Information processing system, information processing method, and program

Abstract

To provide technology that supports companies' marketing activities. [Solution] An information processing system comprising: a VoC acquisition unit that acquires VoC data from customers; and an action generation unit that inputs the VoC data, labels assigned to the VoC data, and prompts including detailed product descriptions into a large-scale language model, and generates action hints for corporate activities based on the VoC data.

Description

Technical Field

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

Background Art

[0002] In Patent Document 1, an information collection processing system for segmenting customers using information such as customer purchase history information and behavior information such as website browsing history is disclosed. In Patent Document 1, it is described that a customer list segmented according to the purchase behavior history is used for marketing measures and branding measures.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Information such as purchase history and website browsing history as used in Patent Document 1 is relatively customer information that is easy for the business operator to obtain. However, it is difficult to grasp customer insights (true feelings, requests, motivations, etc.) only with such superficial information.

[0005] An object of an exemplary embodiment of the present disclosure is to provide a technology for supporting a company's marketing activities.

Means for Solving the Problems

[0006] An information processing system according to an aspect of the present invention a VoC acquisition unit that acquires VoC data from customers, The system includes an action generation unit that inputs the VoC data, labels assigned to the VoC data, and prompts including a detailed description of the product into a large-scale language model, and generates action hints for business activities based on the VoC data.

[0007] According to the present invention, it is possible to efficiently provide actionable hints that contribute to a company's marketing activities by utilizing customer voice data (VoC data: Voice of Customer) such as posts, reviews, word-of-mouth, testimonials, and survey responses submitted by the customers themselves. Further issues and solutions disclosed in this application will be clarified in the section on embodiments of the invention and in the drawings. [Brief explanation of the drawing]

[0008] [Figure 1] Figure 1 is a diagram illustrating the configuration of an information processing system according to one embodiment of the present disclosure. [Figure 2] Figure 2 is a block diagram illustrating the hardware configuration of the management server shown in Figure 1. [Figure 3] Figure 3 is a block diagram illustrating the functions of the management server shown in Figure 1. [Figure 4] Figure 4 is a flowchart showing an example of information processing performed by the information processing system shown in Figure 1. [Modes for carrying out the invention]

[0009] An information processing system according to one embodiment of this disclosure will be described with reference to the drawings. In the attached drawings, identical or similar elements are denoted by identical or similar reference numerals and names, and redundant descriptions of identical or similar elements will be omitted as appropriate in the description of the embodiment. The contents shown in each drawing are illustrative examples for explaining this embodiment and are only shown in a schematic manner to facilitate understanding. They may be modified or changed to the extent that it does not impede technical limitations.

[0010] <System Overview> The information processing system according to this embodiment (hereinafter referred to as "this system") is a system that generates action hints for corporate activities using customer voice data (VoC: Voice of Customer) collected about the user organization's products (goods and / or services).

[0011] In this embodiment, "VoC data" refers to all information relating to opinions, evaluations, experiences, questions, requests, and other expressive acts expressed by customers voluntarily or at their request, regardless of the medium, format, or acquisition route. The data format of VoC data may include data such as text, audio, images, and videos. In this system, labels are assigned to such VoC data as a preprocessing step. In this embodiment, "label" refers to metadata assigned to describe the context and meaning of VoC data, and may include semantic labels, service labels, sentiment labels, 5W1H labels, etc.

[0012] In this system, labeled VoC data as described above is fed to a large-scale language model (LLM), which generates action hints for business activities. In this embodiment, "action hints" mean suggestions that contribute to improving business activities or considering new measures. For example, this system may output suggestions for measures related to new product planning, existing product improvement, appeal improvement, usage patterns, service improvement, or at least one of category entry points. The details of this system will be described below.

[0013] As shown in Figure 1, the information processing system of this embodiment may include, for example, a management server 1, an LLM server 2, one or more corporate terminals 3, and one or more customer terminals 4. Each of these devices is connected to each other so as to be able to communicate via a network NW. In this embodiment, the network NW is mainly assumed to be the internet, but it may also be constructed using a public telephone network, a mobile phone network, a wireless communication network, Ethernet (registered trademark), etc. Furthermore, if necessary, external systems such as social networking services, billing systems, and membership management systems may be connected to the management server 1 and other components so as to be able to communicate via the network NW.

[0014] Management Server 1 is an information processing device managed by the service provider that provides services related to this system. Management Server 1 may be built on-premises using a general-purpose computer such as a workstation or personal computer, or it may be logically configured using cloud computing. Some of the functions of Management Server 1, described later, may be implemented as functions realized by the processor of other devices such as enterprise terminals 3, or they may be implemented in a distributed manner on external servers.

