Information processing device, information processing method and information processing program

Abstract

To provide an information processing device capable of improving the efficiency of editing operation.SOLUTION: A server device 2 being an information processing device includes means for accepting a retrieval condition from a terminal device 3 operated by a user, means for extracting content to be referred to in generating a content idea from disclosed information on the basis of the accepted retrieval condition, means for generating a content idea by an editing format by inputting a first instruction to a large-scale language model, the a first instruction requesting, to the large-scale language model, generation of a content idea by using the extracted content and content to be inputted and generation according to an editing format for determining what kind of an editing policy is adopted for generating the content idea in generating the content idea, and means for providing a terminal device 3 with the generated content idea.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

[0002] Patent Document 1 below discloses a system for displaying newspaper articles. This system displays a list of the top 10 news items of the week selected from newspaper articles for the past week, and displays an article of an item selected from the list on the newspaper page display unit. In other words, the system of Patent Document 1 displays the articles published in the newspaper as they are. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2004-178370 Summary of the Invention [Problem to be solved by the invention]

[0004] However, when manually editing newspaper articles, such as summarizing and reconstructing them, it takes a great deal of time and effort to select the original articles and edit them to fit the post-editing article format, etc. This workload places a heavy burden on the editors performing the editing work.

[0005] The present invention has been made to solve the above-mentioned problems, and one of its objects is to provide an information processing device, an information processing method, and an information processing program that can improve the efficiency of editing work. [Means for solving the problem]

[0006] An information processing device according to one embodiment of the present invention comprises: means for extracting content to be referenced when generating a proposed content from published information based on search conditions received from a terminal device operated by a user; means for inputting a first instruction to a large-scale language model requesting the large-scale language model to generate a proposed content using the extracted content and input content, and to generate the proposed content in accordance with an editing format that defines what editing policy the proposed content will be based on when generating the proposed content; and means for generating the proposed content in the editing format; and means for providing the generated proposed content to the terminal device.

[0007] An information processing method according to another aspect of the present invention is a method executed by a processor, and includes the steps of: extracting content to be referenced when generating proposed content from published information based on search conditions received from a terminal device operated by a user; inputting a first instruction to a large-scale language model requesting the large-scale language model to generate proposed content using the extracted content and input content, and to generate the proposed content in accordance with an editing format that defines what editing policy will be used to generate the proposed content; and generating proposed content in the editing format; and providing the generated proposed content to the terminal device.

[0008] An information processing program according to another aspect of the present invention causes a computer to function as: means for extracting content to be referenced when generating a proposed content from published information based on search conditions received from a terminal device operated by a user; means for inputting a first instruction to a large-scale language model requesting the large-scale language model to generate a proposed content using the extracted content and input content, and to generate the proposed content in accordance with an editing format that defines what editing policy the proposed content will be based on when generating the proposed content; and means for generating the generated proposed content in the editing format; and means for providing the generated proposed content to the terminal device.

[0009] According to these aspects, content to be referenced when generating content suggestions is extracted based on search conditions received from a terminal device, the extracted content and the first instruction are input into a large-scale language model, content suggestions are generated in the editing format included in the first instruction, and are provided to the terminal device.

[0010] This allows users to reduce the work of selecting source content and adjusting the format of edited content when creating content proposals, thereby reducing the workload of editing work and the burden on users who perform editing work. [Effects of the Invention]

[0011] According to the present invention, it is possible to provide an information processing device, an information processing method, and an information processing program that can improve the efficiency of editing work. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a diagram illustrating a configuration of an information processing system including a server device that is an information processing device according to an embodiment; [Figure 2] FIG. 2 is a diagram illustrating a configuration of a server device. [Figure 3] FIG. 2 is a diagram illustrating a configuration of a terminal device. [Figure 4] 10 is an example of a screen displayed on a terminal device. [Figure 5] 10 is an example of a screen displayed on a terminal device. [Figure 6] 10 is a flowchart illustrating an example of an operation when providing an article suggestion. DETAILED DESCRIPTION OF THE INVENTION

[0013] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Note that the same elements are given the same reference numerals and redundant description will be omitted.

