Information processing system, information processing method, and information processing program

Abstract

An information processing system comprising an AI agent and a plurality of information sources, wherein the AI ​​agent comprises: an information generation means for generating characteristic information representing the purpose or function of the information source based on information collected from the aforementioned information sources, and storing the characteristic information in a memory means after establishing an association between the characteristic information and the information source; a trigger analysis means for analyzing an assigned trigger; a processing execution means for performing specific processing on the trigger based on the analysis result of the trigger; and an information collection means for collecting evaluation information including evaluations, reputations, and comments on the aforementioned information sources; furthermore, when performing the aforementioned specific processing, the processing execution means determines whether at least a portion of the analysis result of the trigger is associated with the aforementioned characteristic information; when the two are associated, the aforementioned information source is selected based on the characteristic information, and the specific processing is performed using information obtained from the selected information source; simultaneously, when selecting the aforementioned information source based on the aforementioned characteristic information, the aforementioned evaluation information is considered when selecting the information source.

Description

... , "Information processing system, information processing method, and information processing program of the present invention 5 Technical Field

[0001] The present invention relates to an information processing system, information processing method, and information processing program using an AI agent, and an AI agent. Background Art 10

[0002] In the past, a content delivery system has been proposed, which, by constructing an agent system, enables the server-side agent to detect the information that the user may want next in the initial communication with the user-side agent, even if the agents do not communicate with each other, and can quickly retrieve the information that may be requested in advance, and can display the required information in succession in an instant (for example, see Patent Document 1). [Prior Art Documents] 15 [Patent Documents] <0003> [Patent Document 1] Japanese Patent Application Publication No. 2003-330843 20 Technical Problem to be Solved by the Invention <0004> However, although the above-mentioned Patent Document 1 discloses the technical idea of ​​detecting the information that the user may want next, it does not specifically describe its implementation method. In the system described in the document, the agent has difficulty collecting information efficiently. 25 [

[0005] In view of this actual situation, the present invention aims to provide an information processing system, information processing method, and information processing program that can efficiently collect information used by AI agents and improve the execution speed of AI agent processing. Technical solution adopted to solve the technical problem -1- HK 20134903 A

[0006] The information processing system of the present invention is an information processing system consisting of an AI agent and a plurality of information sources, wherein the aforementioned AI agent has: an information generation means that generates characteristic information representing the purpose or function of the information source based on information collected from the aforementioned information sources, and establishes an association between the characteristic information and the information source and stores it in a memory means; an activation opportunity analysis means that analyzes the activation opportunity given to the AI ​​agent. The mechanism analyzes the initiation opportunity; it performs specific processing on the initiation opportunity based on the analysis results; and it collects evaluation information, including evaluations, reputation, and comments on the aforementioned information source. When performing the specific processing, the aforementioned processing execution means determines whether at least a portion of the analysis results of the initiation opportunity is related to the aforementioned characteristic information. If the two are related, it selects the aforementioned information source based on the characteristic information and uses the information obtained from the selected information source to perform the specific processing. When selecting the aforementioned information source based on the aforementioned characteristic information, the aforementioned processing execution means considers the aforementioned evaluation information when selecting the information source.

[0007] Furthermore, the information processing method of the present invention is an information processing method executed using an AI agent and a plurality of information sources, wherein the aforementioned AI agent performs: an information generation step, generating characteristic information representing the purpose or function of the information source based on information collected from the aforementioned information sources, and storing the characteristic information in a memory means after establishing an association between the characteristic information and the information source; a trigger analysis step, analyzing the assigned trigger; a processing execution step, performing specific processing on the trigger based on the analysis result of the trigger; and an information collection step, collecting evaluation information including evaluations, reputations, and comments on the aforementioned information sources; when performing the aforementioned specific processing, the processing execution step determines whether at least a portion of the analysis result of the trigger is associated with the aforementioned characteristic information, and when the two are associated, selects the aforementioned information source based on the characteristic information, and uses information obtained from the selected information source to perform the specific processing; when selecting the aforementioned information source based on the aforementioned characteristic information, the processing execution means considers the aforementioned evaluation information to select the information source.