[0015] LLM Server 2 is a server that operates a Large Language Model (LLM) to provide services such as text generation and question answering. In this system, the LLM receives prompts and VoC data sent from Management Server 1 as input and executes the tasks instructed by the prompts. The processing results from the LLM are sent back to Management Server 1, which then outputs action hints to Enterprise Terminal 3 based on the LLM processing results. LLM Server 2 may be configured as a physical server installed in an on-premises environment, or as a virtual server built on a cloud computing service.

[0016] Enterprise terminal 3 is an information processing terminal of a user organization that uses the services of this system. Enterprise terminal 3 is used to set prompts and display processing results such as action hints. Enterprise terminal 3 is intended to be used by the system administrator or marketing personnel of the user organization, and may be a general-purpose computer such as a workstation or personal computer, or a mobile terminal such as a smartphone or tablet. In the description of this embodiment, when "user" is used, the "user" refers to the person in charge of the user organization who operates enterprise terminal 3.

[0017] Customer terminal 4 is an information processing terminal used by customers (general consumers) who purchase and / or use products provided by the user organization. Customer terminal 4 is used to input VoC data in predetermined data formats, such as free-response and multiple-choice formats, via media such as survey forms, membership apps, community tools, and social networking services (SNS). Customer terminal 4 can be any device, such as a personal computer, smartphone, or tablet, and is connected to management server 1 via a network (NW).

[0018] <Hardware Configuration> As shown in Figure 2, the management server 1 includes a processor 10, memory 11, storage 12, a transceiver 13, an input / output unit 14, etc., which are electrically connected to each other via a bus 17. Note that the configuration of the management server 1 shown is just one example, and the management server 1 may have a different configuration.

[0019] The processor 10 is a computing unit that controls the operation of the entire management server 1, controls the transmission and reception of data between each element, and performs information processing necessary for application execution and authentication processing. For example, the processor 10 is a CPU (Central Processing Unit) and / or a GPU (Graphics Processing Unit), and executes the program for this system stored in the storage 12 and loaded into the memory 11 to perform the information processing described later.

[0020] The memory 11 includes a main memory composed of a volatile storage device such as a DRAM (Dynamic Random Access Memory), and an auxiliary memory composed of a non-volatile storage device such as a flash memory or an HDD (Hard Disc Drive). The memory 11 is used as a work area of the processor 10, and stores a BIOS (Basic Input / Output System) executed when the management server 1 is started, and various setting information and the like.

[0021] The storage 12 stores various programs such as application programs for executing the information processing shown in this embodiment. A database storing data used for each information processing may be constructed in the storage 12. For example, the storage unit 120 described later may be provided in a part of the storage areas of the memory 11 and / or the storage 12.

[0022] The transmission / reception unit 13 is a communication interface for connecting the management server 1 to a network. The transmission / reception unit 13 may be provided with a short-range communication interface for Bluetooth (registered trademark) and BLE (Bluetooth Low Energy).

[0023] The input / output unit 14 is an input device such as a keyboard and a mouse, and an output device such as a speaker and a display. In the management server 1, the input device and the output device may be mounted on the management server 1 as individual devices, or a touch panel type display having both the functions of the input device and the output device may be mounted on the management server 1.

[0024] The bus 17 is commonly connected to the above elements, and transmits, for example, an address signal, a data signal, and various control signals.

[0025] Furthermore, the hardware configuration of other devices such as LLM Server 2, Enterprise Terminal 3, and Customer Terminal 4 is the same as that of Management Server 1. LLM Server 2, Enterprise Terminal 3, and Customer Terminal 4 are all equipped with processors, memory, storage, transceivers, input / output units, etc., and these are electrically connected to each other via a bus. These components are the same as those of Management Server 1's processor 10, memory 11, storage 12, transceiver 13, input / output unit 14, and bus 17, and therefore their explanation is omitted.

[0026] <Functions (Software Configuration) of Information Processing Systems> Figure 3 is a block diagram illustrating the functions implemented in this system. The management server 1 may include a VoC acquisition unit 101, a preprocessing unit 102, a prompt generation unit 103, a label assignment unit 104, an action generation unit 105, and an output unit 106. The various functional units of the management server 1 are realized by the processor 10 reading data and programs stored in the storage 12 and executing various programs in the working area of ​​the memory 11. The storage unit 120 of the management server 1 may also include an enterprise information storage unit 121, a prompt storage unit 122, and a VoC data storage unit 123. Each of these storage units stores data necessary for various information processing in this system and is configured to be accessible from each functional unit executed by the processor 10.

