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

Abstract

To provide an information processing device, an information processing method, and an information processing program capable of further appropriately supporting optimization of a prompt.SOLUTION: An information processing device according to the present application includes an acquisition unit and an estimation unit. The acquisition unit acquires evaluation information being information indicating evaluation for a service using generation information generated by using generative AI on the basis of a prompt containing a plurality of items of information. The estimation unit estimates importance of a plurality of items on the basis of the evaluation information acquired by the acquisition unit.SELECTED DRAWING: Figure 3

Description

Technical Field

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

Background Art

[0002] In recent years, techniques for generating information using generative AI (Artificial Intelligence) have been known. For example, Patent Document 1 discloses a technique using large language models (LLMs) that generates a prompt by adding a valid sentence as an additional sentence to an input question sentence within a determined character limit range. A prompt is information input to generative AI and indicates, for example, instructions or requests given to generative AI to perform a specific task.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the above prior art, although the large language model into which the user's question sentence is input can generate additional sentences, there is room for improvement in more accurately optimizing the prompt.

[0005] The present application has been made in view of the above, and an object thereof is to provide an information processing apparatus, an information processing method, and an information processing program that can more accurately support the optimization of a prompt.

Means for Solving the Problems

[0006] The information processing apparatus according to the present application includes an acquisition unit and an estimation unit. The acquisition unit acquires evaluation information, which is information indicating an evaluation of a service using generated information generated by a generation AI based on a prompt including information on a plurality of items. The estimation unit estimates the importance levels of the plurality of items based on the evaluation information acquired by the acquisition unit.

Effect of the Invention

[0007] According to one aspect of the embodiment, there is an effect that it is possible to more accurately support the optimization of the prompt.

Brief Description of the Drawings

[0008]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Modes for Carrying Out the Invention

[0009] Hereinafter, embodiments for implementing the information processing apparatus, information processing method, and information processing program according to the present application (hereinafter referred to as "embodiments") will be described in detail with reference to the drawings. Note that the information processing apparatus, information processing method, and information processing program according to the present application are not limited by this embodiment. Also, in the following embodiments, the same parts are denoted by the same reference numerals, and redundant explanations are omitted.

[0010] [1. Example of Information Processing] First, with reference to FIG. 1, an example of information processing according to an embodiment will be described. FIG. 1 is a diagram for explaining the information processing according to the embodiment.

[0011] The information processing apparatus 1 shown in FIG. 1 is an information processing apparatus that cooperates with each terminal device 2 of the user U and provides various types of information to the user U online, and is realized by, for example, one or more servers or a cloud system. The terminal device 2 is, for example, a smartphone, a tablet, or a personal computer.

[0012] As shown in FIG. 1, the information processing apparatus 1 determines each of a plurality of prompts each including information of a plurality of items with different combinations as a target prompt (step S1). Each of the plurality of items is an item indicating the type of a key item or a characteristic item, and is, for example, an item indicating a semantic type such as a semantic field. Further, the information of the item is information indicating the content of the key item or information indicating the content of the characteristic item.

[0013] The target prompt determined in step S1 is information input to a generation AI (Artificial Intelligence) described later in order to generate provided information that is information provided to the user U in the service provided by the information processing apparatus 1. In the following, although it is described that the service provided by the information processing apparatus 1 is an advertisement distribution service, the service provided by the information processing apparatus 1 is not limited to the advertisement distribution service.

[0014] When the service provided by the information processing apparatus 1 is an advertisement distribution service, the prompt determined in step S1 is information input to a generation AI described later in order to generate advertisement content. The advertisement content is content including at least one of, for example, text and an image.

[0015] The plurality of target prompts determined in step S1 are prompts regarding the same target, but they may also be prompts regarding different targets, or may be a plurality of prompts regarding partially different targets. For example, the plurality of target prompts determined in step S1 are prompts for generating advertisement content including link information of a specific landing page, but may also be prompts for generating advertisement content for each of a plurality of mutually different landing pages.

[0016] The prompt input to the generation AI described later for generating advertisement content is, for example, the information of the character string "Please create advertisement content for a newly opened organic coffee shop featuring a comfortable interior. Make the design of the advertisement content target the local community."

[0017] In this case, the information of the plurality of items is the information of the character string "comfortable interior", the information of the character string "newly opened", the information of the character string "organic coffee shop", the information of the character string "local community", etc.

[0018] The information of the character string "comfortable interior" is the information of the item "design inside the store", the character string "newly opened" is the item "status of the store", the information of the character string "organic coffee shop" is the information of the item "category of the store", and the information of the character string "local community" is the information of the item "residence of the target user". The information of the item "residence of the target user" is an example of the information indicating the attributes of the target user U.

[0019] Note that the information of the items is not limited to the above example, and may be, for example, the information of the item "location of the store", the information of the item "event", the information of the item "coupon or discount", the information of the item "product price", the item "age group of the target user", the item "interests of the target user", etc.

[0020] Also, the information of the items may be, for example, the information of the item "tone", the information of the item "prompt format", etc. The information of the item "tone" is, for example, information such as a direct command like "Do it" or an indirect command like "Please do it". The item "prompt format" is, for example, information such as zero-shot, one-shot, few-shot, etc.

[0021] Also, in the above example, the advertisement content of a physical store was used as an example for explanation, but the advertisement content is not limited to the advertisement content of a physical store. For example, the advertisement content may be the advertisement content of a virtual store on an e-commerce site, the advertisement content of a physical store or a product sold in a store, or other advertisement content.

[0022] When the information processing device 1 receives, for example, a prompt determination request transmitted from a business operator terminal that is a terminal device of an advertiser, it extracts the information of a plurality of items from the information included in the prompt determination request. Then, the information processing device 1 determines a plurality of prompts each including the information of two or more items with different combinations among the extracted information of the plurality of items as the above-mentioned plurality of target prompts.

[0023] The information processing device 1 has, for example, dictionary information including a plurality of terms for each item, and can extract the information of a plurality of items from the information included in the prompt determination request using such dictionary information. Also, the information processing device 1 can also extract the information of a plurality of items from the information included in the prompt determination request using a text generation AI such as a large language model.

[0024] In addition, for example, the information processing apparatus 1 can transmit template information for prompting a user to input or select information of a plurality of predetermined items to a business operator terminal, which is an advertiser's terminal device, and display the template information on the business operator terminal. In this case, the advertiser can input or select information of a plurality of items by operating the business operator terminal, and thereby, a prompt determination request including information of a plurality of items is transmitted from the business operator terminal to the information processing apparatus 1.

[0025] The prompt determination request transmitted from the business operator terminal may include information specifying a landing page targeted by the advertisement. In this case, the information processing apparatus 1 can extract information of a plurality of items from the landing page specified in the prompt determination request using the above-described dictionary or generative AI, etc. The information processing apparatus 1 determines a plurality of prompts each including information of two or more items having different combinations among the extracted information of the plurality of items as the above-described plurality of target prompts.

[0026] Note that the target prompt may be, for example, a prompt generated using generative AI. Also, the target prompt is not limited to a text prompt, and may include, for example, image data in addition to the text prompt.

[0027] In this case, in the target prompt, the text prompt is, for example, information of the character string "Please create advertisement content for a newly opened organic coffee shop featuring a comfortable interior based on the input image data.", and the image data is data indicating an image of the advertisement content to be referred to.

[0028] The information of the items included in the target prompt is not limited to the information of the items included in the text prompt. For example, it may include information indicating the type of image shown by the image data or information of the items included in the image shown by the image data. The information of the items included in the image shown by the image data is, for example, an object, a scene, an emotion, a color, etc. shown by the image shown by the image data, but is not limited to such examples.

[0029] Note that when a plurality of prompts are included in the prompt determination request, the information processing apparatus 1 can also determine the prompts included in the prompt determination request as a plurality of target prompts.

[0030] Subsequently, the information processing apparatus 1 inputs, as input information to the generation AI, the information including the target prompt which is a prompt including the information of the plurality of items determined in step S1, and performs, for each target prompt determined in step S1, a process of causing the generation AI to generate generation information corresponding to the input information (step S2).

[0031] The generation AI is, for example, a text generation AI, an image generation AI, or a multimodal generation AI. The text generation AI is, for example, a large language model trained to estimate and output the next token from the input token sequence, and is, for example, a Transformer-based model, an RNN (Recurrent Neural Network)-based model, etc., but may also be a hybrid model thereof. Further, the text generation AI may be a composite system combined with an identification mechanism for preventing unauthorized use.

[0032] The Transformer-based model is, for example, GPT (Generative Pre-trained Transformer), PaLM2 (Pathways Language Model Version 2), etc., but is not limited to such examples. The RNN-based model is, for example, RWKV (Receptance Weighted Key Value), etc., but is not limited to such examples.