[0014] [Outline of information processing system] An information processing system including an information processing device according to an embodiment will be described with reference to Fig. 1. The information processing system 1 includes, for example, a server device 2 which is an information processing device, and a terminal device 3.

[0015] The information processing system 1 according to the embodiment is a system that realizes, for example, a function of providing a user with proposed content. The proposed content is a draft of content that is provided based on existing content. The content is, for example, news, columns, essays, and other articles published in newspapers and magazines, including electronic versions. The existing content is, among the content, an article that has been published in a newspaper or magazine that has already been published. The users include, for example, an editor who edits the article, or a person in charge of the article.

[0016] The server device 2 is, for example, a computer device that executes processes to realize the function of providing the content proposals described above, etc. The terminal device 3 is, for example, a computer device used by a user who requests content proposals.

[0017] The terminal device 3 may be, for example, a PC (personal computer), a notebook PC, a tablet terminal, a smartphone, or other terminal device.

[0018] The terminal device 3 may or may not have installed thereon an application (hereinafter also referred to as an "app") for providing content suggestions.

[0019] If an app for providing content suggestions is not installed on the terminal device 3, it is preferable to exchange data between the terminal device 3 and the server device 2, for example, by accessing a specific site on the server device 2 and using a screen displayed on the terminal device 3 by a web app for providing content suggestions running on the website.

[0020] On the other hand, when an app for providing content suggestions is installed on the terminal device 3, it is preferable to exchange data between the terminal device 3 and the server device 2, for example, using a screen displayed on the terminal device 3 by executing the installed app.

[0021] In the present embodiment, a case will be described in which an application for providing content suggestions is not installed on the terminal device 3, but a web application for providing content suggestions is used.

[0022] The server device 2 and the terminal device 3 are configured to be able to communicate with each other via a network N. The network N may be, for example, the Internet, a LAN, a dedicated line, a telephone line, an in-house network, a mobile communication network, WiFi (Wireless Fidelity), Bluetooth (registered trademark), other communication lines, or a combination thereof, and may be wired or wireless.

[0023] [Server device configuration] As shown in FIG. 2, the server device 2 includes, as its physical configuration, a processor 21, a communication interface 22, and a storage device 23, for example.

[0024] The processor 21 is, for example, a CPU (Central Processing Unit) or an MPU (Micro Processing Unit), etc. The processor 21 executes a program 231 stored in the storage device 23, thereby functioning as a control unit 211 that realizes various functions of the server device 2.

[0025] The control unit 211 realizes, for example, a function of displaying various screens on the terminal device 3 in accordance with requests received from the terminal device 3, a function of receiving content input to various screens from the terminal device 3, a function of executing predetermined processing based on information received from the terminal device 3, and a function of transmitting the processing results to the terminal device 3. Examples of various screens will be described later.

[0026] The communication interface 22 connects to the network N and functions as a communication unit that communicates with other devices on the network N.

[0027] The storage device 23 is a computer-readable recording medium such as a disk drive or a semiconductor memory. The storage device 23 functions as a storage unit that stores a program 231 for implementing various functions of the server device 2, various data 232 used by the program 231, and a large language model (hereinafter also referred to as "LLM") 233. Note that the LLM 233 does not necessarily have to be stored in the server device 2, and may be stored in an external device or system.

[0028] The various data 232 includes, for example, data constituting information about existing content, including existing content.

[0029] LLM233 is a generative AI model that generates content ideas based on input prompts. For example, GPT-3.5, GPT-3.5-16K, GPT-4, etc. can be used as LLM233.

[0030] [Terminal device configuration] As shown in FIG. 3, the terminal device 3 includes, as its physical configuration, a processor 31, a communication interface 32, a storage device 33, an input device 34, and a display device 35, for example.

[0031] The processor 31 is, for example, a CPU, an MPU, etc. The processor 31 executes a program 331 stored in the storage device 33, thereby functioning as a control unit 311 that realizes various functions of the terminal device 3.