[0027] The enterprise information storage unit 121 stores information about the user organization. For example, the enterprise information storage unit 121 may store information about the products provided by the user organization (product name, service name, price, specifications, brand information, sales channel information (information about physical stores, information about e-commerce sites such as EC sites), detailed product description, etc.) associated with identification information (enterprise ID) to uniquely identify the user organization. When VoC data is acquired from external systems such as community tools (member apps, member sites, etc.), EC sites, or SNS, the enterprise information storage unit 121 may also store information about the medium from which the VoC data is acquired (for example, security information required for API integration (API key, secret key), API endpoint (connection URL), etc.) associated with the enterprise ID.

[0028] The prompt storage unit 122 stores prompts used in this system. For example, the prompt storage unit 122 may store prompt templates used for labeling VoC data. Labeling templates may be prepared for each attribute of the label to be assigned, such as semantic labels, service labels, sentiment labels, and 5W1H labels, and each template may include instructional information such as definitions of classification items to be labeled, judgment criteria corresponding to the classification items, output rules, and classification examples.

[0029] Furthermore, the prompt storage unit 122 may store prompt templates used for generating action hints. The action hint templates may include instruction information that specifies the details of the task for which the action hints should be generated, instruction information that specifies rules indicating the thinking guidelines when generating action hints, instruction information that specifies the output format, and information fields for inputting background information that provides a description of the target company or target product.

[0030] The VoC data storage unit 123 stores VoC data collected by the VoC acquisition unit 101. The VoC data may be accompanied by identification information to uniquely identify each data item. The VoC data storage unit 123 can also store VoC data imported in CSV format.

[0031] The VoC data storage unit 123 may store the original VoC data at the time of collection and the normalized data after preprocessing in association with each other. Each VoC data may be associated with information to identify the originating customer, and such information may include the customer account identifier in the acquisition system, the type of the identifier, and the type of media from which the data was acquired. Furthermore, metadata such as the collection time, acquisition name, language, character code, and flags related to data quality (presence or absence of missing data, results of character encoding detection, etc.) may be added and managed. From the perspective of privacy protection, the system may be configured to store highly personally identifiable attributes in accordance with prescribed regulations, by masking or pseudo-naming them.

[0032] Furthermore, the VoC data storage unit 123 may store data indicating labels assigned by the labeling unit 104 (described later) in association with each VoC data, or it may store identification information of the generated action hints in association with the VoC data that served as the basis for the action hint ideas. In addition to the database described above, the storage unit 120 stores information used for information processing in this system. For example, the storage unit 120 may have a database for storing generated action hints.

[0033] The VoC acquisition unit 101 acquires VoC data from customers. The acquisition route for VoC data is not particularly limited. For example, the VoC acquisition unit 101 may acquire VoC data from external systems such as community tools (member apps, member sites, etc.), e-commerce sites, survey systems, and social networking services by using technologies such as API integration.

[0034] For example, when acquiring VoC data from community tools such as membership sites, VoC data such as product reviews and testimonials posted on the tool during a specified period may be acquired via API integration. Alternatively, when collecting VoC data from social media, posts containing the name of the product provided by the user organization may be extracted and acquired using that keyword, or posts made to the user organization's official account may be acquired.

[0035] In addition to acquiring VoC data from external systems as described above, the VoC acquisition unit 101 may also acquire VoC data collected by the user organization from the corporate terminal 3. In this case, the VoC acquisition unit 101 may acquire VoC data by presenting a graphical user interface (hereinafter referred to as GUI) for data import to the corporate terminal 3 and accepting input of the VoC data collected by the user organization via the GUI.

[0036] The VoC acquisition unit 101 may acquire VoC data from multiple sources. Furthermore, when acquiring VoC data, it may acquire supplementary information such as information indicating the source medium, the date and time of posting the VoC data, and information identifying the customer who posted the VoC data (e.g., customer identification information or account information on community tools, e-commerce sites, social media, etc.) and link this information to the VoC data. The acquired VoC data is recorded in the VoC data storage unit 123 described above.

[0037] The preprocessing unit 102 performs rule-based preprocessing on the collected VoC data. For example, as a preliminary step before performing labeling, the preprocessing unit 102 may perform denoising, normalization, or a combination thereof, on the VoC data. Denoising may involve removing or neutralizing non-verbal elements that hinder the interpretation of LLM text, such as emojis, URLs, hashtags, line break tags, special symbols, and machine-dependent characters included in SNS and community posts. It may also include removing unnecessary whitespace, redundant sentences, and formulaic greetings.

