Information processing apparatus, method and program

Abstract

To present an output of a generative AI model together with a corresponding advertisement, and offer a reward.SOLUTION: An information processing apparatus includes: an acquisition unit which inputs a prompt including a user input to a generative AI model to obtain an output of the generative AI model; a search unit which searches a database storing advertisement data, for advertisement data similar to the user input or the output of the generative AI model; an output editing unit which edits the output of the AI model so as to include the searched advertisement data as a part; a presentation unit which presents a result of editing the output of the generative AI model; and a reward offering unit which offers a reward corresponding to the searched advertisement data to a system operator.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

[0002] 2. Description of the Related Art Conventionally, an advertisement distribution device that matches the content of an affiliate site with the content of advertising content posted on that site has been known (Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2006-048430 Summary of the Invention [Problem to be solved by the invention]

[0004] The present invention aims to provide an information processing device, method, and program that can present the output of a generative AI model together with a corresponding advertisement and provide a reward. [Means for solving the problem]

[0005] A first aspect of the present disclosure is an information processing device that includes an acquisition unit that acquires at least one of a prompt including user input and the output of a generative AI model; a search unit that searches a database that stores advertising data for advertising data that is similar to the user input input to the generative AI model or the output of the generative AI model; an output editing unit that edits the output of the generative AI model to include the searched advertising data as part of it; a presentation unit that presents the edited result of the output of the generative AI model to a user; and a reward granting unit that grants a reward corresponding to the searched advertising data to a reward grantee.

[0006] A second aspect of the present disclosure is an information processing method, in which an acquisition unit acquires at least one of a prompt including user input and the output of a generative AI model, a search unit searches a database storing advertising data for advertising data similar to the user input input to the generative AI model or the output of the generative AI model, an output editing unit edits the output of the generative AI model to include the searched advertising data as part of it, a presentation unit presents the edited result of the output of the generative AI model to a user, and a reward granting unit grants a reward corresponding to the searched advertising data to a reward grantee.

[0007] A third aspect of the present disclosure is an information processing program for causing a computer to function as the information processing device of the first aspect. [Effects of the Invention]

[0008] According to the disclosed technology, the output of the generative AI model can be presented along with a corresponding advertisement to provide a reward. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a block diagram showing a configuration of an information processing system according to an embodiment of the present invention; [Figure 2] FIG. 2 is a schematic block diagram of an example of a computer that functions as a management server according to the present embodiment. [Figure 3] FIG. 2 is a block diagram showing the configuration of a management server according to the present embodiment. [Figure 4] FIG. 10 is a diagram showing an example of presenting the edited output of a generative AI model to a user. [Figure 5] 10 is a flowchart showing the contents of a search processing routine of the management server according to the present embodiment. [Figure 6] FIG. 10 is a diagram illustrating an example of generating a search query using a generative AI model and executing a search. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

[0011] <Outline of this embodiment> In conventional search-based advertising, ads were presented in a fixed advertising area separate from the content for each service, particularly outside the content area, depending on the user's search intent (for example, the "Ads" section of Google (registered trademark) search).

[0012] However, with the advent of generative AI models, the generated results themselves will become the output content, narrowing the user's focus area, and it is expected that there will be more situations where it is better or even necessary to display advertisements in the generated results.

[0013] For example, with GPTs, you can publish the customized GPT you created, but the results are kept within the generated field. By displaying advertisements within the generated field, developers and operators can generate revenue.

[0014] This goes beyond text generation. For example, if there is an AI that generates videos specializing in cooking, ads will be inserted as the videos are generated. The ads themselves do not need to be videos.

[0015] The difference between this and static content and ads for viewers on sites like YouTube (registered trademark) is that ads are added at the time of creation, in line with the creator's intent. Conventional technology can be used to add ads to static content once it has been created.

