating dialogue accordingly, styling policies can be determined efficiently and in line with the user's intentions. [Brief explanation of the drawings]

[0011] [Figure 1] 1 is a diagram showing the overall configuration of an information processing system including a server according to an embodiment of the information processing device of the present invention; [Figure 2] 2 is a block diagram showing a hardware configuration of a server included in the information processing system of FIG. 1. FIG. [Figure 3] 3 is a block diagram showing the functional configuration of a server having the hardware configuration of FIG. 2. FIG. [Figure 4] 4 is a flowchart showing the processing operation of a server having the functional configuration of FIG. 3. [Figure 5] FIG. 4 is a diagram showing an outfit list screen presented to a user on a website provided by the server of FIGS. 1 to 3. [Figure 6] FIG. 4 is a diagram showing an item list screen presented to a user on a website provided by the server of FIGS. 1 to 3. [Figure 7] FIG. 4 is a diagram showing a title screen (initial screen) for determining a user's styling policy on a website provided by the server of FIGS. 1 to 3. [Figure 8] FIG. 8 is a diagram showing an interactive screen transitioned from the title screen of FIG. 7. [Figure 9] 9 is a diagram showing a termination screen that is displayed when the dialogue using the dialogue screen of FIG. 8 is terminated. FIG. DETAILED DESCRIPTION OF THE INVENTION

[0012] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 shows the overall configuration of an information processing system including a server according to an embodiment of the information processing device of the present invention. The information processing system shown in FIG. 1 is configured by connecting a server 10 and a user terminal 20 to each other via a predetermined network NW such as the Internet.

[0013] The user terminal 20 is an information processing device managed by a user who has registered as a member of a personal styling service (hereinafter referred to as "this service") realized by the functions of the server 10, and is configured, for example, as a tablet terminal or a smartphone. In this embodiment, an example is shown in which one user terminal is connected to the network NW, but there may be cases in which multiple user terminals managed by multiple users are connected, and the number of user terminals 20 is not limited to one. The server 10 publishes the website of the service on the Internet and provides the service through the website. The server 10 is managed by the service provider of the service. Typically, the server 10 provides styling services by accessing a website at a predetermined URL from a web browsing application program such as a web browser on the user terminal 20 and logging in. Hereinafter, the application program will be abbreviated as "app." Alternatively, the user can access the styling service in the same way as by accessing a website by launching the app for this service, which is pre-installed on the user terminal 20, through a predetermined operation (such as clicking on an icon) and displaying the screen of the app. In this case, the server 10 executes various processes while controlling the operation of the user terminal 20 in cooperation with the app.

[0014] Next, the hardware configuration of the server included in the information processing system of FIG. 1 will be described with reference to FIG. FIG. 2 is a block diagram showing a hardware configuration of a server included in the information processing system of FIG.

[0015] The server 10 includes a CPU (Central Processing Unit) 101, a ROM (Read Only Memory) 102, a RAM (Random Access Memory) 103, a bus 104, an input / output interface 105, an output unit 106, an input unit 107, a memory unit 108, a communication unit 109, and a drive 110.

[0016] The CPU 101 executes various processes according to a program recorded in the ROM 102 or a program loaded from the storage unit 108 into the RAM 103 . The RAM 103 also stores data and the like necessary for the CPU 101 to execute various processes.

[0017] The CPU 101, ROM 102, and RAM 103 are connected to one another via a bus 104. An input / output interface 105 is also connected to this bus 104. An output unit 106, an input unit 107, a storage unit 108, a communication unit 109, and a drive 110 are connected to the input / output interface 105.

[0018] The output unit 106 is composed of a display, a speaker, etc., and outputs various information as images and sounds. The input unit 107 is composed of a keyboard, a mouse, a camera, a microphone, etc., and is used to input various types of information.

[0019] The storage unit 108 is configured with a hard disk, a DRAM (Dynamic Random Access Memory), etc., and stores various data. The communication unit 109 communicates with other devices (such as the user terminal 20 in the example of FIG. 1) via a network NW including the Internet.

[0020] Removable media 111, such as a magnetic disk, optical disk, magneto-optical disk, or semiconductor memory, is appropriately loaded into the drive 110. A program read from the removable media 111 by the drive 110 is installed in the storage unit 108 as needed. Furthermore, the removable medium 111 can also store various data stored in the storage unit 108 in the same manner as the storage unit 108 .

[0021] Next, the functional configuration of a server having the hardware configuration of FIG. 2 will be described with reference to FIG. FIG. 3 is a block diagram showing the functional configuration of a server according to an embodiment of the present invention. As shown in FIG. 3, a database 70 (hereinafter referred to as "DB 70") is provided in one area of ​​the storage unit 108 (FIG. 2) of the server 10. The DB 70 stores various pieces of information about the user acquired by the user information acquisition unit 11. Specifically, DB70 stores user information such as the items available within the service and basic information about the user as a member (name, age, occupation, contact information, login information, payment information, etc. linked to the user's identification information (user ID)). In addition, DB70 stores information about styling related to "preferences" and "what suits" obtained through interactions with users on the service's website as slot information (dataset) linked to the user ID. The DB 70 stores, for example, fashion tastes and colors as fashion elements (also called fashion definitions or indicators) for determining styling policies. Fashion taste refers to the style or taste of outfits. Specifically, fashion taste is classified into cool, feminine, casual, etc. Color is classified into tones such as vivid and grayish. The fashion tastes and colors shown in this embodiment are merely examples, and the number of types of fashion elements and the way of classification may be changed as appropriate and are not limited to those shown here. The processing is executed by writing and reading information to and from the DB 70 using the functional configuration described below.

[0022] In addition, when executing processing for reaching an agreement on a styling policy in a styling service (styling policy agreement processing), the CPU 101 of the server 10 functions as a user information acquisition unit 11, a user information conversion unit 12, a slot generation unit 13, a dialogue unit 14, an interpretability improvement unit 15, and a styling policy determination unit 16.

[0023] The user information acquisition unit 11 acquires information about the user as user information. The acquired user information mainly includes information about styling policies related to "preference" and "what suits you," and information that is not related to "preference" or "what suits you" but may restrict the styling policies.