[0038] The preprocessing unit 102 may perform normalization processing on the text after noise removal. The normalization processing may include a process of unifying different notations having the same meaning. For example, normalization of notation fluctuations such as unifying "出来る", "できる", "出来ます", and "できます" to "できる", or normalization of converting synonyms such as "美味しい", "美味", and "うまい" into representative terms may be performed. Further, it may include a process of unifying alphabetic notations, normalizing full-width and half-width characters, and converting symbols such as emojis representing emotional expressions into word phrases (for example, emotional words such as "嬉しい", "悲しい", and "驚いた"). Thereby, the vocabulary system of the input sentence is standardized, and the accuracy of semantic analysis by the LLM is improved.

[0039] Furthermore, the normalization processing executed by the preprocessing unit 102 may include a process of converting the text of the VoC data into a predetermined data structure. In the normalization processing, for example, the VoC data is divided into sentence units or paragraph units, or divided into word units, and converted into a predetermined data structure defined in advance. For example, it may be converted into structured data such as JSON format having a key-value pair. In this case, "customer ID", "utterance content", "utterance date and time", "media type", "language", "sentence number", etc. may be used as keys, and corresponding values may be stored respectively.

[0040] Also, the preprocessing unit 102 may be provided with validation processing for checking data quality. For example, when the input data includes a sentence with an extremely small number of characters or a sentence whose language identification is difficult, the data may be excluded or output as a different category. The preprocessing result is stored in the VoC data storage unit 123 in the storage unit 120 of the management server 1, and is held in a state that can be referred to by other functional units such as the subsequent labeling unit 104 and action generation unit 105.

[0041] The prompt generation unit 103 generates prompts to be input to the Large-Scale Language Model (LLM). In this embodiment, a "prompt" is instruction information that instructs the execution of a predetermined task, such as labeling or action plan generation. The prompt template may define a set of instructions that includes the task's purpose and definition, judgment criteria, output format, constraints, and contextual information that serves as the premise for the task. The prompt generation unit 103 may acquire a description of the user organization and / or product (prerequisite information) and analysis setting information (target, label type, output format, etc.) input from the enterprise terminal 3, and dynamically construct a case-specific prompt by applying these settings to a templated form.

[0042] For example, in a labeling prompt, the definition of the label classification items is set. Labels are not necessarily limited, and examples of labels that can be used to interpret each VoC data are given, such as semantic labels, service labels, sentiment labels, and 5W1H labels. Semantic labels are labels used to classify the content of statements or descriptions contained in VoC data based on the purpose or intent of the statement. For example, they may be defined as a configuration that includes usage, motivation for use, requests, complaints, recommendations, emotions, and other classification items, and the description of each classification item may be written in natural language in the task definition part of the prompt. Service labels are labels used to identify product names, service names, or functions mentioned in VoC data. For example, classification items may be defined for each product provided by the user organization, or classification items may be defined for each function in the product or service. Sentiment labels are labels that indicate the customer's sentiment expression contained in VoC data. The emotional label classifications could be, for example, a three-level classification of positive, negative, and neutral, or they could be configured to output multiple values ​​by pre-defining specific emotional categories such as joy, anticipation, surprise, anger, disgust, fear, disappointment, relief, and confusion.

[0043] 5W1H labels are labels that annotate the content of customer statements shown in VoC data from six perspectives: When, Where, Who, What, Why, and How. Prompts that instruct the assignment of 5W1H labels may include, for example, basic instructions for the labeling (post classification) task, definitions of 5W1H, and instructions specifying the criteria for each 5W1H item. Other information that may be included is output rule information such as how to write the labels, and reference information that is helpful when performing the task, such as output examples. The criteria for judgment are not necessarily limited. For example, the criteria may include: "For 'what,' extract the product / service name first, and if there is no name, extract the subject"; "For 'when,' extract the scene rather than the time of day"; "For 'where,' emphasize the specificity of the location"; "For 'who,' extract the single most important person"; "For 'how,' describe how it is used"; and "For 'why,' indicate the reason for choosing that product / service."

[0044] In a prompt that instructs the generation of action hints, the category of the action hints to be generated may be specified in the task definition. Examples of categories include new product planning, existing product improvement, appeal improvement, usage patterns, service improvement, or category entry point. The prompt may instruct the generation of action hints by selecting one of the above categories, or it may define each category in the prompt to instruct the generation of action hints belonging to at least one category.