[0016] <System Configuration of This Embodiment> As shown in Fig. 1, an information processing system 100 according to the first embodiment includes a management server 10 installed on the side of a service management company and a user terminal 24 operated by a user. The management server 10 is an example of an information processing device. For simplicity, Fig. 1 shows an example in which two user terminals 24 are provided, but three or more user terminals 24 may be provided.

[0017] The management server 10 and the user terminal 24 are connected via a network 26 such as the Internet.

[0018] The user terminal 24 is composed of a smartphone terminal, a mobile phone, a PDA (Personal Digital Assistant) terminal, a notebook computer terminal, or the like.

[0019] FIG. 2 is a block diagram showing the hardware configuration of the management server 10 of this embodiment.

[0020] 2, the management server 10 includes a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, a storage 14, an input unit 15, a display unit 16, and a communication interface (I / F) 17. Each component is connected to each other via a bus 19 so as to be able to communicate with each other.

[0021] The CPU 11 is a central processing unit that executes various programs and controls each component. That is, the CPU 11 reads programs from the ROM 12 or the storage 14 and executes the programs using the RAM 13 as a work area. The CPU 11 controls the above components and performs various arithmetic processing in accordance with the programs stored in the ROM 12 or the storage 14. In this embodiment, the ROM 12 or the storage 14 stores programs for performing various processing.

[0022] The ROM 12 stores various programs and various data. The RAM 13 temporarily stores programs or data as a working area. The storage 14 is configured with an HDD (Hard Disk Drive) or SSD (Solid State Drive) and stores various programs including the operating system and various data.

[0023] The input unit 15 includes a pointing device such as a mouse and a keyboard, and is used to perform various inputs.

[0024] The display unit 16 is, for example, a liquid crystal display, and displays various information. The display unit 16 may also function as the input unit 15 by adopting a touch panel system.

[0025] The communication interface 17 is an interface for communicating with other devices, and uses standards such as Ethernet (registered trademark), FDDI, and Wi-Fi (registered trademark).

[0026] Next, a description will be given of the functional configuration of the management server 10. As shown in Fig. 3, the management server 10 functionally includes an acquisition unit 30, an extraction unit 32, a subset creation unit 36, a sentence generation unit 38, an evaluation unit 40, a subset extraction unit 42, a search unit 46, an advertisement editing unit 48, an output editing unit 50, a presentation unit 52, a reward granting unit 54, and a storage unit 62.

[0027] The acquisition unit 30 inputs a prompt including a user input received from the user terminal 24 to a generative AI (Artificial Intelligence) model 62A, and acquires the output of the generative AI model 62A.

[0028] The model storage unit 62 stores the generative AI model 62A. The generative AI model 62A may be stored in an external device separate from the management server 10. In this case, the management server 10 may transmit a prompt to the external device and acquire the output of the generative AI model 62A.

[0029] Here, an example of the generative AI model 62A is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the generative AI 62A. The generative AI model is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the generative AI model 62A, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The generative AI model 62A performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0030] The extraction unit 32 extracts a plurality of keywords from an input sentence that includes a user input received from the user terminal 24 and the output of the acquired generative AI model 62A.

[0031] For example, the extraction unit 32 extracts multiple keywords K from the input sentence S using the generative AI model 62A. As an example, the extraction unit 32 inputs a prompt such as "Please guess three to four keywords necessary to explain this input" to the generative AI model 62A, and obtains multiple keywords K from the output of the generative AI model 62A (see FIG. 6).

[0032] The keyword extraction method is not limited to the method using the generative AI model, and other extraction methods may be used.

[0033] Furthermore, multiple keywords may be extracted from an input sentence representing a user input received from the user terminal 24. In this case, the input sentence representing the user input received from the user terminal 24 may be taken as the original input sentence. Alternatively, multiple keywords may be extracted from an input sentence representing the output of the acquired generative AI model 62A. In this case, the input sentence representing the output of the acquired generative AI model 62A may be taken as the original input sentence.

[0034] The subset creation unit 36 creates a plurality of keyword subsets from the extracted plurality of keywords.