[0032] The control unit 311 realizes, for example, a function of transmitting various requests to the server device 2 in accordance with operational instructions from an editor operating the terminal device 3 and displaying various screens on the display device 35, a function of transmitting content input into various screens to the server device 2 in accordance with operational instructions from the editor, a function of executing predetermined processing based on information input into various screens and displaying the processing results on the display device 35, etc. Examples of the various screens will be described later.

[0033] The communication interface 32 connects to the network N and functions as a communication unit that communicates with other devices on the network N.

[0034] The storage device 33 is a computer-readable recording medium such as a disk drive or a semiconductor memory. The storage device 33 functions as a storage unit that stores a program 331 for realizing various functions of the terminal device 3 and various data used by the program 331.

[0035] The input device 34 functions as an input unit that accepts input from an editor. As the input device 34, for example, a touch panel, a pen tablet, a keyboard, a mouse, a microphone, etc. can be used.

[0036] The display device 35 functions as a display unit that displays images, screens, etc. As the display device 35, for example, an organic EL display, a liquid crystal display, etc. can be used.

[0037] [Functions of information processing systems] This section describes various functions of the information processing system 1. In the following explanation, a case will be described in which a draft of a new article (hereinafter also referred to as a "draft article"), which is a proposed content, is generated based on an existing content such as a published digital version of a newspaper or an article published in a newspaper (hereinafter also referred to as a "published article"), and the proposed article is provided to an editor, who is a user. Note that this explanation is merely an example, and the existing content, proposed content, and user are not limited to the examples given.

[0038] The functions of the information processing system 1 are provided, for example, using a screen displayed on the terminal device 3 by a web application for providing content suggestions on the server device 2.

[0039] [Function to extract referenced published articles] The server device 2 extracts published articles to be referenced when generating article ideas, in accordance with instructions from an editor operating the terminal device 3. The published articles are stored in a database as electronic data.

[0040] When an editor requests the server device 2 to extract public articles, the editor specifies conditions for searching public articles. The terminal device 3 adds the conditions for searching public articles specified by the editor (hereinafter also referred to as "search conditions") to an instruction to be sent to the server device 2 when the request to extract public articles is made. The server device 2 extracts public articles to be referenced when generating article suggestions from the database, based on the conditions for searching public articles included in the instruction received from the terminal device 3.

[0041] The conditions for searching for public articles can include various conditions that can be used to search for public articles, such as an identification ID for identifying a public article and keywords for searching for a public article.

[0042] Here, it is preferable to add copyright information to the published articles, and narrow down and extract only published articles whose copyrights are held by the company to which the editor belongs.

[0043] [Function to perform preprocessing for article idea generation process] When generating article ideas for predetermined content, the server device 2 summarizes the extracted published articles as a preprocessing step for generating the article ideas. If there are multiple extracted published articles, it is preferable to summarize them so that the variations in the lengths of the multiple published articles fall within a predetermined range. The contents of each summarized published article may also be combined and summarized. This function may be realized by inputting the extracted published articles into the LLM 233.

[0044] When inputting the extracted published article into the LLM 233, an instruction to request that the published article be summarized is input into the LLM 233. At this time, it is preferable to specify a range of the number of characters to be summarized.

[0045] On the other hand, when there are multiple published articles to be input to the LLM 233, instead of the above instruction, an instruction is input to the LLM 233 to request that each published article be summarized so that the variation in the number of characters of the multiple published articles to be input falls within a predetermined range. At this time, it is preferable to specify an upper and lower limit of the number of characters for determining the predetermined range.

[0046] These instructions are incorporated in advance into a program for the server device 2 to execute pre-processing of the proposed article generation process as prompts to be input to the LLM 233. In other words, these instructions are not input by an editor, but are stored in advance in the server device 2.

[0047] Whether or not the content is a predetermined content can be determined as follows.

[0048] When requesting the server device 2 to generate a draft article, the editor specifies which content the draft article should be generated for. The terminal device 3 includes information for identifying the content specified by the editor as identification information for the generated content in the instruction to be sent to the server device 2 when requesting the generation of the draft article. The server device 2 determines whether the content is the specified content based on the identification information for the generated content included in the instruction received from the terminal device 3.