[0045] The prompt generation unit 103 may, when generating prompts, receive input from the user via the corporate terminal 3, such as the user organization's business content and product description, and reflect this user input information in the prerequisite information field of the prompt template to generate the prompt. Regarding the rules for generating action hints and output rules, rules predefined in this system may be applied, or prompt proposals may be presented to the user, allowing the user to edit the rules.

[0046] As described above, the various instruction information to be set in the prompt may be defined in a pre-prepared template, or instruction information such as task definitions and output rules (rules for classification, rules for hint generation, rules for defining output format, etc.) that are defined by default may be presented to the enterprise terminal 3, and the user may be allowed to edit the instruction information. The prompt generation unit 103 completes the prompt to be given to the LLM by applying the acquired VoC data, such as by attaching it to the prompt template. At this time, the prompt generation unit 103 may present a list of VoC data stored in the VoC data storage unit 123 to the enterprise terminal 3, and allow the user to select the VoC data to be analyzed.

[0047] The labeling unit 104 performs the process of assigning labels to the VoC data. Specifically, the labeling unit 104 inputs the prompts generated by the prompt generation unit 103 and the VoC data into a large-scale language model (LLM), and causes the LLM to perform the labeling process for each VoC data. The types of labels to be assigned are, as mentioned above, semantic labels, service labels, sentiment labels, 5W1H labels, etc., and are not necessarily limited. One type of label from the above label types may be assigned, or two or more types of labels may be assigned to each VoC data.

[0048] The labeling unit 104 may assign multiple types of labels simultaneously with a single request (prompt), or it may instruct classification for only a specific label type with a single request (prompt), or it may switch prompts for each label type and instruct sequential labeling in multiple steps. Parallel or sequential execution of processing is selected according to operational policies and load conditions, and in any method, the label assignment results for each VoC data are acquired and saved in a predetermined format. Specifically, the label assignment results are recorded in association with each VoC data. For example, the labeling unit 104 may perform a process to associate labeling results such as assigned labels, confidence levels, and generation dates with each VoC data and store them in the VoC data storage unit 123 within the storage unit 120.

[0049] The action generation unit 105 executes a process to generate action hints for corporate activities based on VoC data. Specifically, the action generation unit 105 inputs the VoC data, the labels assigned to the VoC data, and prompts including detailed descriptions of the product into a large-scale language model, causing it to generate action hints for corporate activities based on the VoC data. The action hints are, for example, text that describes suggestions that will be useful for improving corporate activities or considering new measures, based on the facts expressed in the VoC data. For example, the action hints may include at least some of the direction of the measures, the target of improvement, the expected effects, and points to note.

[0050] The action generation unit 105 sends prompts and input information such as VoC data to the LLM and receives a response from the LLM. The action generation unit 105 analyzes the response text and formats it into a predetermined output format, and may perform formatting processes such as summarization, normalization, notation standardization, and proper noun identification as needed. The action generation unit 105 may provide multiple VoC data to the LLM at once to generate a single action hint, or it may generate multiple action hints. The VoC data input to the LLM may be classified into clusters that share at least one type of label and input to the LLM cluster by cluster, or VoC data from multiple clusters may be input to the LLM together.

[0051] The action generation unit 105 may also instruct the LLM to list the VoC data that served as the basis for the idea of ​​the action hint, along with the action hint itself. In this case, the action generation unit 105 sends a prompt to the large-scale language model instructing it to identify the VoC data that formed the basis of the generated action hint and include it in the output. Based on the prompt's instructions, the LLM generates an action hint based on at least one of the input VoC data, and then sends back to the action generation unit 105 the action hint along with information identifying the VoC data referenced for generating the idea. This makes it possible to verify the validity and reliability of the action hint.

[0052] The action generation unit 105 may output at least the action hint text to the LLM, and may also generate output in a format that includes the target product name, idea category, title, summary, and VoC data on which the idea originated. These target product name, idea category, title, summary, and VoC data on which the idea originated (founded) may be linked to the action hint as attribute information.

[0053] The categories of action hints are generated based on the definitions specified in the prompt. For example, these categories may include new product planning, existing product improvement, appeal improvement, usage patterns, service improvement, or category entry point. New product planning is a category of action hints related to the development of new products. Existing product improvement is a category of action hints related to improving the functionality and quality of existing products. Appeal improvement is a category of action hints related to improving product advertising and promotion. Usage patterns is a category of action hints related to understanding how customers use products. Service improvement is a category of action hints related to improving customer service. Category entry point is a category of action hints related to entering new markets.