[0008] Furthermore, the information processing program of the present invention is a program of an information processing system comprising an AI agent and a plurality of information sources, wherein the aforementioned AI agent, which is a computer, functions in accordance with the following means: an information generation means that generates characteristic information representing the purpose or function of the information source based on information collected from the aforementioned information sources, and stores the characteristic information in a memory means after establishing an association between the characteristic information and the information source; a trigger analysis means that analyzes the assigned trigger; a processing execution means that performs specific processing on the trigger based on the analysis result of the trigger; and an information collection means that collects evaluation information including evaluations, reputations, and comments on the aforementioned information sources; when performing the aforementioned specific processing, the aforementioned processing execution means determines whether at least a portion of the analysis result of the trigger is associated with the aforementioned characteristic information, and when the two are associated, selects the aforementioned information source based on the characteristic information, and uses the information obtained from the selected information source to perform the specific processing; the aforementioned processing execution means, based on 5 When selecting the aforementioned information source based on the aforementioned characteristic information, the aforementioned evaluation information will be considered when selecting the information source. 10 Effects of the Invention

[0009] According to the information processing system, information processing method, and information processing program of the present invention, information used by the AI ​​agent can be collected efficiently, and the excellent effect of improving the execution speed of AI agent processing can be achieved. Brief Description of the Drawings <0010> [Figure 1] Figure 1 is a schematic diagram showing the general outline of the information processing system 10 of the present embodiment. [Figure 2] Figure 2 is a system configuration diagram showing an example of the configuration of the information processing system 10 of the present embodiment.15 [Figure 3] Figure 3 is a diagram illustrating a specific example of the processing of the information processing system 10 according to this embodiment. (Explanation of symbols) <0011> 10: Information processing system; 12: System terminal; 20 12a: Information generation means; 12b: Initiation trigger analysis means; 12c: Processing execution means; 14: Information source; 16: External terminal; 25 21: CPU; 22: ROM; 23: RAM; 24: Recording medium; -3- HK 20134903 A 25: External memory drive; 26: Memory device; 27: Input device; 28: Display device; 5 29: Communication unit. Detailed Description <0012> Hereinafter, the information processing system 10 according to an embodiment of the present invention will be described using drawings. 10 <0013> <System Overview> First, the overview of the information processing system 10 of this embodiment will be explained using FIG1. ​​FIG1 is a schematic diagram showing the overview of the information processing system 10 of this embodiment. <0014> 15 The information processing system 10 is an information processing system comprising an A1 agent 12 and a plurality of information sources 14 (information sources 14a, 14b, ...). The A1 agent 12 comprises: an information generation means 12a, which generates characteristic information representing the purpose or function of the information source 14 based on information collected from the information source 14, and stores the characteristic information in a memory means (memory device 26) after establishing an association between the characteristic information and the information source 14; a trigger analysis means 12b, which analyzes the assigned trigger; and a processing execution means 12c, which performs specific processing on the trigger based on the analysis result of the trigger. When performing specific processing, the processing execution means 12c determines whether at least a part of the analysis result of the trigger is associated with the characteristic information. When the two are associated, it selects the information source 14 based on the characteristic information and uses the information obtained from the selected information source 14 to perform the specific processing. <0015> 25 The information processing system according to this embodiment can efficiently collect information used by the AI ​​agent and improve the execution speed of the AI ​​agent. <0016> Here, "AI agent" refers to a system that operates as a human agent, has the ability to learn and judge independently, and is a system that integrates various AI technologies to solve complex problems.<0017> The AI ​​agent can perform processing based on externally provided triggers (e.g., user questions (prompt words), signal input from sensors) or internally provided triggers (e.g., activation of internal schedulers, occurrence of an anomaly). <0018> For example, when a question (prompt word) is given as an external trigger (event), tasks such as retrieving or generating an answer to the question are performed, and the answer to the question is output. Similarly, when a problem (prompt word) is given as an external trigger (event), tasks such as retrieving or generating a solution to the problem are performed, and a solution to the problem is output. <0019> Furthermore, "prompt word" refers to an instruction or question given to the system by the user, such as an AI prompt word that includes instructions given to the generative AI, or a command prompt character that provides instructions in the form of a command (instruction). 15 <0010> Various processes performed by AI agents can also be performed by "agent-type AI". <0021> Here, "agent-type AI" refers to a system in which multiple AI agents coordinate their operation. To solve complex problems that a single AI agent cannot solve, agent-type AI has each AI agent assume a specialized responsibility, and while exchanging information with each other, outputs answers to questions or outputs solutions to problems. <0022> "Information source" refers to the source of information or the origin of information. Examples of "information sources" include websites, application software (hereinafter sometimes referred to as "App"), SaaS (Software as a Service), databases, data warehouses, storage services, search engines, etc. <0023> "Characteristic information" refers to information that represents the purpose or function of the information source. As "characteristic information," examples include information provision (news, stock listings, official corporate websites or apps, blogs publishing interests or expertise, etc.), buying and selling goods (e-commerce, hotel or ticket reservations, etc.), communication / dialogue (SNS, bulletin boards, chat rooms, video conferencing, etc.), service provision (storage space, translation, games, etc.), advertising (event announcements, product promotions, etc.), professional information or professional service provision (medical, education, real estate, etc.), development support (cloud IDE, program code management, etc.), operational support (advertising operation support, SNS operation support, etc.), and business support (accounting / finance, human resources / attendance, customer management, etc.).