[0024] Examples of information about styling guidelines related to "taste" and "what suits you" include the following: styling guidelines related to "taste" and "what suits you" that are manually assigned by stylists based only on information about the people and their clothing in photos registered when registering as a member, styling guidelines that are rule-based conversions of "the impression you want others to give you," selection information about outfits you want to wear, selection information about your favorite items, and things you should take into consideration when styling (free description).

[0025] On the other hand, examples of information that may limit styling guidelines include: what you plan to wear to work, the type of clothes you wear to work, the brands you want delivered, your occupation, things that apply to your everyday life (you wear jackets, white coats, work clothes, you often ride a bicycle, etc.), things you want the stylist to take into consideration when choosing clothes (body shape characteristics, etc.), your height, your fit preferences, characteristics of items you do not want delivered (NG information), etc.

[0026] The user information conversion unit 12 converts the acquired user information into information in a format including a plurality of parameters related to the fashion styling policy. Specifically, the user information conversion unit 12 includes a parameterization unit 41 and a classification conversion unit . The parameterization unit 41 quantifies the user's subjective evaluation. The classification conversion unit 42 converts at least one of the item image or description information selected by the user from the group of item images in the item list presented to the user into a fashion element (for example, a fashion taste and color classification).

[0027] For example, for fashion taste, several types of tastes such as cool, feminine, and casual are defined, and the user's preference and degree of suitability for each are expressed as a numerical value between 0 and 1.

[0028] Similarly, for color, several tones such as vivid and grayish are defined, and the user's preference and degree of suitability for each tone are quantified. It should be noted that the examples of fashion elements shown here are merely examples, and these fashion tastes, colors, etc. may be of other types or classifications, and the number of types of fashion tastes and colors, etc. is not limited.

[0029] The slot generating unit 13 generates a data set in which the user information is structured, that is, a slot, based on the converted user information. The slot contains the following information:

[0030] The information that users think is suitable is Styling principle 1:0.1 Styling principle 2:0.7 Styling Policy 3:0.0 Reason: She was mentioned to have been told by people around her that she looks good in clothes like those in Styling Guideline 2.

[0031] The user's preference information is Styling principle 1:0.4 Styling principle 2:0.2 Styling principle 3:0.8 The reason is that she explicitly mentioned that she likes the clothes in styling principle 3.

[0032] Examples of information (restriction information) that may restrict styling policies include occupation: civil servant, planned wear at work: yes, and body type characteristics: broad shoulders. The examples of restriction information given here are just examples, and a wide variety of other information may also be restriction information. This restriction information and styling-related information such as the user's preferences, linked to a user ID, constitute a data set. A structured version of this data set is called a slot. The dialogue unit 14 can efficiently process slot information, which has the effect of making it easier to detect insufficient or ambiguous information.

[0033] The dialogue unit 14 is a dialogue system using a generative AI that can dialogue with the user in free language, and specifically, it displays information such as an inquiry in a text box arranged on a dialogue screen (chat screen) prepared on a website, for example, as shown in Fig. 8, and dialogues with the user through responses entered by the user. In addition to the text box, dialogue with the user may also be conducted by voice conversation, for example.

[0034] The interpretability improvement unit 15 detects insufficiency and ambiguity of information in a slot (dataset) using the dialogue unit 14, and improves the interpretability of the user information by updating the slot information through dialogue by the dialogue unit 14. The dialogue unit 14 generates dialogue with the user according to the insufficiency and ambiguity.

[0035] The interpretability improving unit 15 includes an insufficiency determining unit 51, an ambiguity determining unit 52, a dialogue generating unit 53, a slot updating unit 54, and the like.

[0036] The insufficiency determination unit 51 detects (determines) the insufficiency of slot information. For example, it checks registered information from which a styling policy can be acquired, and determines that the information is insufficient if the styling policy cannot be correctly acquired from the registered information.

[0037] The ambiguity determination unit 52 detects (determines) the ambiguity of the data set. Ambiguity can occur in the following cases a. to c. a. When the priority styling policies for "preference" and "looks good" are different (e.g. preference is feminine, but looks good in casual) b. When there is no clear priority between the styling principles of "preference" and "suitability" (e.g., she seems to like both feminine and casual styles) c. When the preferred styling policy, prohibited items, occupation, brand, etc. do not match (e.g., a civil servant plans to wear the clothing at work, but casual attire seems more appropriate)

[0038] The dialogue generation unit 53 cooperates with the dialogue unit 14 to provide a dialogue function using a generation AI. Specifically, the dialogue generation unit 53 generates an appropriate dialogue sentence with the user according to the detected (determined) insufficiency or ambiguity, and outputs the dialogue sentence to the dialogue unit 14, thereby presenting the dialogue sentence on the dialogue screen (see FIG. 8). Note that in this example, the dialogue generation unit 53 is one function of the interpretability improvement unit 15, but the function of the dialogue generation unit 53 may be included in the dialogue unit 14. for example: "What kind of clothes do you usually wear?" (If information is insufficient) "Since the client is a civil servant, I would like to suggest something that avoids a loose silhouette. Is that okay?" (in the case of ambiguity)

[0039] The slot update unit 54 sequentially updates the slots based on the results of the dialogue. For example, if the user answers "I like cool tastes," the slot update unit 54 updates the slots by increasing the corresponding cool value in the information on the user's slots using data related to "preferences."

[0040] In this way, the interpretability improving unit 15 can propose a more accurate styling policy by detecting problems in the information of the user's slot and resolving the problems through dialogue with the user.

[0041] The styling policy determination unit 16 proposes a styling policy for fashion to the user using the dialogue unit 14 based on the slot information with improved interpretability, and determines the styling policy while updating the slot information based on the user's response to the proposal. The styling policy determination unit 16 includes a styling policy proposal unit 61 , a response acquisition unit 62 , a response type determination unit 63 , a dialogue generation unit 64 , and a slot update unit 65 .

[0042] The styling policy suggestion unit 61 proposes an optimal styling policy based on the slot information. For example, the styling policy is proposed with a message such as, "Understood. For the first styling session, we will consider a style that combines a casual taste with vivid tones. In particular, we will suggest a combination of a vivid color top with casual pants, and a loose-fitting shirt."