[0049] [Function to restructure published articles and generate article ideas] When generating a draft article for specific content by reconstructing a published article to generate a draft article, the server device 2 inputs the extracted published article and an instruction received from the terminal device 3 when requesting the generation of a draft article to the LLM 233, and generates a draft article in the editing format included in the instruction. The editing format can be determined as appropriate for each content of the draft article to be generated.

[0050] Here, when the extracted public articles (including summarized public articles) and instructions including specific information of the generated content received from the terminal device 3 when a request to generate an article proposal is input to the LLM233, instructions are input to the LLM233 to request that the input public articles be used to generate an article proposal corresponding to the specific information of the input generated content, and that the article proposal be generated in accordance with an editing format that defines the editing policy according to which the article proposal will be generated.

[0051] This instruction is incorporated in advance as a prompt to be input to the LLM 233 in the program executed by the server device 2 when the user requests the generation of an article draft. In other words, this instruction is not input by the editor, but is stored in advance in the server device 2.

[0052] The editing format determines the editorial policy based on which the themes and arrangements of each article included in the generated content will be generated when the generated content is posted on a medium such as a website, and is determined in advance for each generated content and stored in the storage device 23. Therefore, by identifying the generated content, the editing format of the generated content can also be identified.

[0053] Furthermore, the instructions input to the LLM 233 are not limited to instructions received from the terminal device 3, but may include predetermined instructions for each content of the proposed article to be generated. Examples of the instructions to be added include "summarize why this news is important," "summarize the main points of the article in three points," "summarize the background, cause, and outlook of the news," and "obtain and summarize articles that are frequently accessed by people under 35 years old." The instructions to be added are not limited to these examples, and various conditions required for generating proposed articles can be appropriately defined for each content of the proposed article.

[0054] In this function, the LLM233 extracts important points or topics from the input published article according to predetermined specific conditions, and generates article ideas by reconstructing the important points or topics as commentary articles according to an editing format, etc. It is preferable that the predetermined specific conditions are incorporated into the program in advance as instructions to be added.

[0055] In this case, when the extracted public articles (including summarized public articles) and instructions including specific information about the content to be generated received from the terminal device 3 when a request to generate an article proposal is input to the LLM233, instructions are input to the LLM233 to request that important points and topics (important content) identified according to predetermined specific conditions be extracted from the input public articles, and that the important points and topics be reconstructed as explanatory articles according to an editing format that defines the editing policy by which the article proposal to be generated will be based.

[0056] This instruction is incorporated in advance as a prompt to be input to the LLM 233 in the program executed by the server device 2 when the user requests the generation of an article draft. In other words, this instruction is not input by the editor, but is stored in advance in the server device 2.

[0057] Whether or not the content is specific content for generating an article suggestion by reconstructing a published article can be determined based on specific information of the content to be generated included in the instruction received from the terminal device 3 when a request for generating an article suggestion is made.

[0058] [A function that summarizes published articles and generates article ideas] When generating a draft article for specific content by summarizing published articles to generate draft articles, the server device 2 extracts a predetermined number of published articles, inputs the extracted predetermined number of published articles and instructions received from the terminal device 3 at the time of the request to generate a draft article to the LLM 233, and generates a draft article in the editing format included in the instructions.

[0059] Specifically, the server device 2 extracts a predetermined number of public articles in accordance with the instructions, and the LLM 233 generates article suggestions by summarizing each of the input predetermined number of public articles and formatting the summarized articles according to an editing format. The public articles to be summarized may be all of the predetermined number of public articles, or may be a portion of the predetermined number of public articles selected based on predetermined selection conditions.

[0060] Here, when the extracted predetermined number of published articles and instructions including specific information of the generated content received from the terminal device 3 when the generation of an article proposal is requested are input to the LLM233, instructions are input to the LLM233 to request that each of the predetermined number of published articles be summarized, and that when generating an article proposal using the summarized articles, the article proposal be formatted and generated in accordance with an editing format that defines the editing policy according to which the article proposal will be generated.

[0061] This instruction is incorporated in advance as a prompt to be input to the LLM 233 in the program executed by the server device 2 when the user requests the generation of an article idea. In other words, this instruction is not input by the editor, but is stored in advance in the server device 2.