<0024> “Characteristic information” can also be more detailed categorized information. For example, “information provision” can be a large category, and the content of “information provision” (e.g., news, stocks, interests, professional knowledge, etc.) can be categorized into smaller categories. Or, the target audience of “information provision” (e.g., men, women, young people, middle-aged and older people, Japanese, foreigners, etc.) can be categorized into smaller categories. <0025> In addition, there can be an information collection means that collects evaluation information including reviews, reputation, and comments on information source 14. When the processing execution means 12c selects information source 14 based on characteristic information, it can also consider the evaluation information to select the information source 14. <0026> If this configuration is adopted, an appropriate information source can be selected from multiple information sources based on the evaluation information, and the accuracy of AI agent processing can be improved. <0027> 20 Furthermore, if the information source includes a website, the information generation means 12a can also generate characteristic information of the website based on tags obtained from the website. <0028> If configured in this way, the accuracy of the website's characteristic information as an information source can be improved. By selecting information sources with high relevance to the analysis results of the activation trigger, the accuracy of AI agent processing can be improved. 25 <0029> Here, "tag" includes HTML tags, conversion tags, and tool control tags. HTML tags refer to tags composed of HTML (Hyper Text Markup Language), the programming language that constitutes the website. Conversion tags are tags used to calculate the number of actions such as purchases or inquiries. By utilizing -6- HK 20134903 A in online advertising such as list ads, the accurate number of conversions can be grasped. Tool control tags are tags embedded in HTML to link the use of tools. When someone accesses a website with embedded tool control tags, user information and inflow paths can be collected, and the page that pops up can be identified. <0020> 5 Furthermore, the information generation means 12a can also generate characteristic information of the website based on the information collected by the website. <0031> If configured in this way, the accuracy of the characteristic information of the website serving as the information source can be improved. By selecting an information source with a high correlation to the resolution results of the activation trigger, the accuracy of AI agent processing can be improved. 10 <0032> Here, "information collected by the website" can include, for example, personal information, browser or OS type and version, location information, visit history, IP address, network type, click history of browsing pages or advertisements, purchase history, or points usage status, etc. <0033> 15 Furthermore, the information source includes application software, and the information generation means 12a can also generate characteristic information of the application software based on the information collected by the application software.<0034> If configured in this way, the accuracy of the characteristic information of the application software as the information source can be improved. By selecting an information source with high correlation to the analysis results of the startup trigger, the accuracy of the AI ​​agent's processing can be improved. <0035> Here, "information collected by the application software" can include, for example, personal information, type and version of the OS, location information, mobile history, IP address, network type, operation history within the App, click history of browsing pages or advertisements, purchase history or points utilization status, etc. 25 <0036> <System Configuration Example> Next, the configuration example of the information processing system 10 of this embodiment will be described using FIG2. FIG2-7- HK 20134903 A is a system configuration diagram showing the configuration example of the information processing system 10 of this embodiment. <0037> The information processing system 10 may be configured, for example, as a system terminal 12 controlling the entire system, and a plurality of information sources 14 and external terminals 16 that can communicate with each other via a network NW. <0038> The system terminal 12 is a terminal that constitutes the agent of A1 and is the terminal that controls the entire information processing system 10, and is composed of conventionally known servers or personal computers. In this example, the system terminal 12 is composed of one server, but it may also be composed of a plurality of servers or personal computers. The hardware configuration of the system terminal 12 and the programs executed by the system terminal 12 will be described later. <0039> The plurality of information sources 14 are configured with information source A14a, information source B14b, etc. <0030> Information source A14a is a web server that provides a website. The content of the website is not particularly limited. For example, websites that provide information (news, stock information, corporate websites, blogs for publishing interests or professional knowledge, etc.), websites for buying and selling goods (e-commerce, hotel or ticket reservations, etc.), websites for communication / dialogue (SNS, bulletin boards, chat rooms, video conferencing, etc.), websites for providing services (storage space, translation, games, etc.), websites for advertising (event announcements, product promotions, etc.), websites for providing professional information or services (medical, education, real estate, etc.), websites for development support (cloud IDE, program code management, etc.), websites for business support (advertising operation support, SNS operation support, etc.), and websites for business support (accounting / finance, human resources / attendance, customer management, etc.). <0041> Information source B14b is the server that provides the App, for example, it can be composed of SaaS (Software as a Service).The content of an app is not particularly limited. For example, apps that provide information (news, stock, official corporate apps, blogs for publishing interests or professional knowledge, etc.), apps for buying and selling goods (e-commerce, hotel or ticket reservations, etc.), apps for communication / dialogue (SNS, bulletin boards, chat rooms, video conferencing, etc.), apps for providing services (storage, translation, games, etc.), apps for advertising (event announcements, product promotions, etc.), apps for providing professional information or services (medical, educational, real estate, etc.), apps for development support (cloud IDE, program code management, etc.), business support (advertising operation support, SNS operation support, etc.), and apps for business support (accounting / finance, human resources / attendance, customer management, etc.). <0042> Furthermore, one, several, or all of the plurality of information sources 14 may be processing devices interconnected with the system terminal 12 via a local area network (e.g., LAN or P2P). Therefore, for example, a website may be an internal network, and an app may be an app stored in the memory of the system terminal 12 (e.g., memory device 26), or an app installed on a mobile terminal (e.g., a tablet or smartphone) connected to the network NW. <0043> External terminal 16 is a terminal used by users of the information processing system 10 (individuals, or specific groups of multiple people (teams, departments, enterprises, etc.)), and may consist of personal computers, tablets, smartphones, etc. The type of external terminal 16 is not particularly limited, but may correspond to, for example, a smartphone, personal computer, tablet used by an individual, or a smartphone, personal computer, tablet used by enterprise employees, etc. <0044> The network NW is the communication line between system terminal 12, information source 14, and external terminal 16. Typically, it consists of a WAN (Wide Area Network), also known as the Internet. Furthermore, the network NW can be wired or wireless, and can be a LAN (Local Area Network), a VPN (Virtual Private Network), or a combination of these dedicated lines. <0045> 25 <System Terminal / Hardware Configuration Example> Next, a hardware configuration example of system terminal 12 will be described.