[0033] An image generation AI is an AI that generates images from text, such as, for example, StackGAN (Generative Adversarial Networks), AttnGAN, T2I (Text-to-Image) with Transformers, Diffusion models, etc., but is not limited to such examples. Examples of Diffusion models include DALL-E and Stable-Diffusion.

[0034] A multimodal generation AI is, for example, a generation AI that generates at least one of text, images, and audio from at least one of text, images, and audio. Multimodal generation AIs include, for example, GPT-4 Turbo with vision, gemini, CM3Leon (Chameleon Multimodal Model), etc., but are not limited to such examples.

[0035] Note that it is desirable for the generation AI to be trained so as not to include personal information, etc. in its generation results. The generation AI is arranged in an external information processing device, and the information processing device 1 uses the generation AI via an API (Application Programming Interface) provided by the external information processing device, but the generation AI may also be arranged within the information processing device 1.

[0036] Subsequently, the information processing device 1 provides a target service, which is a service using generated information that is information generated using the generation AI based on a prompt including a plurality of items of information (step S3). The generated information is generated for each target prompt as described above, and in the target service, the generated information for each target prompt is provided, for example, according to an equal probability or a predetermined rule.

[0037] The information processing device 1 provides each piece of generated information generated using the generation AI to the user U by transmitting, for example, the generated information randomly or selected according to a predetermined rule from among a plurality of pieces of generated information generated using the generation AI to the terminal device 2.

[0038] For example, when the target service of the information processing apparatus 1 is an advertisement distribution service, the information processing apparatus 1 transmits to the terminal device 2 the advertisement content selected randomly or according to a predetermined rule from among a plurality of advertisement contents, thereby providing each advertisement content generated using the generation AI to the user U.

[0039] Subsequently, the information processing apparatus 1 acquires from an external information processing apparatus evaluation information that is information indicating an evaluation of the target service using each piece of generated information generated using the generation AI (step S4). The evaluation of the target service is one or more of an evaluation by the user U who uses the target service and an evaluation based on the behavior of the user U who uses the target service.

[0040] For example, when the generated information generated using the generation AI is advertisement content, the evaluation information is information indicating an index value that is a value of an advertisement effect index. The index value is an evaluation based on the behavior of the user U who uses the advertisement distribution service, and is, for example, CVR (Conversion Rate) or CTR (Click Through Rate).

[0041] In step S4, the information processing apparatus 1 can acquire the index value indicating the advertisement effect of the advertisement content for each context of the user U. The context of the user U is the context when the advertisement content is provided to the user U by the information processing apparatus 1. The context of the user U is the current situation of the user U or the situation surrounding the user U.

[0042] For example, the context of the user U includes the attributes of the user U, the current location of the user U, the current time, the physical environment in which the user U is placed, the social environment in which the user U is placed, the exercise state of the user U, and the emotion of the user U. The attributes of the user U are, for example, demographic attributes, psychographic attributes, and the like.

[0043] Demographic attributes are demographic attributes and include multiple attribute items such as age, gender, occupation, place of residence, annual income, and family structure. Psychographic attributes are psychological attributes and include multiple attribute items such as interests, lifestyle, and values. When the attribute of user U is a demographic attribute, it may be, for example, gender, age (generation), place of residence, occupation, or a combination of two or more of these, but is not limited to such examples.

[0044] The physical environment in which the user U is placed may be, for example, but not limited to, temperature, humidity, weather, illuminance, indoors, outdoors, or a combination of two or more of these. The social environment in which the user U is placed may be, for example, but not limited to, economic conditions, political conditions, popular products and services, or a combination of two or more of these.

[0045] The motion state of the user U may be, for example, running, walking, sitting, etc., but is not limited to these examples. The emotion of the user U may be, for example, laughing, angry, troubled, etc., but is not limited to these examples.

[0046] In addition, the context of user U may be a combination of two or more of, for example, user U's attributes, user U's current location, the current time, the physical environment in which user U is located, the social environment in which user U is located, user U's physical state, and user U's emotions.

[0047] The information processing device 1 can also determine the evaluation of a service that uses generated information generated using a generation AI instead of an external information processing device. For example, the information processing device 1 acquires information indicating whether or not advertising content has been clicked or information indicating whether or not a product targeted by the advertising content has been purchased from the terminal device 2 of the user U or an external information processing device, and calculates an index value for each advertising content based on the acquired information.

[0048] Subsequently, based on the evaluation information acquired in step S4, the information processing apparatus 1 estimates the importance levels of a plurality of items included in the information in the plurality of target prompts used in the generation AI in step S1 (step S5).

[0049] For example, based on the evaluation information for each piece of generated information generated using a plurality of target prompts each including information on a plurality of items with different combinations, the information processing apparatus 1 estimates the importance level of each of the plurality of items included in the plurality of target prompts.

[0050] For example, the information processing apparatus 1 estimates the importance level of each of the plurality of items by regression analysis, a machine learning algorithm, or the like. The machine learning algorithm is, for example, gradient boosting such as XG (eXtreme Gradient) boosting, a neural network, or the like.

[0051] For example, the information processing apparatus 1 uses the evaluation value as the dependent variable (label), each item as the independent variable (feature amount), and uses regression analysis, a machine learning algorithm, or the like to estimate the importance level of each of the plurality of items included in the plurality of target prompts.

[0052] For example, when the information processing apparatus 1 acquires an index value indicating the advertising effect of the advertising content for each context of the user U, it can also estimate the importance level of each of the plurality of items included in the plurality of target prompts for each context of the user U.

[0053] If the information processing apparatus 1 has not determined the information on the plurality of items included in the target prompt in step S1, in step S5, it can identify the information on the plurality of items included in the target prompt. For example, the information processing apparatus 1 extracts the information on the plurality of items from each target prompt using the above-described dictionary or generation AI.

[0054] The case where the information on a plurality of items included in the target prompt has not been determined in step S1 is, for example, when the prompt included in the prompt determination request is determined as the target prompt, when the prompt generated by the generative AI is determined as the target prompt, and the like.

[0055] Further, the information processing apparatus 1 can also calculate the importance for each item combination that is a combination of a plurality of different items. Also in this case, the information processing apparatus 1 estimates the importance for each combination of a plurality of different items by regression analysis, a machine learning algorithm, or the like.

[0056] Also, when the evaluation information acquired in step S4 is evaluation information for each context of the user U, the information processing apparatus 1 can also estimate the importance of each item, the importance of each combination of two or more items, etc. for each context of the user U.

[0057] Subsequently, the information processing apparatus 1 selects the information on one or more items to be included in the new prompt based on the importance of each of the plurality of items estimated in step S5 (step S6). Hereinafter, the item to be included in the new prompt may be described as an extraction item.

[0058] For example, the information processing apparatus 1 can select, as one or more extraction items to be included in the new prompt, the items among the plurality of items for which the importance has been respectively estimated in step S5 and whose importance is equal to or higher than a threshold value.

[0059] Further, the information processing apparatus 1 can select, as one or more extraction items to be included in the new prompt, the m items in descending order of importance among the plurality of items for which the importance has been respectively estimated in step S5. m is an integer of 1 or more.

[0060] In addition, the information processing apparatus 1 can also select, as one or more extraction items to be included in a new prompt, items included in a combination of items whose importance is equal to or higher than a threshold value among the combinations of items whose importance is estimated in step S5, instead of or in addition to items whose importance is equal to or higher than the threshold value.

[0061] In addition, the information processing apparatus 1 can also select, as one or more extraction items to be included in a new prompt, items included in a combination of items with high importance among the combinations of items whose importance is estimated in step S5, instead of or in addition to m items in descending order of importance. n is an integer of 1 or more.

[0062] Subsequently, the information processing apparatus 1 generates a new prompt including information on one or more extraction items selected in step S6 (step S7). For example, the information processing apparatus 1 generates, as a new prompt, a prompt including information on one or more extraction items selected in step S6 and instruction information for instructing generation of generated information using the information on one or more extraction items.

[0063] For example, assume that among the items "design inside the store", "status of the store", "category of the store", and the string "local community", the one or more extraction items selected in step S6 are the items "design inside the store", "category of the store", and the string "local community".

[0064] In this case, the information on the one or more extraction items selected in step S6 is the information on the string "comfortable interior", the information on the string "organic coffee shop", and the information on the string "local community".

[0065] The information processing apparatus 1 includes, for example, information on the string "Please create advertisement content featuring the information of the following items" as instruction information, and further generates, as a new prompt, a prompt including information on one or more extraction items selected in step S6.