[0043] The response acquisition unit 62 acquires the user's response to the suggestion. The response type determination unit 63 determines whether the response is an "agreement," "disagreement," "question," or "request." The dialogue generation unit 64 cooperates with the dialogue unit 14 to provide a dialogue function using a generation AI. Specifically, the dialogue generation unit 64 generates an appropriate dialogue sentence according to the result of the determination by the response type determination unit 63, and outputs the dialogue sentence to the dialogue unit 14, thereby presenting the dialogue sentence on the dialogue screen (see FIG. 8). Note that in this example, the dialogue generation unit 64 is one function of the styling policy determination unit 16, but the function of the dialogue generation unit 64 may be included in the dialogue unit 14.

[0044] The slot update unit 65 updates slot information based on the content of the dialogue. For example, if the user responds negatively, saying, "I prefer colored items, not grayish tones," the slot update unit 65 updates the "preference" information in the slot information and increases the value of vivid tones. This causes the styling policy suggestion unit 61 to generate a new suggestion.

[0045] By repeating this process, it is possible to determine the optimal styling policy that matches the user's preferences and constraints. During this process, the type of user response is determined and a dialogue is generated accordingly, which has the effect of enabling efficient determination of a styling policy that is in line with the user's intentions.

[0046] Now, the operation of the server will be described with reference to FIG. FIG. 4 is a flowchart showing the processing operation of the server having the functional configuration of FIG.

[0047] In the server 10, in step S11 of FIG. 4, the user information acquisition unit 11 acquires user information.

[0048] In step S12, the user information conversion unit 12 converts the user information into a format including parameters related to the fashion styling policy.

[0049] In step S13, the slot generating unit 13 generates a slot (data set) that structures the user information based on the converted user information.

[0050] In step S14, the interpretability improving unit 15 detects insufficiency and ambiguity of the user information of the slot, and improves the interpretability of the user information by updating the slot information through a dialogue by the dialogue unit 14 (dialogue system).

[0051] In step S15, the styling policy determination unit 16 uses the dialogue unit 14 to propose a styling policy based on the slot information (dataset) with improved interpretability, and determines the styling policy while updating the slot information based on the user's response.

[0052] In this way, according to the operation of the server of the embodiment, a slot is generated that structures user information obtained through dialogue with the user, and further dialogue with the user is carried out to resolve insufficiencies and ambiguities in the slot information, after which a styling policy is proposed and the styling policy is determined based on the user's response to the proposal, thereby ultimately improving the user's satisfaction with styling.

[0053] The operation of the server according to the embodiment will be described below with reference to FIGS. 5 to 9, using specific examples. First, the operation of obtaining and converting user information will be described with reference to FIGS. FIG. 5 is a diagram showing an outfit list screen presented to a user on a website provided by the server of FIGS. FIG. 6 is a diagram showing an item list screen presented to a user on a website provided by the server of FIGS.

[0054] In the above-mentioned user information acquisition step S11, since it is more efficient to acquire the information that must be asked of the user first, currently, the user is asked to access the website disclosed by the server 10 from the user terminal 20, and the user information is acquired along with the member registration. Specifically, when proposing a styling policy, two types of information, namely, first information and second information, are acquired. Details of the first and second information are described below.

[0055] The first information is information about styling principles related to the user's "taste" and "what suits them." In order to suggest clothes that match the user's "taste" and "what suits them," the service acquires the following information as information about the user's "taste" and "what suits them" and stores it in DB70.

[0056] The first information includes styling guidelines regarding "preferences" and "what suits" that are manually assigned by stylists based solely on information about the people and their clothing in photos registered when registering as a member. It also includes styling guidelines that are rule-based conversions of the following "impressions you want others to see." Specifically, these include calm, soft, reassuring, lively, sharp, proactive, kind, cute, refreshing, mature, profound, trustworthy, etc.

[0057] Furthermore, the coordinate list screen shown in FIG. On the outfit list screen, Q1 is displayed to prompt the user to make a selection by displaying a message such as "Please put a heart on the outfit that best suits the style you would like to wear using this service." Then, the user selects the outfit they want to wear (the outfit they want) from the outfits displayed on the outfit list screen. Each outfit is tagged with a styling tag, and the percentage of users who select each tag represents the degree to which they like the styling style.

[0058] In addition, the item list screen shown in FIG. The item list screen displays a message Q2 (not shown), such as "Please give a heart to your favorite items on this service," to prompt the user to make a selection. Then, the user selects their preferred items (desired items) from the group of items displayed on the item list screen. Each item is tagged with a styling policy tag, and the selection rate of each tag represents the degree of preference for the styling policy that the user thinks fits.

[0059] The second type of information is information that is not related to "preference" or "suitability" but may restrict styling policy. The service requires that clothes be selected based not only on the user's preferences and what suits them, but also on their lifestyle. For example, if you are pregnant, you will need to choose clothes that fit loosely. Therefore, the following information is obtained as possible restrictions on styling policy. Do you plan to wear it to work ("Yes" or "No"), type of clothes you wear to work (casual, suit, business casual), brand you want delivered (selected brand select option, a pay-per-use service, and registered brands), occupation (management / planning, secretary, general affairs / accounting / clerical, technical / research, professional (legal / consultant / finance), civil servant, medical / healthcare related, fashion related, beauty / esthetics related, food / drinks industry, service / customer service industry, creative / creator, artist / model, media related, self-employed, full-time housewife, sales, public relations, other, things that apply to your daily life (wear jackets, white coats, work clothes, often ride a bicycle, have a preschool child, are breastfeeding, are pregnant), what you want the stylist to consider when choosing clothes (around the face, round, square jaw, big), shoulders (sloping shoulders, hunched shoulders, broad shoulders), upper arms (thick, muscular), chest (small, big), waist (thin, thick), hips (big, small), thighs (thick), calves (thick, muscular), what are you most looking for in a styling service (I want to flatter my size / figure, I want to have more work clothes that I can mix and match, I want to try different tastes and trends, I want to wear clothes that will work in my desired occasions, I want to try a brand I've never tried before, I want to know what clothes suit me now, height (min: 146cm, max: 174cm), fit preference (loose (less revealing), slightly loose, just right, slightly tight, tight (more revealing), leave it to the stylist), characteristics of items you do not want delivered (NG), etc.