[0035] Specifically, the subset creation unit 36 mechanically combines the extracted keywords to create a subset Ks(n) (see FIG. 6). For example, as a total of all combinations for four keywords, 14 (=4C4+4C3+4C2+4C1) subsets Ks(1) to Ks(14) are created.

[0036] Here, when there are a large number of keywords, it is possible to impose restrictions on the number of keywords or a threshold value, rather than all combinations of keywords. For example, multiple subsets may be created with the number of keywords included in each subset set within a predetermined range (for example, 3 to 4, or 3 or less).

[0037] The sentence generation unit 38 inputs a prompt to the generative AI model 62A for each of a plurality of subsets, instructing the model 62A to generate a sentence from the subset without including any keywords not included in the subset, and obtains the generated sentence from the output of the generative AI model 62A.

[0038] Specifically, each subset is used as input, and a subset generation sentence S'(n) is generated so as to better represent the original input sentence and not use keywords that are not included in the subset (see Figure 6). For example, for subsets Ks(1) to Ks(14), subset generation sentences S'(1) to S'(14) are generated.

[0039] Here, the prompt may indicate that if it is difficult to generate a sentence from the subset, no sentence will be generated.

[0040] For example, a prompt such as "Based on a subset of keywords, generate a sentence that is as close as possible to the original input sentence. However, the generated sentence must be based only on the subset of keywords and cannot include other keywords. If it is difficult to generate a sentence using only the subset of keywords, output 'N / A'" is input to the generative AI model 62A, and the generated sentence is obtained from the output of the generative AI model 62A.

[0041] The evaluation unit 40 evaluates the relevance between the input sentence S and the generated sentence S'(n) for each of the plurality of subsets.

[0042] The subset extraction unit 42 extracts at least one subset as an important keyword set in descending order of relevance.

[0043] Specifically, the subset extraction unit 42 extracts, in descending order of the score indicating the relevance, subsets whose scores indicating the relevance are equal to or greater than a threshold as important keyword sets.

[0044] The search unit 46 uses the extracted set of important keywords in descending order of relevance as a search query to perform a search on a database that stores advertisement data.

[0045] Specifically, the search unit 46 uses the extracted set of important keywords in descending order of relevance score as a search query, and performs a search using a search engine, with an advertising database that stores advertising data as the search target database.

[0046] The search unit 46 further selects one of the searched advertisement data based on the relevance of the extracted set of important keywords and the similarity of the searched advertisement data to the search query.

[0047] For example, if the similarity for four keyword sets is 0.3 and the similarity for three keyword sets is 0.5, then one of the searched advertising data is selected through machine learning, rules, or AB testing (human evaluation), taking into consideration both the relevance and similarity of the keyword sets.

[0048] Furthermore, the search unit 46 may select N sets of extracted important keywords in descending order of relevance, and use each of these as a search query to perform a search on a database that stores advertisement data.

[0049] Alternatively, the search unit 46 may use all of the extracted sets of important keywords, regardless of order, as search queries to perform a search on a database that stores advertisement data.

[0050] The advertisement editing unit 48 edits the retrieved advertisement data so that it corresponds to the output of the generation AI model 62 A. Specifically, the advertisement editing unit 48 inputs a prompt to the generation AI model 62 A instructing the generation AI model 62 A to edit the retrieved advertisement data so that it corresponds to the output of the generation AI model 62 A, and obtains the edited result of the retrieved advertisement data from the output of the generation AI model 62 A.

[0051] The output editing unit 50 edits the output of the generation AI model 62A so as to include as part of the editing result of the searched advertising data. Specifically, the output editing unit 50 inputs a prompt to the generation AI model 62A instructing it to edit the output of the generation AI model 62A so as to include as part of the editing result of the searched advertising data, and obtains the editing result of the output of the generation AI model 62A from the output of the generation AI model 62A.