[0062] [A function to check the external format of an article proposal] After generating an article idea, the server device 2 evaluates the external format of the generated article idea, such as the writing style, based on a predetermined format, and if the evaluation level (for example, a threshold value) is not reached, the server device 2 generates the article idea again. The predetermined format can be determined as appropriate for each content of the article idea to be generated.

[0063] Here, the external format is a format for determining what items make up a sentence of generated content. It is preferable that the items that make up a sentence be distinguishable, for example, by adding headings and subheadings. Specifically, the headings and subheadings may include specific phrases such as "Why is it important?", "What is happening?", "Background so far?", and "Summary." These phrases may be detected from the draft article to distinguish the items that make up a sentence.

[0064] The predetermined format is a standard external format established for evaluating generated content, and is determined for each generated content. Preferably, the predetermined format determines, for example, the items required to configure the target generated content and the order in which those items are arranged.

[0065] The server device 2 evaluates the items constituting the content of the generated article suggestion by comparing them with predetermined items defined in a predetermined format of the content, and if the evaluation score does not reach the standard score, generates the article suggestion again.

[0066] Specifically, the system detects headings and subheadings from the content based on the generated article proposal, and determines whether the detected headings and subheadings match the headings and subheadings defined in the predetermined format of the content, whether they are written in the order defined in the predetermined format, etc. Then, it calculates an evaluation score based on the degree of match between the external format of the content based on the generated article proposal and the predetermined format of the content.

[0067] If the calculated evaluation score does not reach the standard score, the article idea is generated again. The standard score can be set as appropriate, such as 90 points, taking into account the quality of the content, such as the degree of completion, of the generated article idea.

[0068] [Function to make corrections by comparing with published articles] After generating a draft article, the server device 2 compares the generated draft article with the extracted published article and corrects the draft article based on the differences between the two. Corrections include adding information that is missing from the draft article, correcting incorrect information, and deleting redundant information. When the draft article is corrected, correction information may be provided to the editor by displaying the corrected parts and content. This function may be realized by inputting the generated draft article and the extracted published article into the LLM 233.

[0069] Here, when the generated article proposal and the extracted published article are input into LLM233, instructions are input into LLM233 to request that the input article proposal be compared with the published article, and that the content of the published article that is missing from the article proposal be added to the article proposal, or that the content of the article proposal that expresses the same topic differently from the published article be corrected to match the expression in the published article.

[0070] Furthermore, if the proposed article contains overlapping content, an instruction may be further input to the LLM 233 to request that one of the overlapping content be deleted.

[0071] These instructions are pre-installed as prompts to be input to the LLM 233 in the program executed by the server device 2 when the user requests revisions based on comparison with the published article. In other words, these instructions are not input by the editor, but are pre-stored in the server device 2.

[0072] [Function to download article proposals] The server device 2 downloads the generated article proposal to the terminal device 3 as editable document data.

[0073] [Function to provide related articles] Based on the generated article suggestion, the server device 2 extracts related articles (related content) related to the article suggestion from a database storing published articles, and provides the extracted related articles to the terminal device 3. The related articles may be downloaded to the terminal device 3 together with the article suggestion.

[0074] Specifically, it is preferable that the server device 2 vectorizes the published articles in advance and stores them in a database for extracting related articles. Note that the published articles to be vectorized may be limited to those to be used as related articles.

[0075] The server device 2 vectorizes the generated article ideas and calculates the similarity between the vectorized article ideas and the vectorized published articles stored in the database. This similarity may be calculated using, for example, cosine similarity. Then, the server device 2 may extract published articles whose similarity is equal to or greater than a predetermined threshold from the database as related articles similar to the article ideas.

[0076] [Function to display comparison results with published articles] The server device 2 uses natural language processing to provide an article draft that visually distinguishes between morphemes that make up the generated article draft and morphemes that differ from the extracted morphemes of the published article. For example, among the wording of the article draft displayed in black, the wording that appears in the published article may be displayed in green, and the wording generated by the LLM 233 may be displayed in red. This allows the editor to visually confirm the rephrasing and additional content created by the LLM 233.