[0060] Much of the acquired user information is in the form of text, or information that requires interpretation, such as "coordinates that users would like to wear using this service." For this reason, it is necessary to convert this user information into information that is easy to handle later. In this case, information about "preference" and "looks good" is extracted based on information about styling policies related to "preference" and "looks good." In particular, taking into consideration the characteristics of this service, the "degree of styling policy" and the reasons for each of the following four types of items and information are set.

[0061] Specifically, the settings include what the user thinks looks good on them (a numerical value is generated by the AI ​​based on free-form information), what the user thinks are their preferences (a numerical value is calculated based on the percentage of tags assigned to the user's favorite items and outfits), what the stylist thinks looks good on the user (the stylist reviews multiple photos of the user and calculates a numerical value based on the percentage of what "looks good" set for each photo), and what the stylist thinks the user likes (the stylist reviews the clothes worn by the user in photos and calculates a numerical value based on the percentage of preferences set for each photo). An example of a numerical representation of what a user thinks looks good on them is styling strategy 1: 0.1, styling strategy 2: 0.7, styling strategy 3: 0.0, etc. The reason why the value for styling strategy 2 is higher than the others is because the user herself mentioned that people around her often tell her that clothes like those in styling strategy 2 look good on her.

[0062] Next, based on information that is not related to "preference" or "what suits" but may be a restriction on styling policy, information is obtained directly as tags such as "occupation: civil servant." However, by separating the information as follows and extracting the necessary information as appropriate, we can expect to reduce mistakes in interpreting the dialogue and the amount of calculations required. Information about preferences (brand, silhouette, etc.), information about personal lifestyle and thoughts (things to consider when choosing clothes, occasions for wearing them, etc.), and personal information (occupation, height, etc.) are basically unchanged during dialogue and will not be updated. The above information group is called a slot, and each time a conversation with the user occurs, the conversation is conducted based on the information in each slot. Also, after one round of conversation, the slot is updated according to the content of the conversation. However, the user information that forms the basis of these slots is not necessarily understandable because input is not required and free text is obtained. Therefore, before proposing styling policies, it is necessary to first check the interpretability of the user information and improve it if necessary.

[0063] An example of improving the interpretability of user information will be described below. The interpretability improving unit 15 checks whether the interpretability of the slot provides sufficient information. If the result of the check indicates that the interpretability is insufficient, the interpretability of this information is improved through dialogue. In this case, the following factors can cause the slot interpretation to deteriorate: For example, there may be cases where there is not enough information to determine a styling policy (insufficient information), or where the interpretation of the styling policy or its details is not unique (ambiguous information). In order to resolve such "insufficiency" or "ambiguity" in the information, the user information registered in DB70 is first judged based on rules such as whether or not it meets pre-set judgment conditions to determine whether it is "insufficient" or "ambiguous."

[0064] When the judgment conditions are met, the interpretability improvement unit 15 uses the dialogue unit 14 (a dialogue system using a generation AI) to dialogue with the user to resolve the issues, thereby resolving each of the issues. Finally, the interpretability improving unit 15 updates the user information (slots in the DB 70) based on these dialogue histories, thereby enabling the styling policy determining unit 16 to propose the next styling policy. When interacting with a user, it is conceivable to have the dialogue unit 14 (a dialogue system using a generative AI) engage in dialogue that solves all problems at once, but it is difficult to confirm whether all problems have been solved correctly.In addition, it is difficult to incorporate the desired dialogue flow into the series of dialogues conducted by the generative AI.

[0065] Therefore, a generation AI is prepared to carry out a dialogue aimed at resolving each issue, and each issue is resolved by sequentially processing the following. Specifically, a dialogue generation unit 53 is provided in the interpretability improvement unit 15, and a dialogue generation unit 64 is provided in the styling policy determination unit 16. Dialogues such as guidance to the user are carried out by the dialogue unit 14. It is determined whether the issue has actually been resolved in the dialogue carried out by each dialogue generation unit 53. In the interpretability improvement unit 15, the dialogue generation unit 64 and the dialogue unit 14 solve each problem through dialogue with the user, and after all problems are finally solved, the styling policy determination unit 16 uses the information to propose a styling policy.

[0066] Here, a solution to the problem will be described. The insufficiency determination is performed by the insufficiency determination unit 51. Here, a method for determining whether slot information is insufficient by the insufficiency determining unit 51 and a method for acquiring new information will be described.

[0067] In this case, first, the insufficiency determination unit 51 reads registered information from which a styling principle can be acquired (information on styling principles related to "preference" and "looks good") from the DB 70. Note that if a styling principle can be correctly acquired from even one piece of registered information, it is deemed not to be insufficient.

[0068] If the insufficiency determination unit 51 determines that the registered information is insufficient as a result of determining the insufficiency, the generation AI such as the dialogue generation unit 53 and the dialogue unit 14 acquires information from the user to resolve the insufficiency by having a dialogue such as the example below in order to align the styling policy. For example, a question such as "What kind of clothes do you usually wear?" is asked. Each time such a dialogue is held, a dialogue history is accumulated, and the slot update unit 54 updates the slot based on the dialogue history.

[0069] As described above, the dialogue between the dialogue generation unit 53 and the dialogue unit 14 is continued to resolve the insufficiency, and when it is determined that the information has been acquired, the insufficiency determination process is terminated. At this time, the method for determining that the information has been acquired will be described later.

[0070] Next, a solution for disambiguation will be described. The ambiguity determination is performed by the ambiguity determination unit 52. Here, a method for determining the ambiguity of slot information by the ambiguity determining unit 52 and a method for acquiring new information will be described.

[0071] Currently, the following cases are considered to contain ambiguity in styling information: a. There are cases where the priority styling policies for "preference" and "suitability" differ. For example, a user may have a feminine preference but look good in casual styles. b. In some cases, it is difficult to determine a single priority policy among the styling policies of "preference" and "suitability." For example, a user may like both feminine and casual styles. c. When the priority styling policy does not match the prohibited items, occupation, brand, etc. For example, if a civil servant plans to wear the clothes to work, but casual clothing would suit them better. For this reason, it is necessary to determine the ambiguity of the above three points a, b, c, etc.

[0072] Specifically, the ambiguity determination unit 52 first compares the degree of "preference" or "suitability" for each slot in order to resolve a. and b., and if the difference in the comparison results is below a predetermined threshold, it determines that ambiguity exists. To resolve c), the ambiguity determination unit 52 first obtains the styling policy that maximizes the degree of "preference" or "suitability" for each slot. Then, the NGs, occupations, brands, etc. for each slot are compared with the obtained styling guidelines, and ambiguity is determined according to pre-set conditions (rules). Based on the above, the ambiguity of the three points a., b., c., etc. is determined.