[0054] The action generation unit 105 may generate action hints for each category defined in the prompt, or it may generate action hints for any one of the categories. Furthermore, a category may be uniquely selected for a single action hint, or multiple categories may be assigned to a single action hint.

[0055] When generating an action hint summary, the action generation unit 105 may distinguish between facts (background) read from the VoC data and implications (reasoning for the suggestion) based on those facts when describing the summary. The format of the summary description (output) is also defined by prompts, similar to the categories. The LLM outputs a summary to be associated with the action hint according to the instructions in the prompts.

[0056] The action generation unit 105 may be equipped with quality control functions to ensure the validity of the generation results. For example, it may be configured to calculate a confidence index based on the degree of compliance with constraints in the prompt, the consistency of the VoC data cited as evidence, and the results of forbidden word detection, and to re-execute or re-format the action if the index is below a threshold. Furthermore, if similar action hints are generated from multiple VoC data, it may perform synonym integration or duplicate removal and select a representative option. In addition, it may be configured to maintain an append / edit history of the generation results and to reflect the edited content in subsequent prompts in an adaptive operation.

[0057] The action generation unit 105 stores the generated action hints in the storage unit 120. The output unit 106 transmits the generated action hints to the enterprise terminal 3. The output format of the action hints is not particularly limited and may be presented in a format viewable on the enterprise terminal 3. The output unit 106 may also generate HTML (HyperText Markup Language) data for displaying the action hints as a web page and transmit it to the enterprise terminal 3. Alternatively, the output unit 106 may provide the action hints in a downloadable format such as CSV or PDF (Portable Document Format). When outputting action hints, the output unit 106 may present the action hint text along with the VoC data that serves as the basis for the action hints in a manner that allows reference.

[0058] Figure 4 is a flowchart illustrating an example of information processing in this system.

[0059] First, the VoC acquisition unit 101 acquires VoC data (step S1). Specifically, customer posts may be collected from community tools, e-commerce sites, social networking services, etc., through API integration with external systems, or a GUI for data import may be presented to the corporate terminal 3, allowing the user organization to acquire the collected VoC data via the GUI. The VoC acquisition unit 101 records the acquired VoC data in the VoC data storage unit 123.

[0060] Next, the preprocessing unit 102 performs preprocessing such as noise reduction and / or normalization (step S2). The data after preprocessing is stored in the VoC data storage unit 123 and can be referenced from subsequent processing. Next, the labeling unit 104 inputs a prompt containing labeling instruction information and the VoC data into the large-scale language model (LLM) to assign a label to each VoC data (step S3). The prompt used to instruct labeling may be generated by the prompt generation unit 103, and labeling may be performed on each VoC data targeting one or more of semantic labels, service labels, sentiment labels, and 5W1H labels. The labeling results are stored in the VoC data storage unit 123, linked to each VoC data.

[0061] Next, the prompt generation unit 103 generates a prompt that defines an action hint generation task (step S4). This prompt may include instruction information that instructs the task to generate action hints by selecting a category from at least one of the following: new product planning, existing product improvement, appeal improvement, usage patterns, service improvement, or category entry point. The prompt generation unit may also accept a selection of a category to be generated from the user and generate a prompt instructing the generation of action hints for the specified category. When generating this prompt, the prompt generation unit 103 may accept input from the user describing the user organization's business and / or a product description, and generate the prompt by applying the user's input to a predetermined template.

[0062] The action generation unit 105 inputs this prompt and the labeled VoC data into the LLM to generate an action hint (step S5). The action generation unit 105 may also cause the LLM to generate an action hint and attribute information such as the target product name, idea category, title, summary, and the VoC data on which the idea originated. The action generation unit 105 links the action hint with the attribute information described above and registers it in the storage unit 120. The output unit 106 transmits the action hint generation result to the enterprise terminal 3 (step S6). When outputting, the action hint text and attribute information may be formatted into a predetermined display format and transmitted to the enterprise terminal 3 as a display screen that allows viewing of a list of the VoC data on which the idea originated or a link to the VoC data.

[0063] Note that the information processing method flow shown in Figure 4 is merely an example, and steps may be added, deleted, modified, or rearranged as appropriate.