[0066] In addition, the information processing apparatus 1 can also generate a new prompt by excluding information other than the information of the extraction items from the information of a plurality of items included in the original prompt. For example, assume that the original prompt is the information of the character string "Please create advertisement content for a new organic coffee shop featuring a comfortable interior. The advertisement content should be designed to target the local community.", and the character string "newly opened" is not the information of the extraction item.

[0067] In this case, the information processing apparatus 1 can generate the information of the character string "Please create advertisement content for an organic coffee shop featuring a comfortable interior. The advertisement content should be designed to target the local community." as a new prompt.

[0068] In addition, the information processing apparatus 1 can cause the generation AI to generate a new prompt, for example, by inputting information including the original prompt, the information of items other than the extraction items, and instruction information for instructing the generation of a new prompt by excluding the information of items other than the extraction items with reference to the original prompt to the generation AI.

[0069] Subsequently, the information processing apparatus 1 generates generated information using the generation AI based on the new prompt generated in step S7 (step S8). For example, the information processing apparatus 1 inputs the information including the new prompt generated in step S7 as input information to the generation AI, and causes the generation AI to generate generated information corresponding to the input information.

[0070] Subsequently, the information processing apparatus 1 provides the target service using the generated information generated in step S8 (step S9). For example, the information processing apparatus 1 provides the generated information generated using the generation AI to the user U by transmitting the generated information generated using the generation AI to the terminal device 2.

[0071] Note that in the above example, the information processing apparatus 1 determines each of a plurality of prompts each including information on a plurality of items with different combinations as the target prompt, but it is also possible to determine each of a plurality of prompts including two or more prompts including information on the same plurality of items as the target prompt.

[0072] Also, in the above example, instead of or in addition to determining each of a plurality of prompts as the target prompt, the information processing apparatus 1 can also determine one prompt as the target prompt.

[0073] In this case, the information processing apparatus 1 inputs the information including the determined one target prompt as input information to the generation AI, causes the generation AI to generate generation information corresponding to the input information, and provides a target service, which is a service using such generation information, to a plurality of users U. The information processing apparatus 1 acquires evaluation information, which is information indicating the evaluation by each of the plurality of users U of the provided target service, from an external information processing apparatus, and estimates the importance of a plurality of items based on the acquired evaluation information.

[0074] The information indicating the evaluation by the user U is, for example, information indicating a comment showing the evaluation of the target service, or information indicating the evaluation by the user U of a plurality of items included in the target prompt. The information processing apparatus 1 estimates the importance of a plurality of items included in the target prompt based on, for example, information indicating a comment showing the evaluation of the target service or information indicating the evaluation by the user U of a plurality of items included in the target prompt.

[0075] For example, the information processing apparatus 1 can input to the generation AI information including information indicating a comment showing the evaluation of the target service and information instructing the estimation of the importance of a plurality of items included in the target prompt from such information, and cause the generation AI to estimate the importance of a plurality of items included in the target prompt.

[0076] Further, the information processing apparatus 1 estimates the importance of a plurality of items included in the target prompt based on information indicating the evaluation of the user U for the plurality of items included in the target prompt, such that the higher the evaluation, the higher the importance.

[0077] Also, the information processing apparatus 1 not only causes the generation AI to generate generated information by inputting information including one target prompt to the generation AI, but also can cause the generation AI to generate generated information by inputting information including a plurality of target prompts to the generation AI as one input information.

[0078] In this way, the information processing apparatus 1 acquires evaluation information, which is information indicating an evaluation of a target service using generated information generated using the generation AI based on a prompt including information of a plurality of items, and estimates the importance of the plurality of items based on the acquired evaluation information. Thereby, the information processing apparatus 1 can more accurately support the optimization of the prompt.

[0079] Also, the information processing apparatus 1 generates a new prompt including information of one or more items selected based on the estimated importance of the plurality of items, and provides the target service using the generated information generated using the generation AI based on the generated new prompt. Thereby, the information processing apparatus 1 can provide information using a new prompt that has been more accurately optimized.

[0080] Hereinafter, the configuration of the information processing system including the information processing apparatus 1 and the plurality of terminal devices 2 that perform such processing will be described in detail.

[0081] 〔2. Configuration of Information Processing System〕 FIG. 2 is a diagram showing an example of the configuration of the information processing system according to the embodiment. As shown in FIG. 2, the information processing system 100 according to the embodiment includes an information processing apparatus 1, a plurality of terminal devices 2, and a business operator terminal 3.

[0082] The plurality of terminal devices 2 are used by different users U. The terminal device 2 is, for example, a notebook PC (Personal Computer), a desktop PC, a smartphone, a tablet PC, or a wearable device. The wearable device is, for example, smart glasses or a smartwatch, but is not limited to such examples.

[0083] The operator terminal 3 is a terminal device of the operator O and is, for example, a notebook PC, a desktop PC, a smartphone, a tablet PC, etc., but is not limited to such examples. When the service provided by the information processing device 1 is an advertisement distribution service, the operator O is an advertiser or an advertisement creator.

[0084] Each of the information processing device 1, the terminal device 2, and the operator terminal 3 is connected to be communicable with each other by wire or wirelessly via the network N. Note that the information processing system 100 shown in FIG. 2 may include a plurality of information processing devices 1 and operator terminals 3.

[0085] The network N includes, for example, a WAN (Wide Area Network) such as the Internet and a mobile communication network such as LTE (Long Term Evolution), 4G (4th Generation), and 5G (5th Generation: the 5th generation mobile communication system).

[0086] The terminal device 2 and the operator terminal 3 can be connected to the network N via a mobile communication network, short-range wireless communication such as Bluetooth (registered trademark), or a wireless LAN (Local Area Network), and communicate with the information processing device 1 and the like.

[0087] 〔3. Configuration of the information processing device 1〕 FIG. 3 is a diagram showing an example of the configuration of the information processing device 1 according to the embodiment. As shown in FIG. 3, the information processing device 1 includes a communication unit 10, a storage unit 11, and a processing unit 12.

[0088] 〔3.1. Communication unit 10〕 The communication unit 10 is realized by, for example, a communication module, a NIC (Network Interface Card), or the like. The communication unit 10 is connected to the network N by wire or wirelessly and transmits and receives information to and from various other devices. For example, the communication unit 10 transmits and receives information to and from the terminal device 2 and the operator terminal 3 via the network N.

[0089] 〔3.2. Storage unit 11〕 The storage unit 11 is realized by, for example, a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk.

[0090] 〔3.3. Processing unit 12〕 The processing unit 12 is a controller and is realized, for example, by a processor such as a CPU (Central Processing Unit) or an MPU (Micro Processing Unit) executing various programs (corresponding to an example of an information processing program) stored in a storage device inside the information processing device 1 using a RAM or the like as a work area.

[0091] Further, the processing unit 12 is a controller and may be realized by an integrated circuit such as an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a GPGPU (General Purpose Graphic Processing Unit).

[0092] As shown in FIG. 3, the processing unit 12 includes a reception unit 20, an acquisition unit 21, a determination unit 22, a generation unit 23, a provision unit 24, an estimation unit 25, and a selection unit 26, and realizes or executes the functions and operations of information processing described below. Note that the internal configuration of the processing unit 12 is not limited to the configuration shown in FIG. 3, and any other configuration may be used as long as it can perform the information processing described later.

[0093] 〔3.3.1. Reception Unit 20〕 The reception unit 20 receives various information and requests via the network N and the communication unit 10. For example, the reception unit 20 receives various requests transmitted from the terminal device 2 via the network N and the communication unit 10.

[0094] For example, when the target service provided by the providing unit 24 is an advertisement distribution service, the reception unit 20 receives an advertisement distribution request transmitted from the terminal device 2. Also, when the target service is an e-commerce service, the reception unit 20 receives a transmission request for a store page or a product page transmitted from the terminal device 2. The advertisement distribution request, the transmission request for the store page, and the transmission request for the product page are examples of service provision requests from the user U.

[0095] Also, when the target service is a text proofreading service, the reception unit 20 receives a text proofreading request transmitted from the terminal device 2. Also, when the target service is an online learning service, the reception unit 20 receives a learning support request transmitted from the terminal device 2. The text proofreading request and the learning support request are examples of service provision requests from the user U. The online learning service is a service that supports the learning of the user U by exchanging information with the user U in a chat format.

[0096] Also, the reception unit 20 receives various requests transmitted from the business operator terminal 3 via the network N and the communication unit 10. For example, the reception unit 20 receives a prompt determination request transmitted from the business operator terminal 3.

[0097] 〔3.3.2. Acquisition Unit 21〕 The acquisition unit 21 acquires various information from an external device or the like via the network N and the communication unit 10, or acquires various information from the storage unit 11.