[0073] If the ambiguity determination unit 52 determines that ambiguity exists as a result of determining whether or not there is ambiguity in this manner, the dialogue generation unit 53 and the dialogue unit 14 acquire information from the user to resolve the ambiguity by having a dialogue with the user as shown in the example below. For example, "Since the customer is a civil servant, I would like to suggest something that avoids a loose silhouette. Is that okay?" Each time such a conversation takes place, a conversation history is accumulated, and the slot update unit 54 updates the slot based on the conversation history.

[0074] As described above, the dialogue between the dialogue generation unit 53 and the dialogue unit 14 is continued to resolve the ambiguity, and when it is determined that the information has been acquired, the ambiguity determination process is terminated. Note that the method for determining that the information has been acquired will be described later.

[0075] As a result, insufficient or ambiguous user information is resolved, making it possible to propose more accurate styling policies.

[0076] Here, a method for determining whether information capable of resolving insufficiency or ambiguity has been acquired will be described. In the dialogue between the user and the dialogue generation unit 53 and the dialogue unit 14 (a dialogue system using a generation AI), when "ending the dialogue because the styling policy has been obtained," three pieces of information are required for each dialogue (start of dialogue, continuation of dialogue, end of dialogue). In this project, the generation AI is used for all decisions except for starting a dialogue. Specifically, it decides whether to continue or end the dialogue based on the dialogue history. Furthermore, when the dialogue continues, the generation AI guides the dialogue to achieve each goal based on the dialogue history and input prompts. This makes it possible to improve the interpretability of user information.

[0077] Next, a method for proposing a styling policy will be explained. A styling principle indicates the atmosphere of the clothes to be styled. In this embodiment, the styling principle is classified into several fashion elements. Specifically, the fashion elements are indicated by several types of fashion tastes (the tastes of DB70 in FIG. 3) and colors (the colors of DB70 in FIG. 3).

[0078] In fashion taste, axes related to the characteristics of clothing are defined, and each is classified based on its qualitative degree. Based on these, the styling policy determination unit 16 determines "preference" and "suitability" in terms of clothing taste.

[0079] The colors are classified into several types with different tones, such as vivid, grayish, and pastel, and each color has defined characteristics such as the impression it gives to the viewer. Based on these, the styling policy determination unit 16 determines "preference" and "suitability" for the color of clothes.

[0080] The examples of the classification of fashion tastes and colors shown here are merely examples, and other classifications may also be used. For example, images of outfits, images of clothing, images of colors (tones), images of celebrities, etc. The items may be classified by the item itself, the attributes or categories of the image items, bone structure, personal color, etc.

[0081] The styling policy determination unit 16 leads the user and the user to an agreement on a fashion styling policy through a dialogue using the dialogue generation unit 64 and the generation AI of the dialogue unit 14 or the like. In this case, first, the previously acquired slots relating to "preference" and "suitability" are converted to make them easier to use. In particular, since the degree of each styling policy is not unique for the user slots stored in the DB 70, a slot is prepared that collects all the information obtained through dialogue with the user. In this embodiment, the styling policy proposing unit 61 of the styling policy determining unit 16 consolidates information about the user obtained from the user by the dialogue generating unit 64. Note that the method of consolidating the information can be a method such as simply adding up the information, and the method of consolidating the information is not limited.

[0082] Next, the styling policy proposal unit 61 inputs the summarized information slot and the dialogue history information into the dialogue generation unit 64, and further inputs the proposal content generated by the dialogue generation unit 64 into the dialogue unit 14, which then presents the proposal content on the application screen of the user terminal 20, thereby proposing a styling policy to the user.

[0083] After proposing a styling policy to the user, the user inputs a response to the proposal on the application screen, and the styling policy proposing unit 61 obtains a response from the user. The styling policy proposing unit 61 determines a response policy based on the received response and updates the slot.

[0084] In order to bring the styling policy proposal closer to the user's expectations and to detect the end of the dialogue, the styling policy proposal unit 61 checks the type of response obtained from the user and determines the type of content to reply based on that type.

[0085] After proposing a styling policy, the response obtained from the user and the next action expected from this response are considered. Specifically, responses obtained from users are classified into, for example, "agreement," "disagreement," "question," "request," and the like. For example, if the response obtained from the user is "agreement," the styling policy proposing unit 61 ends the dialogue. If the response is "oppose," the styling policy proposal unit 61 instructs the slot update unit 65 to change (update) the slot, changes the slot, and re-proposes a styling policy for "oppose." If the response is a "question," the styling policy proposal unit 61 responds to the content of the "question." If the response is a "request," the styling policy proposal unit 61 causes the slot update unit 65 to change the slot and responds to the content of the "request." Note that the above example of response classification is just an example, and responses obtained from users may be classified using a generation AI.

[0086] In this way, the styling policy proposal unit 61 inputs dialogue sentences (prompts) generated by the dialogue generation unit 64 according to the classification into the dialogue unit 14, thereby guiding the dialogue with the user to achieve each objective.

[0087] As long as there is a response from the user after the proposal from the styling policy proposal unit 61, the processing other than proposing the styling policy is repeatedly executed. Note that when a styling policy is proposed again, the processes of updating the slot, reloading the updated slot, creating the styling policy, etc. are performed again. As a result, styling guidelines can be proposed and consensus reached with the user.

[0088] Here, the overall flow of this service will be explained. The user first registers through the app on the user terminal 20, selects images of their favorite outfits and items, and inputs information about their body type, occupation, and lifestyle. This information is acquired by the user information acquisition unit 11 and parameterized by the user information conversion unit 12.

[0089] The converted information is structured as a slot (dataset) and input to the dialogue unit 14. The dialogue unit 14 analyzes the slot information and asks the user questions such as, "From the selection of photos, it seems that you like clothes in grayish tones. What kind of fashion is required in the workplace?"