[0064] As described above, the information processing system of this embodiment can acquire VoC data from customers from diverse data sources, structure it through preprocessing and labeling, and then automatically generate concrete action hints for corporate activities using a large-scale language model. This system can analyze VoC data in detail through multifaceted labeling such as semantic labels, 5W1H labels, service labels, and sentiment labels, and generate classified action hints such as new product planning, existing product improvement, appeal improvement, usage patterns, service improvement, and category entry points. Furthermore, the generated action hints clearly indicate the VoC data that underlies them, allowing for verification of the validity and reliability of the proposals. In addition, the action hints are output in a structured format including product name, idea category, title, and summary, enabling corporate marketing personnel to efficiently utilize the action hints.

[0065] The embodiments described above are merely illustrative to facilitate understanding of this disclosure and are not intended to limit it. This disclosure may be modified and improved without departing from its intent, and its equivalents are included.

[0066] For example, the processing performed by each functional unit of the management server 1 described above may be executed by any of the functional units. Furthermore, different functional units may be added to perform some of the processing performed by each of the functional units described above. Also, the functional units of the management server 1 may be distributed across multiple computers.

[0067] Furthermore, the information stored in each memory unit of the management server 1 may be stored in any of the memory units. That is, the information stored in the multiple memory units mentioned above may be stored in a single memory unit, or a portion of the information stored in one memory unit may be stored in another memory unit.

[0068] <Example 1> In the embodiment described above, an example of generating action hints using a large-scale language model was shown, but it is also possible to use multiple large-scale language models in combination. In this case, the action generation unit 105 generates multiple action hint candidates using the first large-scale language model, and then uses the second large-scale language model to evaluate the quality of each candidate and select the optimal action hint. Specifically, the first large-scale language model is input with a prompt containing VoC data and label information to generate five action hint candidates, and the second large-scale language model is evaluated from the perspectives of feasibility, effectiveness, and cost, and the candidate that receives the highest evaluation is output as the final action hint.

[0069] <Modification 2> In the embodiment described above, an example was shown in which the action generation unit 105 generates action hints with a single prompt. However, it is also possible to generate action hints by sending prompts in stages. In this case, the action generation unit 105 generates a summary of VoC data in the first stage, extracts issues in the second stage, and generates specific action hints in the third stage. Specifically, in the first stage, it sends a prompt that says, "Please summarize the following VoC data," in the second stage, it sends a prompt that says, "Please extract three customer issues from this summary," and in the third stage, it sends a prompt that says, "Please propose specific action hints from the perspective of product improvement to solve these issues."

[0070] <Variation 3> In the embodiment described above, an example was shown in which the action generation unit 105 classifies action hints into predetermined categories. However, categories can also be dynamically generated according to the content of the VoC data. In this case, the action generation unit 105 first has a large-scale language model analyze the VoC data and generate appropriate category names. Specifically, it sends a prompt saying, "Please suggest the most suitable category name for the action hints derived from the following VoC data. If it does not fall under an existing category (new product planning, existing product improvement, appeal improvement, actual usage, service improvement, category entry point), please create a new category name," and then classifies the action hints using the generated category names.

[0071] <Modification 4> In the embodiment described above, an example was shown in which the output unit 106 sends action hints to the enterprise terminal 3. However, it is also possible to assign priorities to the generated action hints before outputting them. In this case, the action generation unit 105 has a large-scale language model evaluate each action hint from the perspectives of feasibility, magnitude of effect, and urgency, and calculates an overall priority score. Specifically, it sends a prompt that says, "Evaluate the following action hints in terms of feasibility (1 to 5 points), magnitude of effect (1 to 5 points), and urgency (1 to 5 points), and calculate an overall score," and based on the obtained scores, the action hints are sorted in order of priority and sent to the enterprise terminal 3 by the output unit 106.

[0072] <Modification 5> This system may include a clustering unit 110 that performs similarity calculations based on the label assignment results. The clustering unit 110 constructs clusters using the distribution of 5W1H labels, semantic labels, and sentiment labels as features, and sorts the VoC data into each cluster. In this case, the action generation unit 105 may generate action hints on a cluster-by-cluster basis, and may associate a representative VoC with a representative action hint for each cluster. The action generation unit 105 may also perform cluster-by-cluster summaries, add annotations such as "candidates for standardization" and "unique needs," and prioritize by category.

[0073] The information processing system described herein may be implemented as a single device, or it may be implemented by multiple devices, some or all of which are connected by a network. For example, the functions of the processor 10 and storage 12 of the management server 1 may be implemented by different servers connected to each other by a network.