[0098] For example, the acquisition unit 21 acquires evaluation information, which is information indicating an evaluation of the target service using a plurality of generated information, which is information generated using a generation AI based on a plurality of prompts including a plurality of items of information, from the terminal device 2 or an external information processing device, etc. The evaluation of the target service is one or more of an evaluation by a user U who uses the target service and an evaluation based on the behavior of the user U who uses the target service.

[0099] When the generated information generated using the generation AI is advertising content, the evaluation information is information indicating an index value, which is an index value of advertising effectiveness. The index value is an evaluation based on the behavior of a user U who uses the advertising distribution service, such as CVR or CTR.

[0100] Furthermore, when the generated information generated using the generation AI is a store page or a product page, the evaluation information is, for example, information indicating the evaluation by a user U who visited the store page or product page, or a CVR indicating the percentage of product purchases by a user U who visited the store page or product page.

[0101] The information indicating the evaluation by the user U is, for example, information indicating the evaluation by the user U of a store or product, for example, an evaluation value indicated on a five-point scale from 1 to 5. In this case, the evaluation information is, for example, the average or median of the evaluation values by the user U.

[0102] Furthermore, when the generated information generated using the generation AI is a proofreading result, the evaluation information is information indicating the evaluation of the text proofreading service by the user U who confirmed the proofreading result. In this case, the evaluation information is, for example, the average or median of the evaluation values by the user U.

[0103] Furthermore, when the generated information generated using the generation AI is a question or answer, the evaluation information is information indicating the evaluation of the online learning service by user U who confirmed the proofreading results. In this case, the evaluation information is, for example, the average or median of the evaluation values by user U.

[0104] The acquisition unit 21 can acquire evaluation information for each context of the user U. The context of the user U is the context when the advertisement content is provided to the user U by the providing unit 24. The context of the user U is the current situation of the user U or the situation surrounding the user U, etc.

[0105] For example, the context of the user U includes the attributes of the user U, the current location of the user U, the current time, the physical environment where the user U is located, the social environment where the user U is located, the movement state of the user U, and the emotion of the user U, etc. The attributes of the user U are, for example, demographic attributes, psychographic attributes, etc.

[0106] Demographic attributes are demographic attributes and include, for example, a plurality of attribute items such as age, gender, occupation, place of residence, annual income, family composition, etc. Psychographic attributes are psychological attributes and include, for example, a plurality of attribute items related to interests, lifestyle, values, etc. The attributes of the user U, when the attribute is a demographic attribute, are, for example, gender, age (age group), place of residence, occupation, or a combination of two or more of these, but are not limited to such examples.

[0107] The physical environment where the user U is located is, for example, temperature, humidity, weather, illuminance, indoor, outdoor, or a combination of two or more of these, but is not limited to such examples. The social environment where the user U is located is, for example, the economic situation, the political situation, popular goods and services, or a combination of two or more of these, but is not limited to such examples.

[0108] The movement state of the user U is, for example, a running state, a walking state, a sitting state, etc., but is not limited to such examples. Also, the emotion of the user U is, for example, a smiling state, an angry state, a troubled state, etc., but is not limited to such examples.

[0109] Note that the context of the user U may be, for example, a combination of two or more of the attributes of the user U, the current location of the user U, the current time, the physical environment in which the user U is placed, the social environment in which the user U is placed, the motion state of the user U, and the emotion of the user U.

[0110] The acquisition unit 21 acquires the information included in the request received by the reception unit 20. For example, the reception unit 20 is included in the prompt determination request received by the reception unit 20.

[0111] In addition, the acquisition unit acquires information indicating a comment showing an evaluation of the target service using the generation information generated based on the target prompt, or information indicating an evaluation of the user U for a plurality of items included in the target prompt, from the terminal device 2 or an external information processing device or the like.

[0112] [3.3.3. Decision unit 22] The decision unit 22 determines one prompt including information on a plurality of items as the target prompt, or determines each of a plurality of prompts including information on a plurality of items as the target prompt.

[0113] For example, the decision unit 22 determines each of a plurality of prompts each including information on a plurality of items with different combinations as the target prompt. The plurality of target prompts include, for example, information on a plurality of items with different combinations.

[0114] Each of the plurality of items included in the target prompt is an item indicating the type of a key matter or a characteristic matter, and is, for example, an item indicating a semantic type such as a semantic field. Also, the information of the item is information indicating the content of the key matter or information indicating the content of the characteristic matter.

[0115] The target prompt is information input to the above-described generation AI in order to generate the provided information that is the information to be provided to the user U in the target service provided by the providing unit 24.

[0116] When the target service is an advertising distribution service, the prompt determined by the determination unit 22 is information that is input to the generation AI described later in order to generate, as generation information, the advertisement content provided by the advertising distribution service to the generation AI. The advertisement content is content including at least one of, for example, text and an image.

[0117] Also, when the target service is an e-commerce service, the prompt determined by the determination unit 22 is information that is input to the generation AI in order to cause the generation AI to generate, as generation information, a store page or a product page provided by the e-commerce service.

[0118] Also, when the target service is a text proofreading service, the prompt determined by the determination unit 22 is information that is input to the generation AI in order to cause the generation AI to generate information indicating the proofreading result for the text included in the text proofreading request received by the reception unit 20.

[0119] Also, when the target service is an online learning service, the prompt determined by the determination unit 22 is information that is input to the generation AI in order to cause the generation AI to generate information indicating information corresponding to the learning support request received by the reception unit 20.

[0120] When a plurality of target prompts are determined by the determination unit 22, these plurality of target prompts may be prompts regarding the same target, or may each be prompts regarding different targets, or may be a plurality of prompts regarding partially different targets. For example, when the target service is an advertising distribution service, the same target is, for example, a specific landing page, and the plurality of targets are a specific plurality of landing pages.

[0121] Also, when the target service is an e-commerce service, the same target is, for example, the same product or service, and multiple targets are multiple products or services. Also, when the target service is a text proofreading service, the same target is, for example, text in the same category, and multiple targets are texts in multiple categories. Also, when the target service is an online learning service, the same target is, for example, the same subject, and multiple targets are multiple subjects.

[0122] Here, the case where the target service is an advertising distribution service will be described. The prompt input to the generation AI to generate the advertisement content provided by the advertising distribution service as generation information is, for example, the information of the character string "Please create advertisement content for a newly opened organic coffee shop featuring a comfortable interior. The advertisement content should be designed to target the local community."

[0123] Also, the information of multiple items is the information of the character string "comfortable interior", the information of the character string "newly opened", the information of the character string "organic coffee shop", the information of the character string "local community", etc.

[0124] The information of the character string "comfortable interior" is the information of the item "design inside the store", the character string "newly opened" is the item "status of the store", the information of the character string "organic coffee shop" is the information of the item "category of the store", and the information of the character string "local community" is the information of the item "residence of the target user". The information of the item "residence of the target user" is an example of the information indicating the attributes of the target user U.

[0125] Note that the information of the item is not limited to the above example, and may be, for example, the information of the item "location of the store", the information of the item "event", the information of the item "coupon or discount", the information of the item "product price", the item "age group of the target user", the item "interests of the target user", etc.

[0126] In addition, in the above example, the advertising content of physical stores was used as an example for explanation, but the advertising content is not limited to the advertising content of physical stores. For example, the advertising content may be the advertising content of virtual stores on e-commerce sites, the advertising content of physical stores or products sold in stores, or other advertising content.

[0127] Next, the case where the target service is a text proofreading service will be described. The prompt input to the generative AI to generate the proofreading result provided by the text proofreading service as generation information is, for example, information such as the character string "You are a professional proofreader. Detect typos, fluctuations, and other misnotations in the input text. The maximum number of detected items is 20. Please check each sentence carefully. The output format is as follows.\n{Error type}\n<Correct>{Original misnotation}\n<Incorrect>{Corrected misnotation}\n\n{Error type} is a character string indicating the error type, {Original misnotation} is a character string indicating the text before the misnotation, and {Corrected misnotation} is a character string indicating the corrected text."

[0128] The information of the character string "You are a professional proofreader." is the information of the item "Role of AI", the information of the character string "Detect typos, fluctuations, and other misnotations in the input text." is the information of the item "Detection target", the information of the character string "The maximum number of detected items is 20." is the information of the item "Number of detections", and the information of the character string "Please check each sentence carefully." is the information of the item "Detection procedure".