[0090] Through such dialogue, slot information is updated based on the user's answers, and the interpretability in the interpretability improving unit 15 is improved. As the dialogue unit 14 continues to dialogue with the user, when the interpretability improvement unit 15 determines that sufficient interpretability has been achieved, the styling policy determination unit 16 proposes a styling policy such as, "For the first styling, we will suggest an outfit that mainly consists of grayish-toned clothes suitable for office casual wear, with one vivid-colored top added as an accent."

[0091] When a user responds to a proposed styling policy with information such as "avoid vivid colors," the proposal is modified based on the user's response, and a final styling policy is determined between the user and the server 10. This policy is used as a guideline when the stylist actually selects items.

[0092] Hereinafter, with reference to FIGS. 7 to 9, examples of interactions on a website (application screen) on the server of the embodiment will be described. FIG. 7 is a diagram showing an initial screen (hereinafter referred to as a "title screen") for determining a user's styling policy on the website provided by the server of FIGS. FIG. 8 is a diagram showing a dialogue screen transitioned from the title screen of FIG. FIG. 9 is a diagram showing a termination screen that is displayed when the dialogue using the dialogue screen of FIG. 8 is terminated.

[0093] On the website (application screen), the title screen shown in FIG. 7 is displayed to a user who has completed membership registration. At the top of the title screen is the heading "Here are some suggestions from your AI stylist assistant." In addition, a text box located in the center of the title screen displays a message beginning with "Thank you very much for registering for this service!" Additionally, at the bottom of the title screen, there is the time until styling begins (12:00:00 remaining), a "See suggestions" button, and a "Skip" button.

[0094] When the user clicks on the "View the proposal" button on the title screen, the dialogue screen shown in FIG. 8 is displayed.

[0095] On the dialogue screen shown in FIG. 8, the dialogue unit 14 presents questions, suggestions, etc. about the user's preferences and what suits them. In this example, a text box with a white background displays a message from the AI ​​stylist assistant. Specifically, the message displayed is, "Based on your photos and favorite items, we feel that vivid and dark colors suit your preferences. Which color would you prefer for your first styling session?" If the user responds to this message by typing, "Vivid, please," the AI ​​stylist assistant will then respond with a thank you message and ask a few questions based on the user's answer. Specifically, a thank you message such as "Thank you very much. We will consider how to meet your request" is displayed. Next, they are asked questions such as, "From the outfits that are close to your preferences and your favorite items, I get the feeling that you have a preference for a cool style. Also, from the photos, it seems that a casual style would suit you particularly well. In your first styling session, would you prioritize clothes that you "like" or clothes that "seem to suit you"?" If the user responds to this message by typing, "A casual style, please!", the AI ​​stylist assistant will provide suggestions based on the user's response. Specifically, "Understood. So, for the first styling session, we will consider a style that combines a casual style with vivid colors. In particular, we will suggest coordinating a vivid top with casual pants, or a loose-fitting shirt. We will select items that can be used both in the office and at home, and suggest coordinating a look that balances your usual style with a new one. If you have any questions or requests, please feel free to contact us.' Suggestions such as these will be presented. If the user responds to this message with a response urging confirmation, such as "That's a great suggestion! I'll go with this!", the AI ​​stylist assistant will interpret the user's response as an agreement with the styling direction proposed to the user and will respond that the user's styling request has been confirmed. Specifically, the response might be something like, "I understand. Now, I will propose a style that combines a casual style with vivid tones, so please look forward to it."

[0096] In this way, when a final styling policy is determined in response to specific requests from the user, the dialogue unit 14 displays an end screen, as shown in FIG. 9, indicating that the dialogue has ended with the user's consent. The closing screen shown in Figure 9 displays a title such as "Thank you for your hard work! Your first styling request has been completed," so the user knows that their first styling request has been completed and can close the app.

[0097] As described above, according to the present invention, a styling policy can be determined through interactive dialogue with a user, taking into consideration information regarding the user's "preferences" and "what suits" the user, as well as constraints on the styling policy in an integrated manner. This allows for a shared understanding between the user and stylist, something that was difficult to achieve with conventional online personal styling services, and can improve satisfaction with the final styling.

[0098] Although one embodiment of the present invention has been described above, the present invention is not limited to the above-described embodiment, and modifications, improvements, etc. within the scope of achieving the object of the present invention are included in the present invention. In the above embodiment, styling is performed at the same time as member registration, but it goes without saying that styling can be performed on other occasions besides the first time after member registration.

[0099] For example, the hardware configuration shown in FIG. 2 is merely an example for achieving the object of the present invention, and is not particularly limited.

[0100] 3 is merely an example and is not particularly limited. That is, it is sufficient for the information processing device to have the function of executing the above-described series of processes as a whole, and the functional blocks and databases used by the information processing device to realize this function are not particularly limited to the example in FIG.

[0101] Furthermore, although the example in which the functional blocks and databases are arranged on one server has been described, the arrangement location is not limited to that shown in FIG. 3, and they may be distributed and arranged on a plurality of information processing devices, or any other suitable location. Furthermore, one functional block and database may be configured as a single piece of hardware, may be provided in separate pieces of hardware, may be configured as a single piece of software, or may be configured as a combination of these.

[0102] When the processing of each functional block is executed by software, the program that constitutes the software is read from a network or a recording medium into a computer such as a CPU of an information processing device. The computer may be a computer built into dedicated hardware, or may be implemented in an information processing device capable of executing various functions by loading various programs, such as a server, or a general-purpose smartphone or personal computer.

[0103] The recording medium containing such a program may not only be composed of removable media that is distributed separately from the device itself in order to provide the program to each user, but may also be composed of recording media that are provided to each user in a state where they are pre-installed in the device itself.

[0104] In this specification, the steps describing the program to be recorded on the recording medium include not only processes that are performed in chronological order, but also processes that are not necessarily performed in chronological order but are performed in parallel or individually.

[0105] In addition, in this specification, the term "system" refers to an overall device that is made up of a plurality of devices, a plurality of means, and the like.

[0106] In the above embodiment, an example of a service that suggests fashion styling has been described, but the present invention can be widely used in other fields as well. The mechanism of the present invention can also be applied to fields other than fashion, such as interior coordination and makeup advice, personalized services according to the preferences and characteristics of users.

[0107] In addition to styling, there are also systems that listen to customer feedback and requests and formulate styling policies for the next step, systems that can update user information based on the dialogue, and systems that formulate policies based on customer feedback and requests and then procure inventory. , and can be applied to various systems and services.