[0077] Specifically, the server device 2 uses natural language processing to divide the generated article ideas into morphemes and determines whether the divided morphemes exist in the text of the published article used to generate the article ideas. When displaying the article ideas, the server device 2 may, for example, display morphemes of the article ideas that exist in the text of the published article in green, and display morphemes of the article ideas that do not exist in the text of the published article in red.

[0078] At this time, users can specify specific parts of speech, such as particles, auxiliary verbs, or suffixes, from the morphemes that make up the article proposal to exclude from comparison with the published article text. Morphemes excluded from comparison will remain in black. This makes it possible to visually identify parts of the article proposal that are expressed in red (parts generated independently by LLM233) that differ from the content of the published article used to generate the article proposal.

[0079] [Example of screen displayed on terminal device 3] An example of a screen displayed on the terminal device 3 when article suggestions are provided in the information processing system 1 will be described below.

[0080] 4 is an example of a screen 4a displayed on the terminal device 3. This screen 4a is provided, for example, with an instruction input field 4b for inputting instructions to the LLM 233, and a display field 4c for displaying article ideas and the like generated based on the input instructions.

[0081] The instruction input field 4b is provided with, for example, option buttons 4ba, 4be, 4bg, and 4bh, check boxes 4bb and 4bc, and text boxes 4bd and 4bf. The option buttons 4ba, 4be, 4bg, and 4bh are examples of option buttons for selecting conditions to instruct the LLM 233. The check boxes 4bb and 4bc are examples of check boxes for specifying conditions to instruct the LLM 233. The text boxes 4bd and 4bf are examples of text boxes for inputting conditions to instruct the LLM 233. Each button and box will be described below.

[0082] The option button 4ba is a button that allows the editor to select the content to be created. Screen 4a provides five exemplary content options for the content to be created. The content can include, for example, content that explains world events in an easy-to-understand way so that readers can feel that they are personally involved, or content that picks out a few news stories from the day's news that are considered "the bare minimum that should be read" to grasp the trends of the world and provides a short summary.

[0083] Check box 4bb is a check box for setting whether or not to execute the function of executing corrections based on comparison with published articles described above. When this check box 4bb is checked, the generated article proposal is compared with the published article that was referenced when the article proposal was generated, and the article proposal is corrected based on the differences between them.

[0084] Check box 4bc is a check box for setting whether or not to execute the function of providing related articles described above. When this check box 4bc is checked, in addition to the generated article proposal, related articles related to the article proposal are provided to the editor.

[0085] The text box 4bd is a field for inputting the theme of the information you want to include in the article proposal or related article. In the screen 4a, "Generative AI" is input as an example.

[0086] The option button 4be is a button for selecting the method of extracting published articles to be referenced when generating article ideas. Screen 4a provides two extraction methods as options, for example, "Article Specification" and "Keyword Specification." "Article Specification" is a method of extracting published articles by specifying the URL (Uniform Resource Locator) where the article is posted. "Keyword Specification" is a method of extracting published articles using specified keywords.

[0087] The text box 4bf is a field for inputting the URL where the published article is posted or a keyword for extracting the published article.

[0088] The option button 4bg is a button for selecting the medium of the published article to be extracted. On the screen 4a, for example, two mediums, "electronic version" and "print version", are provided as options.

[0089] The option button 4bh is a button for selecting which part of a published article to search when extracting published articles. For example, the search target can be the preamble of the article or the entire article.

[0090] The items provided in the instruction input field 4b are not limited to the items exemplified in the drawing, and necessary items can be provided as appropriate according to the conditions for extracting published articles, the editing format of the content to be created, etc. For example, items such as the publication period (publication start date and publication end date) of the published articles to be searched, the number of articles to be output, and the number of characters of the article ideas to be output may be provided in the instruction input field 4b.

[0091] 4, a search button 4ca and a headline display field 4cb are illustratively provided. The search button 4ca is a button that is pressed to instruct a search for published articles to be referenced when generating a proposed article. The headline display field 4cb is a field that displays headlines selected by an editor from headlines included in the searched published articles.