[0129] In addition, the character string "The output format is as follows.\n{Error type}\n<Correct>{Original misnotation}\n<Incorrect>{Corrected misnotation}\n\n{Error type} is a character string indicating the error type, {Original misnotation} is a character string indicating the text before the misnotation, and {Corrected misnotation} is a character string indicating the corrected text." is the information of the item "Output format".

[0130] Note that the information in the item "Output Format" may further be information on subdivided items. For example, it may be information such as the item "Output Format - Misspelling Type" information, the item "Output Format - Misspelling Content" information, the item "Output Format - Correction Content" information, etc., and may also include information such as the item "Output Format - Detection Reason" information.

[0131] Next, the case where the target service is an online learning service will be described. The prompt input to the generative AI to generate information indicating problems and answers provided in the online learning service as generative information is, for example, information such as the character string "You are a professional teacher. Under the following constraints, follow the following procedures to provide an educational service to the user.\nConstraints\n1. Correspond to the needs of individual learners: Provide customized guidance tailored to the learner's knowledge level, interests, and learning style\n2. Abide by educational principles: Based on teaching methods for including educational accuracy and understanding...\nProcedures\n1. Grasp the learning goals of the user: Confirm the topics and goals that the user wants to learn\n2. Create a customized learning plan: Create an individual learning plan based on the user's needs and goals..."

[0132] The information of the character string "You are a professional teacher." is the information of the item "Role of AI", the information of the character string "1. Correspond to the needs of individual learners:... " is the information of the item "Constraints - Needs Correspondence", the information of the character string "2. Abide by educational principles:... " is the information of the item "Constraints - Educational Principles",..., the information of the character string "1. Grasp the learning goals of the user:..." is the information of the item "Procedures - Goal Grasping",..., the information of the character string "2. Create a customized learning plan:..." is the information of the item "Procedures - Learning Plan Creation, etc.".

[0133] The information of the items is not limited to the above examples. For example, it may be information of the item "tone", information of the item "prompt format", etc. The information of the item "tone" is, for example, information such as a direct command like "do it" or an indirect command like "please do it". The item "prompt format" is, for example, information such as zero-shot, one-shot, few-shot, etc.

[0134] When the determination unit 22 receives, for example, a prompt determination request transmitted from the business operator terminal 3, the determination unit 22 extracts information of a plurality of items from the information included in the prompt determination request. Then, the determination unit 22 determines a plurality of prompts each including information of two or more items with different combinations among the extracted information of the plurality of items as the plurality of target prompts described above.

[0135] For example, the determination unit 22 has dictionary information including a plurality of terms for each item, and can extract information of a plurality of items from the information included in the prompt determination request using such dictionary information. Also, the determination unit 22 can extract information of a plurality of items from the information included in the prompt determination request using a text generation AI such as a large language model.

[0136] Further, the determination unit 22 can transmit, for example, template information for allowing the prompt user to input or select information of a plurality of predetermined items to the business operator terminal 3 and display the template information on the business operator terminal 3. In this case, the business operator O can input or select information of a plurality of items by operating the business operator terminal 3, and thereby a prompt determination request including information of a plurality of items is transmitted from the business operator terminal 3 to the information processing apparatus 1.

[0137] The prompt determination request sent from the business operator terminal 3 may include information necessary to generate a prompt containing information on a plurality of items instead of the template information. For example, when the target service is an online learning service, information specifying the landing page targeted by the advertisement may be included. In this case, the determination unit 22 can extract information on a plurality of items from the landing page specified in the prompt determination request using the above-described dictionary or generative AI. The determination unit 22 determines a plurality of prompts each including information on two or more items with different combinations among the extracted information on the plurality of items as the above-described plurality of target prompts.

[0138] Note that the target prompt may be, for example, a prompt generated using generative AI. Further, the target prompt is not limited to a character prompt, and may include, for example, image data in addition to the character prompt.

[0139] The information on the items included in the target prompt is not limited to the information on the items included in the character prompt, and may include, for example, information indicating the type of image shown by the image data or information on the items included in the image shown by the image data. The information on the items included in the image shown by the image data is an object, scene, emotion, color, etc. shown by the image shown by the image data, but is not limited to such examples.

[0140] Note that when the prompt determination request includes a plurality of prompts, the determination unit 22 can also determine the prompts included in the prompt determination request as a plurality of target prompts.

[0141] 〔3.3.4. Generation unit 23〕 The generation unit 23 performs a process of inputting, as input information, information including the target prompt, which is a prompt including information on a plurality of items determined by the determination unit 22, to the generative AI, and causing the generative AI to generate generation information corresponding to the input information.

[0142] When there are multiple target prompts determined by the determination unit 22, the generation unit 23 inputs information including the target prompts as input information to the generation AI for each target prompt, and causes the generation AI to generate generation information corresponding to the input information.

[0143] Also, when there are multiple target prompts determined by the determination unit 22, the generation unit 23 can also input information including two or more target prompts among these multiple target prompts as input information to the generation AI, and cause the generation AI to generate generation information corresponding to the input information.

[0144] When there is one target prompt determined by the determination unit 22, the generation unit 23 inputs information including the one target prompt as input information to the generation AI, and causes the generation AI to generate generation information corresponding to the input information.

[0145] The generation AI is, for example, a text generation AI, an image generation AI, or a multimodal generation AI. The text generation AI is, for example, a large language model trained to estimate and output the next token from the input token sequence, and is, for example, a Transformer-based model, an RNN-based model, etc., but may also be a hybrid model of these. Also, the text generation AI may be a composite system combined with an identification mechanism for preventing illegal use.

[0146] The Transformer-based model is, for example, GPT, PaLM2, etc., but is not limited to such examples. The RNN-based model is, for example, RWKV, etc., but is not limited to such examples.

[0147] The image generation AI is an AI that generates an image from text, and is, for example, StackGAN, AttnGAN, T2I with Transformers, Diffusion model, etc., but is not limited to such examples. Examples of the Diffusion model include DALL-E and Stable-Diffusion.

[0148] A multimodal generation AI is a generation AI that generates at least one of text, images, and audio from at least one of text, images, and audio. Examples of multimodal generation AIs include, but are not limited to, GPT-4 Turbo with vision, Gemini, CM3Leon, etc.

[0149] Note that it is desirable for the generation AI to be trained so as not to include personal information or the like in its generation results. The generation AI is arranged in an external information processing device, and the generation unit 23 uses the generation AI via an API provided by the external information processing device, but the generation AI may be arranged within the information processing device 1.

[0150] When the target service is an advertising distribution service or an e-commerce service, the generation unit 23 inputs, as input information to the generation AI, information including a target prompt that is a prompt including information on a plurality of items determined by the determination unit 22, and performs, for each target prompt determined by the determination unit 22, a process of causing the generation AI to generate generation information corresponding to the input information.

[0151] Also, when the target service is a text correction service or an online learning service, and there is a service provision request from the user U, the generation unit 23 inputs, as input information to the generation AI, information including a target prompt randomly or selected according to a predetermined rule from among a plurality of target prompts determined by the determination unit 22 and information included in the service provision request, and causes the generation AI to generate generation information corresponding to the input information.

[0152] For example, when the target service is a text correction service, the information included in the service provision request from the user U is information indicating a text. Also, when the target service is an online learning service, the information included in the service provision request from the user U is information in the form of a character string indicating a question or an answer.

[0153] Further, the generation unit 23 generates, as a new prompt, a prompt including information on one or more items selected based on the importance levels of a plurality of items estimated by the estimation unit 25.

[0154] For example, the generation unit 23 generates, as a new prompt, a prompt including information on one or more items selected by the selection unit 26 based on the importance levels of a plurality of items estimated by the estimation unit 25. Further, the generation unit 23 can also generate, as a new prompt, a prompt including information on one or more items selected by the business operator O based on the importance levels of a plurality of items estimated by the estimation unit 25.

[0155] For example, the generation unit 23 generates, as a new prompt, a prompt including information on one or more extracted items selected by the selection unit 26 and instruction information for instructing generation of generated information using the information on the one or more extracted items.

[0156] For example, assume that one or more extracted items selected by the selection unit 26 are the item "design inside the store", the item "state of the store", the item "category of the store", and the string "local community", and among them, the item "design inside the store", the item "category of the store", and the string "local community".

[0157] In this case, the information on one or more extracted items selected by the selection unit 26 is the information on the string "comfortable interior", the information on the string "organic coffee shop", and the information on the string "local community".

[0158] The generation unit 23 generates, as a new prompt, a prompt including, for example, the information on the string "Please create advertisement content characterized by the information on the following items" as instruction information and further including the information on one or more extracted items selected by the selection unit 26.