[0052] The presentation unit 52 presents the edited result of the output of the generative AI model 62A to the user terminal 24. Here, the advertising data includes a website link. As shown in Fig. 4, the edited result of the output of the generative AI model 62A presented to the user terminal 24 includes a website link of the advertising data.

[0053] The reward granting unit 54 grants a reward corresponding to the searched advertising data to the reward grantee. Specifically, the reward granting unit 54 grants a reward to the reward grantee when an edited result of the output of the generative AI model 62A is presented to a user, or when a website link included in the edited result of the output of the generative AI model presented to the user terminal 24 is clicked. The reward grantee is, for example, a system operator. Here, the system operator may be a company that provides the generative AI model itself, or a developer that provides a service using the generative AI model 62A (for example, a service provider using GPTs).

[0054] <Operation of information processing system> Next, the operation of the information processing system 100 according to this embodiment will be described.

[0055] When the management server 10 receives a user input representing an instruction statement for instructing the generated AI model 62A to perform processing from the user terminal 24, the management server 10 executes a search processing routine shown in Fig. 5. The search processing routine is an example of an information processing method.

[0056] In step S100, the acquisition unit 30 acquires a user input received from the user terminal 24.

[0057] In step S102, the acquisition unit 30 inputs a prompt including a user input received from the user terminal 24 to the generative AI model 62A, and acquires the output of the generative AI model 62A.

[0058] In step S104, the extraction unit 32 extracts a plurality of keywords from an input sentence that includes the user input received from the user terminal 24 and the output of the acquired generative AI model 62A.

[0059] In step S106, the subset creating unit 36 creates a plurality of keyword subsets from the extracted plurality of keywords.

[0060] In step S108, the sentence generation unit 38 inputs a prompt to the generative AI model 62A for each of the multiple subsets, instructing the generative AI model 62A to generate a sentence from the subset without including any keywords that are not included in the subset, and obtains the generated sentence from the output of the generative AI model 62A.

[0061] In step S110, the evaluation unit 40 evaluates the relevance between the input sentence S and the generated sentence S'(n) for each of the plurality of subsets.

[0062] In step S112, the subset extraction unit 42 extracts at least one subset as an important keyword set in descending order of relevance.

[0063] In step S114, the search unit 46 uses the extracted set of important keywords in descending order of relevance as a search query to perform a search on an advertisement database that stores advertisement data.

[0064] In step S116, the search unit 46 further selects one of the searched advertisement data based on the relevance of the extracted set of important keywords and the similarity of the searched advertisement data to the search query.

[0065] In step S118, the advertisement editing unit 48 edits the searched advertisement data so as to correspond to the output of the generative AI model 62A.

[0066] In step S120, the output editing unit 50 edits the output of the generative AI model 62A so as to include the edited results of the searched advertisement data as a part of the output.

[0067] In step S122, the presentation unit 52 outputs to the user terminal 24 a screen presenting the edited results of the output of the generated AI model 62A.

[0068] In step S124, the reward granting unit 54 grants a reward corresponding to the searched advertisement data to the reward grantee, and ends the search processing routine.

[0069] As described above, the information processing system according to this embodiment inputs a prompt including a user input to a generative AI model, obtains the output of the generative AI model, searches a database storing advertising data for advertising data similar to the user input or the output of the generative AI model, edits the output of the generative AI model to include the searched advertising data as part of it, presents the edited output of the generative AI model to a user, and awards a reward corresponding to the searched advertising data to a reward recipient. This allows the output of the generative AI model to be presented together with the corresponding advertisement, and a reward to be awarded.

[0070] Furthermore, multiple keyword subsets are created from multiple keywords extracted from an input sentence including a user input and the output of a generative AI model, and for each subset, a sentence is generated using a generative AI model so as not to include keywords not included in the subset, and a search is performed on an advertising database using the subsets in descending order of relevance to the original input sentence as important search queries. This makes it possible to create search queries from input sentences using a generative AI model and perform searches.