[0074] Furthermore, the series of processes performed by the information processing system described herein may be implemented using software, hardware, or a combination of software and hardware. It is also possible to create a computer program to implement each function of the management server 1 according to this embodiment and implement it on a PC or the like. A computer-readable recording medium on which such a computer program is stored can also be provided. Examples of recording media include magnetic disks, optical disks, magneto-optical disks, flash memory, etc. In addition, the above-mentioned computer program may be distributed, for example, via a network, without using a recording medium.

[0075] Furthermore, the effects described herein are merely descriptive or illustrative and not limiting. In other words, the technology relating to this disclosure may produce other effects that will be apparent to those skilled in the art from the description herein, in addition to or in lieu of the effects described herein.

[0076] The information processing system, information processing method, and program disclosed herein have, for example, the following configuration. [Item 1] A VoC acquisition unit that acquires VoC data from customers, An action generation unit inputs the aforementioned VoC data, labels assigned to the VoC data, and prompts including detailed product descriptions into a large-scale language model, and generates action hints for corporate activities based on the VoC data. An information processing system equipped with the following features. [Item 2] The information processing system described in item 1, wherein the action generation unit generates the action hint along with the category to which the action hint belongs. [Item 3] The aforementioned category is an information processing system as described in item 2, which is at least one of the following: new product planning, existing product improvement, appeal improvement, usage data, service improvement, or category entry point. [Item 4] The information processing system described in item 1, wherein the action generation unit lists the VoC data that served as the basis for the idea of ​​the action hint, along with the action hint. [Item 5] The information processing system described in item 1, wherein the action generation unit generates output in a format that includes product name, idea category, title, summary, and VoC data on which the idea is based. [Item 6] The summary includes, clearly separating the facts read from the VoC data from the proposed content derived based on those facts, as described in item 5 of the information processing system. [Item 7] Steps to obtain VoC data from customers, The steps include inputting the VoC data, the labels assigned to the VoC data, and prompts including detailed product descriptions into a large-scale language model to generate action hints for business activities based on the VoC data, Information processing methods including [Item 8] The computer is used to acquire VoC data from customers. The VoC data, the labels assigned to the VoC data, and prompts including detailed product descriptions are input into a large-scale language model, and action hints for business activities are generated based on the VoC data. program. [Explanation of symbols]

[0077] 1. Management Server 2 LLM Servers 3. Enterprise terminals 4 Customer terminals 101 VoC acquisition department 105 Action Generation Unit

Claims

1. A VoC acquisition unit that acquires VoC data from customers, A prompt that instructs a labeling task for the VoC data, including label definitions and output rules; a labeling unit that inputs the VoC data into a large-scale language model and assigns multiple types of labels to the VoC data that describe the context and / or meaning of the VoC data; An action generation unit inputs the VoC data, the labels assigned to the VoC data, and prompts instructing the task of generating action hints, including a detailed description of the product, into a large-scale language model, and generates action hints for business activities based on the VoC data. An information processing system equipped with the following features.

2. The information processing system according to claim 1, wherein the action generation unit generates the action hint along with the category to which the action hint belongs.

3. The information processing system according to claim 2, wherein the category is at least one of new product planning, existing product improvement, appeal improvement, usage patterns, service improvement, or category entry point.

4. The information processing system according to claim 1, wherein the action generation unit lists the VoC data that served as the basis for the idea of ​​the action hint, along with the action hint.

5. The information processing system according to claim 1, wherein the action generation unit generates output in a format that includes a product name, idea category, title, summary, and VoC data that formed the basis of the idea for the action hint.

6. The information processing system according to claim 5, wherein the summary clearly separates the facts read from the VoC data from the proposed content obtained based on those facts.

7. Steps to obtain VoC data from customers, A prompt instructing a labeling task for the VoC data, including label definitions and output rules; a step of inputting the VoC data into a large-scale language model and assigning multiple types of labels to the VoC data that describe the context and / or meaning of the VoC data; The steps include inputting the VoC data, the labels assigned to the VoC data, and prompts instructing the task of generating action hints, including a detailed description of the product, into a large-scale language model, and generating action hints for business activities based on the VoC data; A method of information processing performed by a computer.

8. The computer is used to acquire VoC data from customers. A prompt instructing a labeling task for the VoC data, including label definitions and output rules; a step of inputting the VoC data into a large-scale language model and assigning multiple types of labels to the VoC data that describe the context and / or meaning of the VoC data; The VoC data, the labels assigned to the VoC data, and prompts instructing the task of generating action hints, including a detailed description of the product, are input to a large-scale language model, and action hints for business activities are generated based on the VoC data. program.

Images