<0046> As shown in Figure 2, the system terminal 12 is configured, for example, with a CPU 21, a ROM 22, RAM 23 connected to the CPU 21 via a bus, an external memory driver 25, a memory device 26, an input device 27, a display device 28, and a communication unit 29. <0047> The CPU 21 is the control means for the overall system terminal 12, executing the processing of application software and operating system (OS) stored in the ROM 22 or memory device 26, and storing the data or files required for the execution program in the RAM 23 or memory device 26. <0048> The ROM 22 is a memory means for storing basic I / O programs and various data, and is composed of, for example, PROM or flash memory. The RAM 23 is a memory means for temporarily storing data, and is composed of, for example, SDRAM or DRAM. External memory drive 25 is a control means for reading and writing data to recording media 24 such as magnetic tape and DVD, and is composed of, for example, magnetic tape storage and DVD drive. <0049> Memory device 26 is a memory means for storing application software, OS, control programs, associated programs, various information, etc., and is composed of, for example, hard disk (HDD) and solid-state drive (SSD). Input device 27 15 is used for inputting commands (instructions) to system terminal 12, and is composed of, for example, keyboard, pointing device (mouse, etc.), touch panel, etc. <0040> Display device 28 is used for displaying commands input by input device 27, response output of system terminal 12 to the command, and various displays, and is composed of, for example, liquid crystal display, plasma display, OLED, etc. Communication unit 29 is a control means for controlling communication with information source 14 and external terminal 16 via network NW, and is composed of, for example, communication card. <0051> <System Terminal / Function> 25 Next, the function of system terminal 12 will be described. <0052> The memory device 26 of system terminal 12 stores a program (Information-10-HK 20134903 A processing program) that enables system terminal 12 to function as information generation means 12a, activation trigger analysis means 12b, and processing execution means 12c. <0053> <System Terminal / Function / Information Generation Means> 5 Next, information generation means 12a will be described.<0054> The information generation means 12a is a means of generating characteristic information representing the purpose or function of information source 14 based on information collected from information source 14, and storing the characteristic information in a memory means (memory device 26) after establishing an association between the characteristic information and information source 14. In this example, it consists of a program stored in the memory device 26 of the system terminal 12 10, the memory device 26, etc. <0055> The information generation means 12a cycles (crawls) a plurality of information sources 14 (information source A14a, information source B14b, ...) at a predetermined period, and generates characteristic information representing the purpose or function of information source 14 based on the information obtained from information source 14, and stores the characteristic information in the memory device 26 after establishing an association between the characteristic information and information source 14. <0056> As mentioned above, "information source" refers to the source of information or the source of information transmission. As an "information source," examples include websites, application software (hereinafter sometimes referred to as "App"), SaaS (Software as a Service), databases, data warehouses, storage services, search engines, etc. "Characteristic information" refers to information representing the purpose or function of the information source. Examples of "characteristic information" include information provision (news, stock information, official corporate websites or apps, blogs publishing interests or professional knowledge, etc.), buying and selling goods (e-commerce, hotel or ticket reservations, etc.), communication / dialogue (SNS, bulletin boards, chat rooms, video conferencing, etc.), service provision (storage space, translation, games, etc.), advertising (event announcements, product promotions, etc.), professional information or professional service provision (medical, educational, real estate, etc.), development support (cloud IDE, program code management, etc.), operational support (advertising operation support, SNS operation support, etc.), business support (accounting / finance, human resources / attendance, customer management, etc.). -11- HK 20134903 A <0058> “Characteristic Information” can also be more detailed categories of information. For example, “Information Provided” can be classified as a large category, and the content of “Information Provided” (such as news, stocks, interests, expertise, etc.) can be classified as a small category, or the target audience of “Information Provided” (such as men, women, young people, middle and old age groups, Japanese, foreigners, etc.) can be classified as a small category.<0059> For example, the information generation means 12a crawls (crawls) information source A14a, which is composed of web servers, at a predetermined period. If the website provided by information source A14a is an EC website, as shown in Figure 3(a), it generates "commodity trading (EC website)" as one of the characteristic information of information source A14a, and associates this characteristic information (commodity trading (EC website)) with the identifier of information source A14a (e.g., the ID of information source A14a) and stores it in the memory device 26. <0050> If the website provided by information source A14a is a news publishing website, as shown in Figure 3(b), the information generation means 12a generates "information provision (news)" as one of the characteristic information of information source A14a, and associates this characteristic information (information provision (news)) with the identifier of information source A14a (e.g., the ID of information source A14a) and stores it in the memory device 26. <0061> If the website provided by information source A14a is a promotional website for a certain product (e.g., cosmetics), as shown in Figure 3(c), the information generation means 12a generates "promotional advertisement (cosmetics)" as one of the characteristic information of information source A14a, and associates this characteristic information (promotional advertisement) with the identifier of information source A14a (e.g., the ID of information source A14a) and stores it in the memory device 26. <0062> Furthermore, the information generation means 12a generates the characteristic information of the website based on tags obtained from the website or information collected from the website. 25 <0063> Here, "tags" include HTML tags, transformation tags, and tool control tags. Furthermore, as "information collected by the website," examples include personal information, browser or OS type and version, location information, visit history, IP address, network type, click history of browsing pages or advertisements, purchase history, or points usage status, etc. <0064> For example, when the information generation means 12a obtains personal information (age, gender) 5 from the labels obtained from the website provided by the information source A14a, which is composed of a web server, or from the information collected by the website, as shown in Figures 3(a) and (c), it associates this characteristic information (personal information (age, gender)) with the identifier of the information source A14a (e.g., the ID of the information source A14a) and stores it in the memory device 26.