[0092] A more detailed explanation follows. When the search button 4ca is pressed, published articles are extracted based on the conditions set in the instruction input field 4b. When the "▼" mark at the right end of the headline display field 4cb is pressed, a list 4cc that displays a list of the headlines of the extracted published articles is displayed below the headline display field 4cb. When the editor selects a desired headline from the list 4cc, the selected headline is added and displayed in the headline display field 4cb.

[0093] In the heading display field 4cb in Fig. 4, two headings are displayed as examples. When the list 4cc is closed in this state, the execute button 4cd, which was hidden under the list 4cc, appears under the heading display field 4c, as shown on the screen 5a in Fig. 5. This execute button 4cd is a button that is pressed when instructing the LLM 233 to generate an article proposal. When the execute button 4cd is pressed, the public article displayed in the heading display field 4c out of the extracted public articles and instructions based on the content set in the instruction input field 4b are input into the LLM 233.

[0094] When the published article and instructions are entered into the LLM 233, a message indicating the progress of processing is displayed below the execute button 4cd. For example, screen 5a displays three messages: "Processing first article," "Processing second article," and "Processing has started. Please wait for output."

[0095] Thereafter, an article draft conforming to the editing format included in the instructions is output from the LLM 233, and the output article draft is displayed below the processing progress message.

[0096] [Behavior when providing article ideas] An example of the operation when providing article suggestions will be described with reference to FIG.

[0097] First, an editor operates the terminal device 3 to input conditions for searching for published articles, for example, in the instruction input field 4b of Figure 5, and presses the search button 4ca.The server device 2 then extracts published articles from the database based on the input conditions (step S101).

[0098] Next, the editor operates the terminal device 3 to input conditions for generating an article proposal, for example, in the instruction input field 4b of Figure 5, and presses the execute button 4cd, and the server device 2 summarizes the extracted published articles as a preprocessing step for generating the article proposal (step S102).

[0099] Next, the server device 2 inputs the summarized article and the instruction received from the terminal device 3 when the execute button 4cd is pressed to the LLM 233, and generates an article draft in the editing format included in the instruction (step S103).

[0100] Next, the server device 2 checks the external format of the generated article idea, such as how it is written, based on a predetermined format (step S104). If this check finds that the predetermined standard is not met, the server device 2 generates the article idea again.

[0101] Next, the server device 2 compares the generated article suggestion with the extracted published article, and corrects the article suggestion based on the difference between the two (step S105).

[0102] Next, the server device 2 extracts related articles related to the proposed article from a database storing published articles based on the generated proposed article (step S106).

[0103] Next, the server device 2 provides the generated article idea and the extracted related articles to the editor (step S107). Providing the idea to the editor includes displaying the idea on the terminal device 3 and downloading the idea to the terminal device 3.

[0104] As described above, according to the information processing system 1 including the server device 2, published articles to be referenced when generating article suggestions can be extracted based on search conditions received from the terminal device 3, instructions can be input to the LLM233 to request the LLM233 to generate article suggestions using the extracted published articles and the published articles, and article suggestions can be generated in the editing format included in the instructions and provided to the terminal device.

[0105] This allows editors to reduce the work of selecting published articles to reference when creating article ideas and the work of editing them to match the post-editing article format, thereby reducing the workload of editing work and the burden on users who perform editing work.

[0106] Therefore, the information processing system 1 including the server device 2 makes it possible to improve the efficiency of editing work by an editor.

[0107] The present invention is not limited to the above-described embodiment, and can be embodied in various other forms without departing from the spirit of the present invention. Therefore, the above-described embodiment is merely an example in all respects and should not be interpreted as being limiting. For example, the order of the above-described processing steps can be arbitrarily changed or executed in parallel as long as no contradiction occurs in the processing content.