[0159] In addition, the generation unit 23 can also generate a new prompt by excluding information other than the information of the extraction item from the information of a plurality of items included in the original prompt. For example, assume that the original prompt is the information of the character string "Please create advertisement content for a new organic coffee shop featuring a comfortable interior. Make the advertisement content designed to target the local community.", and the character string "newly opened" is not the information of the extraction item.

[0160] In this case, the generation unit 23 can generate the information of the character string "Please create advertisement content for an organic coffee shop featuring a comfortable interior. Make the advertisement content designed to target the local community." as a new prompt.

[0161] In addition, the generation unit 23 can cause the generation AI to generate a new prompt, for example, by inputting information including the original prompt, the information of items other than the extraction item, and instruction information for instructing the generation of a new prompt by excluding the information of items other than the extraction item with reference to the original prompt into the generation AI.

[0162] The generation unit 23 inputs the information including the generated new prompt as input information into the generation AI, and causes the generation AI to generate generation information corresponding to the input information. When the generation unit 23 generates a plurality of new prompts, the process of inputting the information including each of the plurality of new prompts as input information into the generation AI and causing the generation AI to generate generation information corresponding to the input information is performed for each new prompt.

[0163] When one or more extraction items to be included in the new prompt are selected by the selection unit 26 for each context of the user U, the generation unit 23 can perform, for each context of the user U, the process of inputting the information including the generated new prompt as input information into the generation AI and causing the generation AI to generate generation information corresponding to the input information.

[0164] 〔3.3.5. Provision Unit 24〕 The providing unit 24 provides the target service using a plurality of pieces of generated information, which is information generated using a generation AI based on each of a plurality of target prompts. The generated information is generated for each target prompt as described above, and in the target service, the generated information for each target prompt is provided, for example, with equal probability or according to a predetermined rule.

[0165] The providing unit 24 provides each piece of generation information generated using the generation AI to the user U, for example, by transmitting generation information selected randomly or according to predetermined rules from among multiple pieces of generation information generated using the generation AI to the terminal device 2.

[0166] For example, when the target service is an advertising distribution service, the providing unit 24 provides each advertising content generated using the generation AI to the user U by transmitting advertising content selected from multiple advertising contents randomly or according to predetermined rules to the terminal device 2.

[0167] In addition, when the target service is an e-commerce service, the providing unit 24 provides each store page and each product page generated using the generation AI to the user U by sending a store page or product page selected randomly or according to predetermined rules from among multiple store pages or product pages to the terminal device 2.

[0168] In addition, when the target service is a text proofreading service, the providing unit 24 provides the proofreading result generated using the generation AI to the terminal device 2 using a target prompt selected by the generating unit 23 randomly or according to predetermined rules from among multiple target prompts, thereby providing the proofreading result generated using the generation AI to the user U.

[0169] Further, when the target service is an online learning service, the providing unit 24 transmits information such as questions and answers to the user U generated using the generation AI with the target prompt selected by the generation unit 23 according to a random or predetermined rule among a plurality of target prompts to the terminal device 2, thereby providing the user U with information such as questions and answers to the user U generated using the generation AI.

[0170] In addition, the providing unit 24 provides the target service using the generated information generated using the generation AI based on the new prompt generated by the generation unit 23. For example, the providing unit 24 transmits the generated information generated using the generation AI based on the new prompt generated by the generation unit 23 to the terminal device 2, thereby providing the user U with the generated information generated using the new prompt by the generation AI.

[0171] The providing unit 24 can provide the user U with the generated information corresponding to the context of the user U generated using the new prompt by the generation AI by transmitting the generated information generated using the generation AI based on the new prompt generated by the generation unit 23 and corresponding to the context of the user U to the terminal device 2.

[0172] When there is no information included in the service provision request or when the information included in the service provision request is not used for generating the generated information, the providing unit 24 can provide the same generated information to a plurality of users U. Further, when the information included in the service provision request is used for generating the generated information, the providing unit 24 can provide different generated information to the user U for each piece of information included in the service provision request.

[0173] In addition, when one target prompt is determined by the determination unit 22, the providing unit 24 can provide the generated information provided by the generation unit 23 using such one target prompt as the target service to a plurality of users U.

[0174] 〔3.3.6. Estimation Unit 25〕 Based on the evaluation information acquired by the acquisition unit 21, the estimation unit 25 estimates the importance levels of multiple items.

[0175] For example, based on the evaluation information for each piece of generated information generated using a plurality of target prompts each containing information on multiple items with different combinations, the estimation unit 25 estimates the importance level of each of the multiple items included in the plurality of target prompts.

[0176] For example, the information processing apparatus 1 estimates the importance level of each of the multiple items by means of regression analysis, machine learning algorithms, etc. The machine learning algorithm is, for example, gradient boosting such as XG boosting, a neural network, or the like.

[0177] For example, the estimation unit 25 uses the evaluation value as the dependent variable (label), each item as the independent variable (feature), and uses regression analysis, machine learning algorithms, etc. to estimate the importance level of each of the multiple items included in the plurality of target prompts.

[0178] For example, when the estimation unit 25 acquires an index value indicating the advertising effect of the advertising content for each context of the user U, it can also estimate the importance level of each of the multiple items included in the plurality of target prompts for each context of the user U.

[0179] When the information on the multiple items included in the target prompt has not been determined by the determination unit 22, the estimation unit 25 can identify the information on the multiple items included in the target prompt. For example, the estimation unit 25 extracts the information on the multiple items from each target prompt using the above-described dictionary or generative AI.

[0180] The case where the information on the multiple items included in the target prompt has not been determined by the determination unit 22 is, for example, when the prompt included in the prompt determination request is determined as the target prompt, or when the determination unit 22 determines the prompt generated by the generative AI as the target prompt.

[0181] Further, the estimation unit 25 can also calculate the importance for each combination of items that are combinations of a plurality of different items. Also in this case, the estimation unit 25 estimates the importance for each combination of a plurality of different items by regression analysis, machine learning algorithms, or the like.

[0182] Further, when the evaluation information acquired by the acquisition unit 21 is evaluation information for each context of the user U, the estimation unit 25 can also estimate the importance of each item, the importance of each combination of two or more items, etc. for each context of the user U.

[0183] Further, when one target prompt is determined by the determination unit 22, the estimation unit 25 estimates the importance of the plurality of items included in the target prompt based on information indicating a comment showing an evaluation of the target service using the generated information generated based on such target prompt or information indicating an evaluation of the user U for the plurality of items included in the target prompt.

[0184] For example, the estimation unit 25 can input information including information indicating a comment showing an evaluation of the target service and information instructing the estimation of the importance of the plurality of items included in the target prompt from such information to the generation AI, and cause the generation AI to estimate the importance of the plurality of items included in the target prompt.

[0185] Further, the estimation unit 25 estimates the importance of the plurality of items included in the target prompt such that the higher the evaluation, the higher the importance, based on the information indicating the evaluation of the user U for the plurality of items included in the target prompt.

[0186] Note that the estimation unit 25 can also determine an evaluation of a service using generated information generated using a generation AI instead of an external information processing device. For example, the acquisition unit 21 of the information processing device 1 acquires information indicating whether an advertisement content has been clicked, information indicating whether a product targeted by the advertisement content has been purchased, etc. from the terminal device 2 of the user U, an external information processing device, etc., and the estimation unit 25 calculates an index value of each advertisement content based on the information acquired by the acquisition unit 21.

[0187] [3.3.7. Selection unit 26] The selection unit 26 selects information on one or more extraction items, which is information on one or more items to be included in a new prompt, from among information on a plurality of items based on the importance levels of the plurality of items estimated by the estimation unit 25.

[0188] For example, the selection unit 26 can select, as one or more extraction items to be included in a new prompt, items among the plurality of items whose importance levels have been respectively estimated by the estimation unit 25 and whose importance levels are equal to or higher than a threshold value.

[0189] Further, the selection unit 26 can select, as one or more extraction items to be included in a new prompt, m items in descending order of importance among the plurality of items whose importance levels have been respectively estimated by the estimation unit 25. m is an integer equal to or greater than 1.

[0190] Further, instead of or in addition to items whose importance levels are equal to or higher than a threshold value, the selection unit 26 can also select, as one or more extraction items to be included in a new prompt, items included in an item combination whose importance level is equal to or higher than a threshold value among the item combinations whose importance levels have been respectively estimated by the estimation unit 25.

[0191] Further, instead of or in addition to m items in descending order of importance among the item combinations whose importance levels have been respectively estimated by the estimation unit 25, the selection unit 26 can also select, as one or more extraction items to be included in a new prompt, items included in an item combination with a high importance level. n is an integer equal to or greater than 1.

[0192] Further, when the importance of each item and the importance of each combination of two or more items are estimated for each context of the user U by the estimation unit 25, the selection unit 26 can select one or more extraction items to be included in the new prompt by the same method as the above-described process.