<0065> Furthermore, the information generation means 12a crawls the information source 10 B14b composed of SaaS at a predetermined period. If the App provided by the information source B14b is an accounting / financial service, as shown in Figure 3 (d), "business support (accounting and finance)" is generated as one of the characteristic information of the information source B14b. The characteristic information (business support (accounting and finance)) is associated with the identifier of the information source B14b (e.g., the ID of the information source B14b) and then stored in the memory device 26. <0066> 15 Furthermore, if the App provided by information source B14b is a bulletin board, as shown in FIG3(e), the information generation means 12a generates "communication dialogue (bulletin board)" as one of the characteristic information of information source B14b, and associates this characteristic information (communication dialogue (bulletin board)) with the identifier of information source B14b (e.g., the ID of information source B14b) and stores it in the memory device 26. <0067> 20 Furthermore, if the App provided by information source B14b is a blog for publishing home appliance information, as shown in FIG3(f), the information generation means 12a generates "information provision (home appliance)" as one of the characteristic information of information source B14b, and associates this characteristic information (information provision (home appliance)) with the identifier of information source B14b (e.g., the ID of information source B14b) and stores it in the memory device 26. <0068> 25 Furthermore, the information generation means 12a generates characteristic information of the App based on the information collected by the App. <0069> Here, "information collected by the application software" can include, for example, personal information, OS type and version, location information, mobile history, IP address, network type, operation history within the App, click history of browsing pages or advertisements, purchase history or points usage status, etc. <0060> For example, when the information generation means 12a obtains browsing history from the information collected by the App 5 provided by the information source 814b, which is composed of SaaS, as shown in Figures 3(d) to (f), it associates the characteristic information (browsing history) with the identifier of the information source 814b (e.g., the ID of the information source 814b) and stores it in the memory device 26. <0071> 10 <System Terminal / Function / Startup Trigger Resolution Method> Next, the startup trigger resolution method 12b will be explained.<0072> The activation trigger analysis means 12b is a means of analyzing the given activation trigger. In this example, it consists of a program stored in the memory device 26 of the system terminal 12, the memory device 26, etc. 15 <0073> As described above, an "activation trigger" can be, for example, an activation trigger given externally to the AI ​​agent 12 (e.g., a user's question (prompt word), a signal input from a sensor), or an activation trigger given internally to the AI ​​agent 12 (e.g., the activation of an internal scheduler, the occurrence of an anomaly). <0074> 20 When a prompt word is given by a user of the information processing system 10 (e.g., an individual, or a specific group (team, department, enterprise, etc.) consisting of multiple people) using the external terminal 16 as an activation trigger, the activation trigger analysis means 12b obtains the prompt word input by the user, associates the obtained prompt word with the user's ID, and stores it in the memory device 26. <0075> 25 Here, "prompt word" refers to the information (data) of instructions or questions input by the user of the information processing system 10 using an input means (keyboard, pointing device (mouse, etc.), touch panel, microphone, etc.) of the external terminal 16. The data format of "prompt word" is not particularly limited, and can be any of the following: text data, voice data, still image data, moving image data, files, etc. -14- HK 20134903 A <0076> As "prompt words", examples include text data or voice data of instructions or questions to the information processing system 10, files uploaded by the user, still image data, or moving image data, etc. <0077> 5 Next, the trigger analysis means 12b is activated to analyze the prompt words stored in the memory device 26 and generate the analysis result of the prompt words. <0078> Here, the analysis of prompt words is performed based on, for example, the content of the prompt words and the user's expectations inferred from the content of the prompt words. 10 <0079> For example, when the content of the prompt word is used as the object of analysis, the trigger analysis method 12b is activated. The text data, voice data, image data, etc. contained in the prompt word are used to identify the content of the prompt word (e.g., article, section, single word, keyword, feature point, etc.), and the identified content is used as the analysis result of the prompt word. After being associated with the user's ID, it is stored in the memory device 26.15 <0070> Furthermore, for example, when the user's expectation inferred from the content of the prompt is used as the object of analysis, the trigger analysis means 12b is activated. Using the text data, voice data, video data, etc. contained in the prompt, the user's expectation inferred from the content of the prompt (e.g., requesting the specified analysis, requesting translation, requesting the generation of a program, etc.) is identified, and the identified user expectation is used as the result of the prompt analysis. After being associated with the user's ID, it is stored in the memory device 26. <0081> <System Terminal / Function / Processing Execution Means> Next, the processing execution means 12c will be described. 25 <0082> The processing execution means 12c performs specific processing on the activation trigger based on the parsing result of the activation trigger. It also determines whether at least a portion of the activation trigger parsing result is associated with characteristic information. When the two are associated, it selects information source 14 based on the characteristic information and uses information obtained from the selected information source -15- HK 20134903 A 14 to perform the specific processing. In this example, it consists of a program stored in the memory device 26 of the system terminal 12, the memory device 26, etc. <0083> When a question (prompt word) is given to the user of the information processing system 10 as an activation trigger, the processing execution means 12c retrieves or generates an answer to the question and outputs the answer to the question. <0084> Furthermore, when retrieving an answer to a question, the processing execution means 12c determines whether at least a portion of the initiation trigger parsing result is associated with characteristic information. If the two are associated, information source 14 is selected based on the characteristic information, and the answer to the question is output using the information obtained from the selected information source 14. 10 <0085> For example, consider the question "What are some recommended cosmetics suitable for women in their 20s?" given as a prompt word. The parsing result of the initiation trigger (prompt word) stores the information "target age is women in their 20s, target product is cosmetics, and the most popular cosmetic among all cosmetics" in the memory device 26, and the characteristic information of information source A14a stores the information "product purchase and sale, personal information (age, gender)" in the memory device 26, and the personal information (age, gender) of the characteristic information includes information about women in their 20s. <0086> In this case, the processing execution means 12c determines that "target age is a woman in her 20s" as part of the parsing result of the initiation trigger (prompt word) is consistent (related) with "personal information (woman in her 20s)" as the characteristic information of information source A14a, and selects information source A14a (EC website in this example) based on the characteristic information.