[0108] In the above embodiment, it was explained that the user information obtained from the user is stored in DB70 and then read from DB70 for use, but the user information stored in DB70 can be changed, and the changed user information can be converted into a slot.

[0109] Furthermore, if the user terminal 20 is, for example, a smartphone, a function for recognizing people, clothing, etc. can be added to the app, and user information can be obtained from a photograph (image) taken by the user that includes the entire body, thereby reducing the work required for the user to input information in the form of text.

[0110] In summary, the information processing apparatus to which the present invention is applied is sufficient if it has the following configuration, and can take on a variety of different embodiments. (1) That is, an information processing device to which the present invention is applied (for example, the server 10 in FIG. 3) In an information processing device (for example, the server 10 in FIG. 3) that supports the determination of a styling policy in fashion, information acquisition means (for example, the user information acquisition unit 11 in FIG. 3) for acquiring information about a user as user information; an information conversion means (for example, the user information conversion unit 12 of FIG. 3) for converting the user information into information in a format including a plurality of parameters related to a fashion styling policy; a data set generating means (for example, the slot generating unit 13 in FIG. 3) for generating a data set (for example, a slot) in which the user information is structured based on the user information whose format has been converted; an interpretability improving means (e.g., the interpretability improving unit 15 of FIG. 3 ) for detecting insufficiency and ambiguity of information in the data set (e.g., slot) using a dialogue system (e.g., the dialogue unit 14 of FIG. 3 ) and updating the data set (e.g., slot) through dialogue with the dialogue system (e.g., the dialogue unit 14 of FIG. 3 ), thereby improving the interpretability of the user information; a styling policy determination means (e.g., styling policy determination unit 16 in FIG. 3) that proposes a fashion styling policy to the user using the dialogue system (e.g., dialogue unit 14 in FIG. 3) based on the dataset (e.g., slots) with improved interpretability, and determines the styling policy while updating the dataset (e.g., slots) based on the user's response to the proposal; Equipped with. This allows us to resolve insufficient or ambiguous information through dialogue based on slots (datasets) that structure subjective information such as the user's "preferences" and "what suits them" as well as constraints, and determine styling guidelines that will satisfy the user. As a result, by using generative AI to agree on styling guidelines with the user before actually styling in fashion, it is possible to ultimately improve styling satisfaction.

[0111] (2) In the information processing device (for example, the server 10 in FIG. 3 ), The dialogue system (for example, the dialogue unit 14) uses natural language processing technology to carry out a dialogue with the user. By using natural language processing technology, users can converse in the natural language they normally use, enabling more intuitive and flexible information gathering and interpretation.

[0112] (3) In the information processing device (for example, the server 10 in FIG. 3), The dialogue system (for example, the dialogue unit 14) executes processing using a generation AI. In this way, by using a generative AI in a dialogue system (for example, the dialogue unit 14), it is possible to generate advanced dialogue that understands the context and to resolve the ambiguity that is unique to the fashion field, which requires complex judgments.

[0113] (4) In the information processing device (for example, the server 10 in FIG. 3), The user information is Information about the user's "preferences"; Information about what "looks good" to the user; Information that may be a limitation of the styling policy; Includes. By including the user's "preference" and "suitability" as restriction information, it is possible to suggest appropriate styling guidelines that the user himself / herself may not be aware of. This is particularly effective in eliminating ambiguity regarding the clothes that should be styled based on "preference" and "suitability."

[0114] (5) In the information processing device (for example, the server 10 in FIG. 3), The information conversion means (for example, the user information conversion unit 12 in FIG. 3) a parameterization means (for example, the parameterization unit 41 in FIG. 3) for quantifying the user's subjective evaluation; a conversion means (for example, the classification conversion unit 42 in FIG. 3) for converting at least one of the image and description information selected by the user into fashion elements (fashion taste and color classification); Includes. In this way, by converting the user's selection into numerical values ​​of subjective evaluation and fashion elements (such as fashion taste and color classification), the processing accuracy of the dialogue system (e.g., dialogue unit 14) is improved, and it becomes possible to take into account specialized fashion elements.

[0115] (6) In the information processing device (for example, the server 10 in FIG. 3), The interpretability improving means (for example, the interpretability improving unit 15 in FIG. 3) Insufficiency determination means (e.g., insufficiency determination unit 51 in FIG. 3) for determining the insufficiency of the data set (e.g., slot); Ambiguity determination means (e.g., the ambiguity determination unit 52 of FIG. 3) for determining the ambiguity of the data set (e.g., slot); A dialogue generating means (for example, the dialogue generating unit 53 of FIG. 3) that generates a dialogue for the user in accordance with the insufficiency and the ambiguity; a data set update unit (e.g., slot update unit 54 in FIG. 3) that sequentially updates the data set (e.g., slot) based on the interaction; Includes. This allows us to individually determine the insufficiency and ambiguity of a dataset and generate appropriate dialogues for each, thereby gradually and efficiently improving interpretability.

[0116] (7) In the information processing device (for example, the server 10 in FIG. 3), The styling policy determination means (for example, the styling policy determination unit 16 in FIG. 3) a styling policy suggestion unit (e.g., styling policy suggestion unit 61 in FIG. 3) that proposes the styling policy based on the data set (e.g., slots); response acquisition means (for example, the response acquisition unit 62 in FIG. 3) for acquiring the user's response to the proposal; A response type determination means (for example, the response type determination unit 63 in FIG. 3) for determining the type of the response (for example, agreement, disagreement, question, request); A dialogue generating means (for example, the dialogue generating unit 64 in FIG. 3) for generating a dialogue for the user in accordance with the type of response (for example, agreement, opposition, question, or request); a data set updating means (for example, the slot updating unit 65 in FIG. 3) for updating the data set based on information obtained through the interaction with the user; Includes. This allows the system to determine the type of user response (e.g., agreement, disagreement, question, request) and generate a dialogue accordingly, thereby determining a styling policy that is efficient and in line with the user's intentions.