[0108] In addition, the programs in the above-described embodiments can be installed or loaded onto a computer through various recording media (computer-readable recording media) such as optical disks such as CDs and DVDs, magnetic disks, and semiconductor memories, or by downloading them from a server device (medium) via a communication network, etc. [Explanation of symbols]

[0109] 1...information processing system, 2...server device, 3...terminal device, 21...processor, 22...communication interface, 23...storage device, 31...processor, 32...communication interface, 33...storage device, 34...input device, 35...display device, 211...control unit, 231...program, 232...data, 233...LLM, 311...control unit, 331...program

Claims

1. means for extracting content to be referenced when generating content proposals from published information based on search conditions received from a terminal device operated by a user; means for inputting a first instruction to the large-scale language model to request the large-scale language model to generate the proposed content using the extracted content and the input content, and to generate the proposed content in accordance with an editing format that defines an editing policy on which the proposed content will be generated; and generating the proposed content in accordance with the editing format; means for providing the generated content proposal to the terminal device; An information processing device comprising:

2. the generating means, when there are a plurality of extracted contents, inputs a second instruction to the large-scale language model to request the large-scale language model to summarize each of the plurality of contents so that variation in length of the plurality of contents falls within a predetermined range, and inputs each of the summarized contents output from the large-scale language model together with the first instruction to the large-scale language model, thereby generating the proposed content in the edited format.

2. The information processing device according to claim 1.

3. the generating means generates the proposed content in the edited format by inputting the extracted content and the first instruction to the large-scale language model in addition to the first instruction to generate the proposed content based on important content identified from the input content in accordance with a predetermined specific condition; 2. The information processing device according to claim 1.

4. The extracting means extracts a predetermined number of the contents from the published information based on the search conditions; the generating means, when inputting the extracted predetermined number of pieces of content and the first instruction into the large-scale language model, summarizes each of the input predetermined number of pieces of content, and inputs the summarized pieces of content into the large-scale language model together with the first instruction to use the summarized pieces of content when generating the proposed pieces of content, thereby generating the proposed pieces of content in the edited format.

2. The information processing device according to claim 1.

5. after generating the content proposal, the generating means evaluates the items constituting the generated content proposal by comparing them with predetermined items provided for evaluating the content proposal, and if the evaluation level does not reach a standard level, generates the content proposal again; 2. The information processing device according to claim 1.

6. after generating the content proposal, the generating means compares the generated content proposal, the extracted content, and the input content proposal with the content, and modifies the content proposal by inputting a third instruction to the large-scale language model to request the large-scale language model to add content of the content that is missing from the content proposal to the content proposal, and to modify content of the content proposal that is different from the content on the same theme to the expression of the content; 2. The information processing device according to claim 1.

7. the providing means downloads the generated content proposal to the terminal device as editable document data; 2. The information processing device according to claim 1.

8. The published information is vectorized, The extracting means vectorizes the generated content proposal, and further extracts related content related to the content proposal from the vectorized published information based on the vectorized content proposal; the providing means further provides the extracted related content to the terminal device.

2. The information processing device according to claim 1.

9. the providing means compares morphemes constituting the content proposal generated using the extracted content with the text of the content used to generate the content proposal, and further provides the content proposal in such a manner that parts of the generated content proposal that are expressed in a manner different from the content of the content used to generate the content proposal can be visually identified; 2. The information processing device according to claim 1.

10. The published information is information published in an electronic version of a newspaper or in a newspaper print version, and the content is an article; 2. The information processing device according to claim 1.

11. 1. A processor-implemented method comprising: extracting content to be referenced when generating content proposals from published information based on search conditions received from a terminal device operated by a user; inputting a first instruction to the large-scale language model to request the large-scale language model to generate the proposed content using the extracted content and the input content, and to generate the proposed content in accordance with an editing format that defines an editing policy on which the proposed content will be generated; and generating the proposed content in accordance with the editing format; providing the generated content proposal to the terminal device; An information processing method including:

12. Computer, means for extracting content to be referenced when generating content proposals from published information based on search conditions received from a terminal device operated by a user; a means for inputting a first instruction to the large-scale language model to request the large-scale language model to generate the proposed content using the extracted content and the input content, and to generate the proposed content in accordance with an editing format that defines an editing policy on which the proposed content will be generated; and means for providing the generated content proposal to the terminal device; An information processing program that functions as a

Images