<0087> Then, the answer to the question is retrieved using the selected information source A14a (EC website), or an answer is generated using information obtained from information source A14a (e.g., information about cosmetics), and the answer to question 25 is output (e.g., textual data such as "the best-selling cosmetic is cosmetic X"). <0088> For example, consider the case where the question "What are the recommended cosmetics for women in their 20s?" is given as a prompt word, and the parsing result of the trigger (prompt word) stores the information "target age is women in their 20s, target product is cosmetics, most popular cosmetic among all cosmetics" in memory device 26, and the information "advertisement (cosmetics), personal information (age, gender)" is stored in memory device 26 as characteristic information of information source A14a. <0089> 5 In this case, processing execution means 12c determines that "the target product is cosmetics" as part of the parsing result of the initiation trigger (prompt word) is consistent with "advertisement (cosmetics)" as characteristic information of information source A14a, and selects information source A14a (in this example, a cosmetics promotion website) based on the characteristic information. <0080> Then, it retrieves an answer to the question using the selected information source A14a (cosmetics promotion website), or generates an answer using information obtained from information source A14a (e.g., information about the cosmetics being promoted), and outputs the answer to the question (e.g., text data such as "What do you think of cosmetic Y?"). <0091> When a user of the information processing system 10 provides a topic (prompt word), processing execution means 12c retrieves or generates a solution to the topic, and outputs a solution to the topic. 15 <0092> Furthermore, when processing execution means 12c retrieves solutions to a problem, it determines whether at least a portion of the initiation trigger analysis result is related to characteristic information. If the two are related, it selects information source 14 based on the characteristic information and uses the information output obtained from the selected information source 14 to provide solutions to the problem. 20 <0093> For example, consider the scenario where a question is given as a prompt word: "Please create a ranking of home appliances that may become popular in Japan in the future, up to number 100." The analysis result of the initiation trigger (prompt word) stores the information "target region is Japan, survey object is home appliances, ranking from number 1 to number 100" in the memory device 26, and the characteristic information of information source 814b stores the information "information provided (home appliances), browsing history" in the memory device 26.<0094> In this case, processing execution means 12c determines that "the survey target is home appliances" as part of the analysis result of the initiation trigger (prompt word) is consistent with "information provision (home appliances)" as characteristic information of information source 814b -17- HK 20134903 A, and selects information source B14b (in this example, a blog that publishes home appliance information) based on the characteristic information. <0095> Then, it retrieves answers to the question using the selected information source B14b (blog that publishes home appliance information), or generates an answer using information obtained from information source B14b (e.g., articles that evaluate or introduce home appliances, or the profiles of readers who viewed the article), and outputs the answer to the question (e.g., image data of home appliance ranking icons). <0096> Furthermore, considering the prompt word "Please create a ranking of home appliances that may be popular in Japan in the future, up to the 100th place," the analysis result of the activation trigger (prompt word) stores the information "target region 10 is Japan, the survey object is home appliances, and the ranking is from 1st to 100th place" in the memory device 26, and the characteristic information of information source B14b stores the information "communication dialogue (bulletin board) and browsing history" in the memory device 26, and the communication dialogue (bulletin board) of the characteristic information contains discussion threads about home appliances. <0097> 15 In this case, the processing execution means 12c determines that "the survey object is home appliances" as part of the analysis result of the activation trigger (prompt word) is consistent with "discussion threads about home appliances in the communication dialogue (bulletin board)" as the characteristic information of information source B14b, and selects information source B14b (bulletin board in this example) based on the characteristic information. <0098> 20 Then, the selected information source B14b (bulletin board) is used to retrieve answers to the questions, or information obtained from information source B14b (such as reviews and comments about home appliances) is used to generate answers, and the answers to the questions are output (such as the URL information of a website that publishes a ranking of home appliances up to number 100). <0099> According to this example, the information used by the AI ​​agent can be collected efficiently, and the execution speed of AI agent processing can be improved. <0090> In addition, there can be an information collection means to collect evaluation information including evaluations, reviews and comments about information source 14. When the processing execution means 12c selects information source 14 based on characteristic information, it can also consider the evaluation information to select the information source 14. <0101> If this configuration is adopted, an appropriate information source can be selected from multiple information sources based on the evaluation information, and the accuracy of AI agent processing can be improved.5 <0102> <Information Processing System / Summary> As explained above, the information processing system of this embodiment (e.g., the information processing system 10 shown in Figures 1 and 2) is an information processing system consisting of an AI agent (e.g., the AI ​​agent 12 shown in Figure 1, the system terminal 12 of 10 shown in Figure 2) and a plurality of information sources (e.g., the plurality of information sources 14 shown in Figure 1). The aforementioned AI agent includes: an information generation means (e.g., information generation means 12a shown in Figures 1-3), which generates characteristic information representing the purpose or function of the information source based on information collected from the aforementioned information sources, and associates the characteristic information with the information source and stores it in a memory means (e.g., memory device 26 shown in Figures 1-3); a trigger analysis means (e.g., trigger analysis means 12b shown in Figure 1 or 2), which analyzes the assigned trigger; and a processing execution means (e.g., processing execution means 12c shown in Figures 1-3). Based on the analysis results of the aforementioned initiation opportunity, specific processing of the initiation opportunity is performed; when performing the aforementioned specific processing, the aforementioned processing execution means determines whether at least a part of the analysis results of the aforementioned initiation opportunity is associated with the aforementioned characteristic information. When the two are associated, the aforementioned information source is selected based on the characteristic information, and the information obtained from the selected information source is used to perform the specific processing.