[0193] [4. Processing Procedure] Next, the information processing procedure by the processing unit 12 of the information processing apparatus 1 according to the embodiment will be described. FIG. 4 is a flowchart showing an example of information processing by the processing unit 12 of the information processing apparatus 1 according to the embodiment.

[0194] As shown in FIG. 4, the processing unit 12 of the information processing apparatus 1 determines whether the target prompt determination timing has arrived (step S10). The target prompt determination timing is, for example, the timing when a predetermined period has arrived, the timing when a prompt request transmitted from the business operator terminal 3 has been received, etc., but is not limited to such examples.

[0195] When the processing unit 12 determines that the target prompt determination timing has arrived (step S10: Yes), it determines a plurality of target prompts (step S11). Then, the processing unit 12 determines whether the target service, which is the service that provides the generated information using the target prompt in step S11, is a specific service (step S12).

[0196] The specific service is, for example, an advertisement distribution service or an e-commerce service, but is not limited to such examples. The target services other than the specific service are, for example, a text proofreading service or an online learning service, etc., but are not limited to such examples.

[0197] When the processing unit 12 determines that the target service is a specific service (step S12: Yes), it inputs the information including each of the plurality of target prompts into the generation AI, and generates a plurality of generated information using the generation AI (step S13).

[0198] When the process in step S13 ends, when it is determined that the target service is not a specific service (step S12: No), or when it is determined that it is not the target prompt determination timing (step S10: No), the processing unit 12 determines whether it is the service provision timing (step S14). The service provision timing is, for example, the timing when a service provision request transmitted from the terminal device 2 of the user U is received, but is not limited to such an example.

[0199] When the processing unit 12 determines that it is the service provision timing (step S14: Yes), it determines whether the target service is a service other than the specific service (step S15). When the processing unit 12 determines that the target service is a service other than the specific service (step S15: Yes), it inputs information including the selected target prompt among the plurality of target prompts to the generation AI, and generates generated information using the generation AI (step S16).

[0200] When the process in step S15 ends, or when it is determined that the target service is not a service other than the specific service (step S15: No), the processing unit 12 provides the target service using the generated information generated in step S13 or step S16 (step S17).

[0201] When the process in step S17 ends, or when it is determined that it is not the service provision timing (step S14: No), the processing unit 12 determines whether a new prompt determination timing has arrived (step S18). The new prompt determination timing is, for example, the timing when a predetermined period has elapsed after the generated information generated using the target prompt is provided, or the timing when the number of times the generated information generated using the target prompt is provided to the user U reaches a predetermined number or more, but is not limited to such an example.

[0202] When the processing unit 12 determines that it is the timing to determine a new prompt (step S18: Yes), it performs prompt generation processing (step S19). The processing in step S19 is the processing of steps S30 to S35 shown in FIG. 5, which will be described in detail later.

[0203] When the processing in step S19 by the processing unit 12 is completed, or when it determines that it is not the timing to determine a new prompt (step S18: No), it determines whether it is the timing to end the operation (step S20). The processing unit 12 determines that it is the timing to end the operation, for example, when the power of the information processing apparatus 1 is turned off.

[0204] When the processing unit 12 determines that it is not the timing to end the operation (step S20: No), it shifts the processing to step S10, and when it determines that it is the timing to end the operation (step S20: Yes), it ends the processing shown in FIG. 4.

[0205] FIG. 5 is a flowchart showing an example of prompt generation processing by the processing unit 12 of the information processing apparatus 1 according to the embodiment. As shown in FIG. 5, the processing unit 12 acquires evaluation information, which is information indicating an evaluation of a target service using a plurality of generated information that is information generated using a generation AI based on a plurality of target prompts including information on a plurality of items (step S30).

[0206] Subsequently, the processing unit 12 estimates the importance of a plurality of items based on the evaluation information acquired in step S30 (step S31). Then, the processing unit 12 extracts one or more items from the plurality of items based on the importance of the plurality of items estimated in step S31 (step S32).

[0207] Subsequently, the processing unit 12 determines the prompt including one or more extraction items extracted in step S32 as a new prompt (step S33). Then, the processing unit 12 inputs the information including the new prompt generated in step S33 into the generation AI and generates generated information using the generation AI (step S34). The processing unit 12 provides the target service using the generated information generated in step S34 (step S35) and ends the processing shown in FIG. 5.

[0208] 〔5. Variation〕 The target service is not limited to the above-described example, and may be, for example, a text creation service, a text summarization service, a programming support service, or the like.

[0209] In the above-described example, the information indicating the evaluation of the target service is, for example, an evaluation value indicated by a five-level evaluation from 1 to 5, but may be an evaluation value of four levels or less or an evaluation value of six levels or more. Further, the information indicating the evaluation of the target service may be, for example, the number of posts and reposts to X (old Twitter), or an evaluation value calculated from the review of the user U.

[0210] Also, in the above-described example, the generation unit 23 inputs the information including the prompt into the generation AI as input information and causes the generation AI to generate generated information, but is not limited to such an example. For example, the generation unit 23 can also generate generated information based on the information output from the generation AI by inputting the information including the prompt into the generation AI as input information.

[0211] For example, the prompt may be a prompt including intention definition information for extracting an intention type and intention content. The intention definition information includes, for example, instruction information including an instruction to extract an intention type and intention content from user setting information, and definition information including the intention type and information defining the intention type. When using the API provided by OpenAI (registered trademark), for example, the generation unit 23 can cause the generation AI to generate intention information by using the function calling function.

[0212] The intention type is, for example, the type of intention of the additional information, and the information indicating the intention type is, for example, information specifying a function or a function corresponding to the intention type. The intention content is, for example, the content of the intention of the additional information, and the information indicating the intention content is, for example, information indicating an argument of a function or a parameter of a function corresponding to the intention type.

[0213] In this case, the generation AI generates intention information including information indicating the intention type and information indicating the intention content. The generation unit 23 acquires information from an external information processing device or an internal storage unit using the intention information generated by the generation AI, inputs information including the acquired information and instruction information indicating an instruction to generate generated information using the acquired information to the generation AI, and causes the generation AI to generate the generated information.

[0214] The processing unit 12 may also include an extraction unit that extracts information about items included in a prompt using a technique such as slot filtering. In this case, when the extraction unit extracts information about multiple items from a prompt, the determination unit 22 can determine the extracted prompt as a target prompt. In slot filtering, each item is called a slot, but each item is not limited to a slot and may be any item that indicates a type of key or characteristic item, as described above.

[0215] The extraction unit included in the processing unit 12 extracts information on multiple items from the prompt, for example, on a rule basis. For example, the extraction unit has dictionary information containing multiple terms for each item, and can extract information on multiple items from the prompt using the dictionary information.

[0216] The extraction unit also has an extraction model for extracting multiple items of information contained in the prompt, and can extract multiple items of information contained in the prompt using this extraction model.

[0217] The extraction model is a model trained to extract information on a plurality of items from a prompt using learning information that includes, for each prompt, a combination of the prompt and information on the plurality of items included in the prompt. The extraction model is, for example, an LSTM (Long Short-Term Memory) or Transformer-based model, but may also be the above-described generative AI or other models.

[0218] Also, the extraction unit can extract information on a plurality of items from the prompt using a language model such as a text generation AI. For example, the extraction unit can input information including instruction information indicating an instruction to extract information on a plurality of items from the prompt and the prompt into the text generation AI, and have the text generation AI extract information on the plurality of items.

[0219] [6. Hardware Configuration] The information processing apparatus 1 according to the above-described embodiment is realized by a computer 80 having a configuration as shown in FIG. 6, for example. FIG. 6 is a hardware configuration diagram showing an example of a computer 80 that realizes the functions of the information processing apparatus 1 according to the embodiment. The computer 80 includes a CPU 81, a RAM 82, a ROM (Read Only Memory) 83, an HDD (Hard Disk Drive) 84, a communication interface (I / F) 85, an input / output interface (I / F) 86, and a media interface (I / F) 87.

[0220] The CPU 81 operates based on a program stored in the ROM 83 or the HDD 84 and controls each part. The ROM 83 stores a boot program executed by the CPU 81 when the computer 80 is started up, a program dependent on the hardware of the computer 80, and the like.

[0221] The HDD 84 stores programs executed by the CPU 81, data used by such programs, and the like. The communication interface 85 receives data from other devices via the network N (see FIG. 2), sends it to the CPU 81, and transmits the data generated by the CPU 81 to other devices via the network N.