[0117] (8) An information processing method executed by an information processing device (e.g., the server 10 of FIG. 3) for interactively determining a styling policy in fashion, an information acquisition step (for example, step S11 in FIG. 4) of acquiring information about a user as user information; an information conversion step (e.g., step S12 in FIG. 4) of converting the user information into information in a format including a plurality of parameters related to fashion styling principles; a data set generation step (e.g., step S13 in FIG. 4) of generating a data set (e.g., slots) that structures the user information based on the user information whose format has been converted; an interpretability improvement step (e.g., step S14 in FIG. 4 ) of detecting insufficient and ambiguous information in the data set using a dialogue system (e.g., dialogue unit 14 in FIG. 3 ) and updating the data set (e.g., slots) through dialogue with the dialogue system (e.g., dialogue unit 14 in FIG. 3 ) to improve the interpretability of the user information; a styling policy determination step (e.g., step S15 in FIG. 4) of proposing a fashion styling policy to the user using the dialogue system (e.g., the dialogue unit 14 in FIG. 3) based on the dataset (e.g., slots) with improved interpretability, and determining the styling policy while updating the dataset (e.g., slots) based on the user's response to the proposal; Contains In this way, by realizing it as a method executed by an information processing device (for example, the server 10 in FIG. 3), it becomes possible to use it on various platforms and systems, thereby improving practicality and scalability.

[0118] (9) Computers that interactively decide on styling policies in fashion. an information acquisition step (for example, step S11 in FIG. 4) of acquiring information about a user as user information; an information conversion step (e.g., step S12 in FIG. 4) of converting the user information into information in a format including a plurality of parameters related to fashion styling principles; a data set generation step (e.g., step S13 in FIG. 4) of generating a data set (e.g., slots) that structures the user information based on the user information whose format has been converted; an interpretability improvement step (e.g., step S14 in FIG. 4 ) of detecting insufficient and ambiguous information in the data set using a dialogue system (e.g., dialogue unit 14 in FIG. 3 ) and updating the data set (e.g., slots) through dialogue with the dialogue system (e.g., dialogue unit 14 in FIG. 3 ) to improve the interpretability of the user information; a styling policy determination step (e.g., step S15 in FIG. 4) of proposing a fashion styling policy to the user using the dialogue system (e.g., the dialogue unit 14 in FIG. 3) based on the dataset (e.g., slots) with improved interpretability, and determining the styling policy while updating the dataset (e.g., slots) based on the user's response to the proposal; The control process including the above is executed. In this way, by realizing the functions of an information processing device (for example, the server 10 in FIG. 3) as a program, it becomes possible to use it on various platforms and systems, thereby improving practicality and scalability. [Explanation of symbols]

[0119] 10...server, 20...user terminal, 101...CPU, 102...ROM, 103...RAM, 104...bus, 105...input / output interface, 106...output unit, 107...input unit, 108...storage unit, 109...communication unit, 110...drive, 111...removable media, 11...user information acquisition unit, 12...user information conversion unit, 13...slot generation unit, 14...interaction unit, 15...interpretability improvement unit, 16...styling policy determination unit, 41...parameterization unit, 42...classification conversion unit, 51...insufficiency determination unit, 52...ambiguity determination unit, 53...dialogue generation unit, 54...slot update unit, 61...styling policy proposal unit, 62...response acquisition unit, 63...response type determination unit, 64...dialogue generation unit, 65...slot update unit, 70...DB, NW...network

Claims

1. An information processing device for supporting decisions on fashion styling policies, information acquisition means for acquiring information about a user as user information; an information conversion means for converting the user information into information in a format including a plurality of parameters related to fashion styling principles; a data set generating means for generating a data set in which the user information is structured based on the user information whose format has been converted; an interpretability improvement means for detecting insufficiencies and ambiguities in the data set using a dialogue system and updating the data set through dialogue with the dialogue system to improve the interpretability of the user information; a styling policy determination means for proposing a fashion styling policy to the user using the dialogue system based on the dataset with improved interpretability, and determining the styling policy while updating the dataset based on the user's response to the proposal; An information processing device comprising:

2. The dialogue system uses natural language processing technology to dialogue with the user. The information processing device according to claim 1 .

3. The dialogue system performs the process using generative AI. The information processing device according to claim 2 .

4. The user information is Information about the user's "preferences"; Information about what "looks good" to the user; Information that may be a limitation of the styling policy; The information processing device according to claim 1 ,

5. The information conversion means parameterization means for quantifying the user's subjective evaluation; a conversion means for converting at least one of the image and description information selected by the user into a fashion element; The information processing device according to claim 1 ,

6. The interpretability improving means includes: an insufficiency determination means for determining the insufficiency of the data set; an ambiguity determination means for determining the ambiguity of the data set; a dialogue generating means for generating a dialogue with the dialogue system according to the insufficiency and the ambiguity; a dataset update unit that sequentially updates the dataset based on the interaction; The information processing device according to claim 1 ,

7. The styling policy determination means a styling policy suggestion means for suggesting the styling policy based on the data set; a response acquisition means for acquiring the response of the user; a response type determination means for determining the type of the response; a dialogue generation means for generating a dialogue with the dialogue system according to the type of response; a dataset updating means for updating the dataset in response to the interaction; The information processing device according to claim 1 ,

8. An information processing method executed by an information processing device for interactively determining styling policies in fashion, comprising: an information acquisition step of acquiring information about the user as user information; an information conversion step of converting the user information into information in a format including a plurality of parameters related to fashion styling principles; a data set generation step of generating a data set in which the user information is structured based on the user information whose format has been converted; an interpretability improvement step of detecting insufficient and ambiguous information in the dataset using a dialogue system and updating the dataset through dialogue with the dialogue system to improve the interpretability of the user information; a styling policy determination step of proposing a fashion styling policy to the user using the dialogue system based on the dataset with improved interpretability, and determining the styling policy while updating the dataset based on the user's response to the proposal; An information processing method including:

9. Computers that interactively decide on styling policies in fashion an information acquisition step of acquiring information about the user as user information; an information conversion step of converting the user information into information in a format including a plurality of parameters related to fashion styling principles; a data set generation step of generating a data set in which the user information is structured based on the user information whose format has been converted; an interpretability improvement step of detecting insufficient and ambiguous information in the dataset using a dialogue system and updating the dataset through dialogue with the dialogue system to improve the interpretability of the user information; a styling policy determination step of proposing a fashion styling policy to the user using the dialogue system based on the dataset with improved interpretability, and determining the styling policy while updating the dataset based on the user's response to the proposal; A program that executes control processing including:

Images