<0103> Furthermore, regarding the information processing method of this embodiment (e.g., the method executed by the information processing system 10 shown in FIG1 and FIG2), which is an information processing method executed using an AI agent (e.g., AI agent 12 shown in FIG1, system terminal 12 shown in FIG2) and a plurality of information sources (e.g., a plurality of information sources 14 shown in FIG1) 25, it includes: an information generation step (e.g., the processing executed by the information generation means 12a shown in FIG1 to FIG3), which generates characteristic information representing the purpose or function of the information source based on information collected from the aforementioned information sources, and establishes an association between the characteristic information and the information source and stores it in a memory means (e.g., the memory device 26 shown in FIG1 to FIG3); a trigger analysis step (e.g., the processing executed by the trigger analysis means 12b shown in FIG1 or FIG2), which analyzes the assigned trigger; and a processing execution step (e.g., FIG1, ..., FIG3). The processing execution means 12c shown (the processing executed) performs a specific processing on the aforementioned initiation opportunity based on the parsing result of the aforementioned initiation opportunity; when performing the aforementioned specific processing, the aforementioned processing execution step determines whether at least a part of the parsing result of the aforementioned initiation opportunity is associated with the aforementioned characteristic information. When the two are associated, the aforementioned information source is selected based on the characteristic information, and the information obtained from the selected information source is used to perform the specific processing.<0104> Furthermore, regarding the program of the information processing system of this embodiment (e.g., the information processing system 10 shown in Figures 1 and 2), it is a program of an information processing system consisting of an AI agent (e.g., the AI ​​agent 12 shown in Figure 1, the system terminal 12 shown in Figure 2) and a plurality of information sources (e.g., the plurality of information sources 14 shown in Figure 1), wherein the aforementioned AI agent, which is a computer, functions as follows: an information generation means (e.g., the information generation means 12a shown in Figures 1 and 3), which generates characteristic information representing the purpose or function of the information source based on information collected from the aforementioned information source, and stores the characteristic information in a memory means (e.g., the memory device 26 shown in Figures 1, 2, and 3); a start-up opportunity analysis means (e.g., the start-up opportunity analysis means 12b shown in Figures 1 or 2), which analyzes the assigned start-up opportunity; and a processing execution means (e.g., Figures 1, 2, and 3). The processing execution means 12c shown performs specific processing on the aforementioned initiation trigger based on the parsing result of the aforementioned initiation trigger; when performing the aforementioned specific processing, the aforementioned processing execution means determines whether at least a portion of the parsing result of the aforementioned initiation trigger is associated with the aforementioned characteristic information. When the two are associated, the aforementioned information source is selected based on the characteristic information, and the information obtained from the selected information source is used to perform the specific processing.<0105> Furthermore, the AI ​​agent in this embodiment (e.g., the AI ​​agent 12 shown in FIG1, the system terminal 12 shown in FIG2) operates as a human agent, possessing the ability to learn and make judgments independently. The aforementioned AI agent, acting as a computer, functions through the following means: an information generation means (e.g., information generation means 12a shown in FIG1, ..., FIG3), which generates characteristic information representing the purpose or function of an information source based on information collected from information sources (e.g., a plurality of information sources 14 shown in FIG1), and associates this characteristic information with the information source and stores it in a memory means (e.g., memory devices 26 shown in FIG1-FIG3); a trigger analysis means (e.g., trigger analysis means 12b shown in FIG1 or FIG2), which analyzes the assigned trigger; and a processing execution means (e.g., in FIG1-FIG3). The processing execution means 12c shown performs specific processing on the aforementioned initiation trigger based on the analysis result of the aforementioned initiation trigger. When performing the aforementioned specific processing, the processing execution means determines whether at least a portion of the analysis result of the aforementioned initiation trigger is associated with the aforementioned characteristic information. When the two are associated, the aforementioned information source is selected based on the characteristic information, and the specific processing is performed using information obtained from the selected information source. Furthermore, the aforementioned initiation trigger analysis result and the aforementioned characteristic information are repeatedly assigned as learning data, and the aforementioned information source suitable for the aforementioned specific processing is learned through machine learning. <0106> According to the information processing system, information processing method, information processing program, and AI agent of this embodiment, the information used by the AI ​​agent can be collected efficiently, and the execution speed of the AI ​​agent processing can be improved. 15 <0107> Furthermore, an information collection means (or information collection step) may be provided to collect evaluation information including assessments, reputations, and comments regarding the aforementioned information sources. When selecting the aforementioned information source based on the aforementioned characteristic information, the aforementioned processing execution means (or the aforementioned information collection step) may also consider the aforementioned evaluation information when selecting the information source. 20 <0108> If configured in this way, an appropriate information source can be selected from a plurality of information sources based on the evaluation information, and the accuracy of AI agent processing can be improved. <0109> Furthermore, if the aforementioned information source includes a website, the aforementioned information generation means (or the aforementioned information generation step) 25 may also generate the aforementioned characteristic information of the website based on tags (e.g., HTML tags, conversion tags, tool control tags) obtained from the aforementioned website.<0100> If configured in this way, the accuracy of the characteristic information of the website serving as the information source can be improved. By selecting information sources with high relevance to the startup trigger resolution results, the accuracy of AI agent processing can be improved. <0111> Furthermore, the aforementioned information generation method can also generate the characteristic information of the website based on information collected by the website (such as personal information, browser or OS type and version, location information, visit history, IP address, network type, page or advertisement click history, purchase history, or points usage status, etc.). <0112> If configured in this way, the accuracy of the characteristic information of the website serving as the information source can be improved. By selecting information sources with high relevance to the startup trigger resolution results, the accuracy of AI agent processing can be improved. 