[0222] The CPU 81 controls output devices such as displays and printers, and input devices such as keyboards or mice, via the input / output interface 86. The CPU 81 acquires data from the input devices via the input / output interface 86. Also, the CPU 81 outputs the data generated via the input / output interface 86 to the output devices.

[0223] The media interface 87 reads a program or data stored in the recording medium 88 and provides it to the CPU 81 via the RAM 82. The CPU 81 loads such a program from the recording medium 88 onto the RAM 82 via the media interface 87 and executes the loaded program. The recording medium 88 is, for example, an optical recording medium such as a DVD (Digital Versatile Disc) or PD (Phase change rewritable Disk), a magneto-optical recording medium such as an MO (Magneto-Optical disk), a tape medium, a magnetic recording medium, or a semiconductor memory.

[0224] For example, when the computer 80 functions as the information processing apparatus 1 according to the embodiment, the CPU 81 of the computer 80 realizes the functions of the processing unit 12 by executing the program loaded on the RAM 82. Also, the data in the storage unit 11 is stored in the HDD 84. The CPU 81 of the computer 80 reads and executes these programs from the recording medium 88, but as another example, these programs may be acquired from other devices via the network N.

[0225] [7. Others] In addition, among the processes described in the above embodiments, all or part of the processes described as being automatically performed can be manually performed, or all or part of the processes described as being manually performed can be automatically performed by a known method. In addition, the processing procedures, specific names, and information including various data and parameters shown in the above documents and drawings can be arbitrarily changed unless otherwise specified. For example, the various information shown in each figure is not limited to the illustrated information.

[0226] In addition, each component of each device shown in the drawings is a functional concept, and does not necessarily have to be physically configured as shown in the drawings. That is, the specific form of the distribution and integration of each device is not limited to that shown in the drawings, and all or part of it can be functionally or physically distributed and integrated in any unit according to various loads, usage conditions, etc.

[0227] For example, the information processing apparatus 1 described above may be realized by a terminal device and a server computer, or may be realized by a plurality of server computers. Also, depending on the function, the configuration can be flexibly changed, such as by calling an external platform or the like using an API or network computing.

[0228] In addition, the above-described embodiments and modifications can be appropriately combined as long as the processing contents do not conflict.

[0229] 〔8. Effects〕 As described above, the information processing apparatus 1 according to the embodiment includes an acquisition unit 21 and an estimation unit 25. The acquisition unit 21 acquires evaluation information indicating an evaluation of a service using generated information, which is information generated using a generation AI based on a prompt including information of a plurality of items. The estimation unit 25 estimates the importance of a plurality of items based on the evaluation information acquired by the acquisition unit 21. Thereby, the information processing apparatus 1 can more accurately support the optimization of the prompt.

[0230] The estimation unit 25 estimates the importance of each of the plurality of items. Thereby, the information processing apparatus 1 can more accurately support the optimization of the prompt.

[0231] The estimation unit 25 estimates the importance for each combination of the plurality of items. Thereby, the information processing apparatus 1 can more accurately support the optimization of the prompt.

[0232] Further, the estimation unit 25 estimates the importance of the plurality of items based on evaluation information that indicates an evaluation of a service using a plurality of pieces of generated information each generated using a generation AI based on a corresponding prompt among a plurality of prompts including information on a plurality of items with different combinations.

[0233] Further, the information processing apparatus 1 includes a selection unit 26 that determines information on one or more items to be included in a new prompt from among the information on the plurality of items based on the importance of the plurality of items estimated by the estimation unit 25. Thereby, the information processing apparatus 1 can more accurately support the optimization of the prompt.

[0234] Further, the information processing apparatus 1 includes a generation unit 23 that generates a new prompt including information on one or more items selected by the selection unit 26, and a provision unit 24 that provides a service using generated information that is information generated using a generation AI based on the new prompt generated by the generation unit 23. Thereby, the information processing apparatus 1 can provide information using a more accurately optimized prompt.

[0235] Further, the acquisition unit 21 acquires the evaluation information for each context of the user U who uses the service, the estimation unit 25 estimates the importance of the plurality of items for each context of the user U, and the selection unit 26 selects information on items to be included in a new prompt from among the information on the plurality of items for each context of the user U. Thereby, the information processing apparatus 1 can more accurately support the optimization of the prompt.

[0236] In addition, the evaluation of the service is one or more of the evaluation by the user U who uses the service and the evaluation based on the behavior of the user U who uses the service. Thereby, the information processing apparatus 1 can more accurately support the optimization of the prompt.

[0237] In addition, the service is an advertisement distribution service, and the generated information is advertisement content distributed by the advertisement distribution service. Thereby, the information processing apparatus 1 can more accurately support the optimization of the prompt for generating advertisement content in the advertisement distribution service.

[0238] As described above, the embodiments of the present application have been described in detail with reference to the drawings. However, this is an example, and the present invention can be implemented in other forms with various modifications and improvements based on the knowledge of those skilled in the art, including the aspects described in the column of the disclosure of the invention.

[0239] In addition, the "section (section, module, unit)" described above can be read as "means" or "circuit". For example, the acquisition unit can be read as an acquisition means or an acquisition circuit.

Explanation of Reference Numerals

[0240] 1 Information processing apparatus 2 Terminal device 3 Operator terminal 10 Communication unit 11 Storage unit 12 Processing unit 20 Reception unit 21 Acquisition unit 22 Decision unit 23 Generation unit 24 Provision unit 25 Estimation unit 26 Selection unit N Network

Claims

1. An acquisition unit that acquires evaluation information indicating an evaluation of a service using generated information generated using a generative AI based on a prompt including information on a plurality of items; An estimation unit that estimates the importance of the plurality of items based on the evaluation information acquired by the acquisition unit, characterized in that the information processing apparatus comprises: An information processing apparatus characterized by the above.

2. The estimation unit: Estimates the importance of each of the plurality of items, characterized in that the information processing apparatus according to claim 1 comprises: The information processing apparatus according to claim 1, characterized in that the information processing apparatus comprises:

3. The estimation unit: Estimates the importance of each combination of the plurality of items, characterized in that the information processing apparatus according to claim 1 comprises: The information processing apparatus according to claim 1, characterized in that the information processing apparatus comprises:

4. The estimation unit: Estimates the importance of the plurality of items based on evaluation information indicating an evaluation of a service using a plurality of pieces of generated information each generated using a generative AI based on a corresponding prompt among a plurality of the prompts including information on different combinations of the plurality of items, characterized in that the information processing apparatus according to any one of claims 1 to 3 comprises: The information processing apparatus according to any one of claims 1 to 3, characterized in that the information processing apparatus comprises:

5. A selection unit that determines information on one or more items to be included in a new prompt among the information on the plurality of items based on the importance of the plurality of items estimated by the estimation unit, characterized in that the information processing apparatus according to any one of claims 1 to 3 comprises: The information processing apparatus according to any one of claims 1 to 3, characterized in that the information processing apparatus comprises:

6. A generation unit that generates a new prompt including information on one or more items selected by the selection unit; A provision unit that provides the service using the generated information generated using the generative AI based on the new prompt generated by the generation unit, characterized in that the information processing apparatus according to claim 5 comprises: The information processing apparatus according to claim 5, characterized in that the information processing apparatus comprises:

7. The acquisition unit: Acquires the evaluation information for each context of a user who uses the service; The estimation unit: Estimates the importance of the plurality of items for each context of the user; The selection unit: Selects information on items to be included in a new prompt among the information on the plurality of items for each context of the user, characterized in that the information processing apparatus according to claim 5 comprises: The information processing apparatus according to claim 5, characterized in that the information processing apparatus comprises:

8. The evaluation of the service: Is one or more of an evaluation by a user who uses the service and an evaluation based on the behavior of a user who uses the service, characterized in that the information processing apparatus according to any one of claims 1 to 3 comprises: The information processing apparatus according to any one of claims 1 to 3, characterized in that the information processing apparatus comprises:

9. The service: Is an advertisement distribution service; ​ The advertisement content distributed by the advertisement distribution service The information processing apparatus according to claim 6, characterized in that

10. An information processing method executed by a computer, comprising An acquisition step of acquiring evaluation information, which is information indicating an evaluation of a service using generated information generated using a generation AI based on a prompt including information of a plurality of items; An estimation step of estimating the importance of the plurality of items based on the evaluation information acquired in the acquisition step The information processing method is characterized by the above

11. An acquisition procedure for acquiring evaluation information, which is information indicating an evaluation of a service using generated information generated using a generation AI based on a prompt including information of a plurality of items; An estimation procedure for estimating the importance of the plurality of items based on the evaluation information acquired by the acquisition procedure, and causing a computer to execute the estimation procedure The information processing program is characterized by the above

Images