[0071] The present invention is not limited to the above-described embodiment, and various modifications and applications are possible without departing from the spirit and scope of the present invention.

[0072] For example, if the output of the generating AI model 62A acquired by the acquisition unit 30 includes search results for external knowledge, and the search results for the external knowledge include advertising data, the search unit 46 may use the advertising data included in the search results for the external knowledge in place of the advertising data being searched, and edit the output of the generating AI model to include the advertising data as part of it.

[0073] Specifically, when the generation AI model 62A generates output in the acquisition unit 30 in the form of RAG (Retrieval-Augmented Generation), the search results in RAG include a blog article (e.g., an article about health), and there is an advertisement (e.g., a supplement) published in that blog article.

[0074] The advertisements in this blog post are advertisements that the blog creator wants to display (advertisements that have been approved). The output of the generative AI model may be edited to include the advertising data embedded in the blog post as part of the advertisement. This allows the output of generative AI model 62A to be presented in a state where it is filtered to only the advertisements that the blog creator wants to display (advertisements that have been approved), and also makes it easy to award rewards to the owner of the blog post.

[0075] For example, if a blog article included in the search results of RAG contains banner advertisements, video advertisements, etc., it may be possible to continue displaying these as they are.

[0076] In addition, if a blog creator provides information to the AI generation service system and registers an account, they can simply provide proof of ownership of the blog and specify their bank details. After the reward corresponding to the advertising data is first given to the system operator, at least a portion of that reward can be returned to the blog creator.

[0077] Additionally, the display of the blog post may be embedded and displayed as part of the output of the generative AI model. In this case, the display of an advertisement in the blog post is also reproduced on the generative AI service. For example, only the advertisement portion of the blog post may be inherited and displayed on the generative AI service side. Alternatively, the blog page may be opened in a pop-up.

[0078] Furthermore, when a blog article is included in the search results of external knowledge, and the generation AI model determines that the blog article and the blog article's poster's past blog articles do not violate predetermined terms of use or prohibited matters, the presentation unit 52 may present to the user an edited result of the output of the generation AI model including the search results of external knowledge. Specifically, the presentation unit 52 inputs the terms of use and other prohibited matters of the service platform on which the data is ultimately posted or the terms of use and other prohibited matters of the service provider that provides the advertising database, the output of the generation AI model including the blog article, and the blog article's past blog articles, and determines whether or not there is a conflict in the generation AI model. If there is a conflict, the presentation unit 52 does not present the edited result of the output of the generation AI model.

[0079] The search unit 46 may also select one of the searched advertisement data based on a history of actions taken by the user on the advertisement data. Specifically, based on the history of actions taken by the user on the advertisement data, the search unit 46 may preferentially select advertisement data clicked by the user from among advertisement data previously output. For example, the search unit 46 may select one of the searched advertisement data based on the following score S(Q, K, A, P):

[0080] S(Q,K,A,P)=α·S_rel(Q,K)+β·S_search(K,A)+γ·S_person(A,P)

[0081] where Q is a variable indicating an input sentence, K is a variable indicating one keyword set, A is a variable indicating one advertising data, and P is a variable indicating a user. Furthermore, S_rel(Q,K) is the relevance between the input sentence Q and the keyword set K, S_search(K,A) is the similarity of the searched advertising data A to the keyword set K, and S_person(A,P) is a score indicating whether or not user P previously clicked on A or an advertisement similar to A. α, β, and γ are predetermined constants.

[0082] This allows for more appropriate advertisements to be displayed based on, for example, not only the prompt the user has just input, but also the prompts the user has input in the past. It also makes it easier to present advertisements that are useful to the user.

[0083] Furthermore, the search unit 46 may select one of the searched advertising data based on the relevance of the extracted set of important keywords and a predetermined importance level for the searched advertising data. If each piece of advertising data has an importance level, the advertising data may be selected by taking into account both the relevance of the set of important keywords and the importance level of the advertising data. In this way, when taking into account both the relevance of the set of important keywords and the importance level of the advertising data, any method may be used, such as linear weighting or leaving the weighting decision to the generation AI model 62A.