10 <0113> Furthermore, the aforementioned information source includes application software, and the aforementioned information generation means (or the aforementioned information generation steps) can also generate the aforementioned characteristic information of the application software based on the information collected by the aforementioned application software (such as personal information, type and version of OS, location information, mobile history, IP address, network type, operation history within the App, click history of browsing pages or advertisements, purchase history or points utilization status, etc.). <0114> If configured in this way, the accuracy of the characteristic information of the application software as the information source can be improved, and by selecting information sources with high correlation to the analysis results of the startup trigger, the accuracy of AI agent processing can be improved. 20 <0115> In addition, the information processing system, information processing method, information processing program, and AI agent of the present invention are of course not limited to the above-described embodiments, and various modifications can be made without departing from the spirit of the present invention. <0116> 25 Therefore, for example, in the above embodiments, the "characteristic information" of the present invention is only information representing the purpose or function of the information source, and is not limited to the illustrated information. Furthermore, the "startup trigger" of the present invention is not limited to a prompt word, but can also be an externally given startup trigger (e.g., a signal input from a sensor), or an internally given startup trigger (e.g., the start of an internal scheduler, the occurrence of an abnormality -22- HK 20134903 A). <0117> Furthermore, the "information source" of the present invention is not limited to a website or App, but can also be a database, data warehouse, storage service, search engine, etc. 5 [Industrial Applicability] 10 <0118> The information processing system, information processing method, information processing program, and AI agent of the present invention are widely applicable to manufacturing, service, and retail industries.-23- HK 20134903 A Claim 1. An information processing system comprising an AI agent and a plurality of information sources, wherein the AI ​​agent comprises: 5. an information generation means for generating characteristic information representing the purpose or function of the information source based on information collected from the aforementioned information sources, and for storing the characteristic information in a memory means after establishing an association between the characteristic information and the information source; 6. a trigger analysis means for analyzing an assigned trigger; 7. a processing execution means for performing 10 specific processing on the trigger based on the analysis result of the trigger; and 8. an information collection means for collecting evaluation information including evaluations, reputations, and comments on the aforementioned information sources. 8. When performing the aforementioned specific processing, the aforementioned processing execution means determines whether at least a portion of the analysis result of the aforementioned initiation opportunity is associated with the aforementioned characteristic information. When the two are associated, the aforementioned information source is selected based on the characteristic information, and the specific processing is performed using information obtained from the selected information source. When selecting the aforementioned information source based on the aforementioned characteristic information, the aforementioned processing execution means considers the aforementioned evaluation information to select the information source. 2. The information processing system as described in claim 1, wherein, 20, the aforementioned information source includes a website; the aforementioned information generation means generates the aforementioned characteristic information of the website based on tags obtained from the aforementioned website. 3. The information processing system as described in claim 1 or 2, wherein, the aforementioned information source includes application software; 25, the aforementioned information generation means generates the aforementioned characteristic information of the application software based on information collected by the aforementioned application software. 4. An information processing method using an AI agent and a plurality of information sources, wherein the AI ​​agent performs: -1- HK 20134903 A Information generation step, generating characteristic information representing the purpose or function of the information source based on information collected from the aforementioned information sources, and establishing an association between the characteristic information and the information source and storing it in a memory means; 5 Initiation opportunity analysis step, analyzing the assigned initiation opportunity; 6 Processing execution step, performing specific processing on the initiation opportunity based on the analysis result of the aforementioned initiation opportunity; and 7. Information collection step, collecting evaluation information including evaluations, reputations and comments on the aforementioned information sources. When performing the aforementioned specific processing, the processing execution step determines whether at least a portion of the analysis result of the aforementioned initiation opportunity is associated with the aforementioned characteristic information. When the two are associated, the aforementioned information source is selected based on the characteristic information, and the specific processing is performed using information obtained from the selected information source; When selecting the aforementioned information source based on the aforementioned characteristic information, the processing execution step considers the aforementioned evaluation information when selecting the information source.15 5. A program for an information processing system comprising an AI agent and a plurality of information sources, wherein the AI ​​agent, acting as a computer, functions as follows: an information generation means for generating characteristic information representing the purpose or function of the information source based on information collected from the aforementioned information sources, and storing the characteristic information in memory after associating it with the information source; a trigger analysis means for analyzing an assigned trigger; a processing execution means for executing a specific processing of the trigger based on the analysis result of the trigger; and an information collection means for collecting evaluation information including evaluations, reputations, and comments of the aforementioned information sources; wherein, when executing the specific processing, the processing execution means determines whether at least a portion of the analysis result of the trigger is associated with the aforementioned characteristic information, and when the two are associated, selects the aforementioned information source based on the characteristic information, and uses information obtained from the selected information source to execute the specific processing; When selecting the aforementioned information source based on the aforementioned characteristic information, the aforementioned processing execution method will consider the aforementioned evaluation information to select the information source. 5 -3- HK 20134903 A HK 20134903 A HK 20134903 A HK 20134903 A.

Images