[0084] In the above embodiment, the searched advertising data is edited to correspond to the output of the generative AI model, but the present invention is not limited to this. The searched advertising data may be presented as is without being edited.

[0085] Furthermore, in the above embodiment, an example has been described in which the relevance between the input sentence S and the generated sentence S'(n) is evaluated, but this is not limiting. The sentence generation unit 38 may acquire embeddings of the generated sentence from the output of the generative AI model 52A, and the evaluation unit 40 may evaluate the relevance between the embeddings of the input sentence and the embeddings of the generated sentence for each of a plurality of subsets. In this way, the sentence generation unit 38 does not necessarily need to generate S' as a sentence; embeddings may be used as long as they allow for a comparison of Ks(n) with the original S.

[0086] The evaluation unit 40 may further evaluate the coverage of the input sentence for a set of sentences generated for each subset whose relevance score is equal to or greater than a threshold. If the coverage score is less than the threshold, the evaluation unit 40 adds the sentences generated for the subset whose relevance score is less than the threshold to the set of sentences generated for each subset whose relevance score is equal to or greater than the threshold, and evaluates the coverage of the result for the input sentence. Furthermore, the evaluation unit 40 identifies a subset for which the coverage score is equal to or greater than the threshold when the addition of the sentences generated for the subset whose relevance score is less than the threshold. The search unit 44 prioritizes the identified subset as a search query and performs a search. In this way, coverage may be taken into account. If S″ obtained from the set of adopted sentences S′ does not cover the input sentence S but can be covered by adding S′(n), S′(n) can be prioritized and adopted as a search query.

[0087] In addition, the edited output of the generative AI model presented to the user may be configured not to violate predetermined terms of use or prohibited items. Specifically, the edited output of the generative AI model is input to the terms of use and other prohibited items of the service platform on which the data will ultimately be published, and the generative AI model determines whether or not there is a violation. If there is a violation, the edited output is not presented. Alternatively, the search unit 46 may use the extracted set of important keywords as a search query to search the advertising database, adding potentially related prohibited items as text to the search query, and searching for advertising data that matches those conditions.

[0088] Furthermore, the presentation unit 52 may present the edited output of the generative AI model to the user when the generative AI model determines that the prompt including the user input or the output of the generative AI model is not misleading or does not violate predetermined terms of use or prohibitions. Specifically, for example, the presentation unit 52 may take a prompt entered by a system operator as input, determine whether the prompt is misleading in the generative AI model, and if so, not present the edited output of the generative AI model. Alternatively, the presentation unit 52 may take a prompt including user input or the output of the generative AI model as input, determine whether the prompt is misleading in the generative AI model, and if so, not present the edited output of the generative AI model. Alternatively, the presentation unit 52 may take a prompt including user input or the output of the generative AI model as input, determine whether the prompt violates predetermined terms of use or prohibitions in the generative AI model, and if so, not present the edited output of the generative AI model. [Explanation of symbols]

[0089] 10 Management Server 11 CPU 14. Storage 15 Input section 16 Display section 17 Communication Interface 24 User terminals 30 Acquisition Department 32 Extraction part 36 Subset Creation Unit 38 Sentence generation section 40 Evaluation Department 42 Subset extraction part 46 Search Section 48 Advertising Editorial Department 50 Output Editorial Department 52 Presentation section 54 Reward Division 62 Storage section 62 Model memory section 62A Generative AI Model< / url:>

Claims

1. an acquisition unit that acquires at least one of a prompt including a user input and an output of a generated AI model; a search unit that searches a database storing advertising data for advertising data similar to the user input input to the generative AI model or the output of the generative AI model; an output editing unit that edits the output of the generative AI model so as to include the searched advertising data as a part of it; A presentation unit that presents an edited result of the output of the generative AI model to a user; a reward granting unit that grants a reward corresponding to the searched advertisement data to a reward grantee; An information processing device comprising:

2. Further comprising an advertisement editing unit that edits the searched advertisement data so as to correspond to the output of the generative AI model; The information processing device according to claim 1 , wherein the output editing unit edits the output of the generative AI model so as to include as part of the edited result of the searched advertising data.

3. The information processing device according to claim 1 , wherein the search unit searches for advertising data similar to the user input or the output of the generating AI model using similarity of embedding vectors or similarity of keywords.

4. 3. The information processing device according to claim 2, wherein the advertising editing unit inputs a prompt to the generative AI model instructing the generative AI model to edit the searched advertising data to correspond to the output of the generative AI model, and obtains the editing results of the searched advertising data from the output of the generative AI model.

5. The information processing device described in claim 1, wherein the output editing unit inputs a prompt to the generative AI model instructing it to edit the output of the generative AI model so as to include the searched advertising data as part of it, and obtains the edited result of the output of the generative AI model from the output of the generative AI model.

6. the advertising data includes a website link; The information processing device of claim 1, wherein the reward granting unit grants the reward to the reward grantee when the edited result of the output of the generative AI model is presented to the user, or when a link to the website included in the edited result of the output of the generative AI model presented to the user is clicked.

7. an extraction unit that extracts a plurality of keywords from the user input or the output of the generative AI model acquired by the acquisition unit; a subset creation unit that creates a plurality of keyword subsets from the extracted plurality of keywords; An evaluation unit that evaluates the relevance of each of the plurality of subsets between the user input or the output of the generative AI model acquired by the acquisition unit and the subset; a subset extraction unit that extracts at least one of the subsets as an important keyword set in descending order of the relevance, The information processing device according to claim 1 , wherein the search unit uses the extracted set of important keywords as a search query to search a database storing advertising data for advertising data related to the user input or the output of the generative AI model.

8. The evaluation unit obtains an embedding of the generated sentence from the output of the generative AI model, The information processing device according to claim 7, wherein for each of the plurality of subsets, the relevance between the embedding of the user input or the output of the generative AI model acquired by the acquisition unit and the embedding of the generated sentence is evaluated.

9. 8. The information processing device according to claim 7, wherein the search unit further selects one of the searched advertisement data based on the relevance of the extracted set of important keywords and the similarity of the searched advertisement data.

10. The acquisition unit inputs a prompt including a user input to a generative AI model and acquires an output of the generative AI model including a search result of external knowledge; The information processing apparatus according to claim 7 , wherein the search unit further presents the advertisement data included in the search results of the external knowledge when the advertisement data is included in the search results of the external knowledge.

11. The information processing device according to claim 7 , wherein the search unit selects one of the searched advertisement data based on a history of actions taken by the user on the advertisement data.

12. 8. The information processing device according to claim 7, wherein the search unit further selects one of the searched advertising data based on the relevance of the extracted set of important keywords and a predetermined importance of the searched advertising data.

13. The information processing device according to claim 1 , wherein the editing result of the output of the generative AI model presented to the user is determined by the generative AI model to be an editing result that does not violate predetermined terms of use or prohibited items.

14. The information processing device of claim 1, wherein the presentation unit presents the edited result of the output of the generative AI model to the user when the generative AI model determines that the prompt including the user input or the output of the generative AI model is not misleading or does not violate predetermined terms of use or prohibited items.

15. The acquisition unit acquires at least one of a prompt including a user input and an output of the generating AI model; A search unit searches a database storing advertisement data for advertisement data similar to the user input input to the generative AI model or the output of the generative AI model; an output editing unit editing the output of the generative AI model to include the searched advertising data as a part of the output; A presentation unit presents an edited result of the output of the generative AI model to a user, A reward granting unit grants a reward corresponding to the searched advertisement data to a reward grantee. Information processing methods.

16. An information processing program for causing a computer to function as the information processing device according to any one of claims 1 to 14.

Images