Data structure and information processing system using the data structure

HK20137565APending Publication Date: 2026-09-11D4ALL CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
HK22026127119
Authority / Receiving Office
HK · HK
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-08-07
Filing Date
2026-08-05
Publication Date
2026-09-11

Smart Images

  • Figure 00000021_0000
    Figure 00000021_0000
  • Figure 00000022_0000
    Figure 00000022_0000
  • Figure 00000023_0000
    Figure 00000023_0000
Patent Text Reader

Abstract

A data structure for data used by an AI agent includes: service information that identifies the services used by the AI ​​agent; and data information that identifies the data used by the AI ​​agent. The preceding data information includes: a service information ID representing the preceding service information. The AI ​​agent identifies the preceding data information based on the parsing result of a given prompt, and identifies the preceding service information based on the preceding service information ID contained in the identified data information. Based on the identified preceding service information, the AI ​​agent accesses the preceding service and retrieves the preceding data identified based on the preceding data information from the service. This structure is used in the processing of the AI ​​agent when initiating AI processing for the preceding prompt based on the retrieved preceding data.
Need to check novelty before this filing date? Find Prior Art

Description

5 10 15 20 25 Specification Data Structure, and Information Processing System Using the Data Structure Technical Field

[0001] This invention relates to a data structure for data used by an agent employing AI, and an information processing system using the data structure. Background Art

[0002] Previously, systems utilizing external services in AI-based systems were known. For example, Patent Document 1 proposes a system comprising: means for allowing a user to input presentation data and voice data; means for providing appropriate feedback by generative AI; means for inputting voice data into text and vectorizing volume or emotion data, and returning feedback on speaking style; the system comprises: means for vectorizing emotion data of voice data using an external service’s “emotion analysis service”. [Prior Art Documents] [Patent Documents]

[0003] [Patent Document 1] The technical problem to be solved by the invention disclosed in Japanese Patent Application Publication No. 2025-50770

[0004] However, in previous systems that used AI, there is still room for improvement in the processing when utilizing external services. -1- HK 20137565 A 5 10 15 20 25

[0005] In view of this fact, the present invention attempts to provide a data structure and an information processing system using the data structure, which can improve the quality or speed of processing the triggering opportunity such as prompts for users by selecting the best service from the various services utilized by AI.The technical solution adopted to solve the technical problem

[0006] The data structure described in this invention is a data structure for data used by an AI agent, characterized in that it contains: service information that can identify the services used by the AI ​​agent; and data information that can identify the data used by the AI ​​agent; the preceding data information contains: a service information ID representing the preceding service information; the preceding AI agent specifies the preceding data information based on the parsing result of the given prompt, and specifies the preceding service information based on the preceding service information ID contained in the specified data information; accesses the preceding service based on the specified preceding service information, and obtains the preceding data specified based on the preceding data information from the service; it is used in the processing of the preceding AI agent when starting AI processing for the preceding prompt based on the obtained preceding data. Effects of the Invention

[0007] According to the data structure described in this invention and the information processing system using the data structure, by selecting the best service from the various services utilized by AI, the quality or speed of processing triggers such as prompts for users can be improved, thus achieving such excellent results. Brief Description of the Drawings

[0008] [Figure 1] A schematic diagram of the data structure of the data described in this embodiment. [Figure 2] A system configuration diagram of an example of the configuration of the information processing system 10 to which the data structure of the data described in this embodiment is applicable. [Figure 3] An illustration of an example of the use of the data described in this embodiment. -2- HK 20137565 A 5 10 15 20 25 (Symbol Explanation)

[0009] 10: Information processing system; 12: System terminal; 14: Service; 12a: Startup trigger analysis unit; 12b: Processing execution unit; 12c: Database; 16: External terminal; 21: CPU; 22: ROM; 23: RAM; 24: Recording medium; 25: External memory drive; 26: Memory device; 27: Input device; 28: Display device; 29: Communication unit; SVI: Service information; DTI: Data information; DSI: Data structure information. Detailed Embodiments

[0010] Hereinafter, the data structure of the data described in the embodiments of the present invention will be explained using drawings.

[0011] (Summary of Data Structure of Data) -3- HK 20137565 A 5 10 15 20 25 First, using FIG1, a summary of the data structure of the data described in this embodiment will be explained. FIG1 is a summary diagram of the data structure of the data described in this embodiment.

[0012] The data structure of the data described in this embodiment is a data structure for data used by an agent (AI agent 12) using AI, characterized in that it contains: service information SVI that can identify the service 14 (service 14a, service 14b, service 14c, ...) used by AI; and data information DTI that can identify the data used by AI (e.g., data used in service 14); the agent 12 specifies the combination of service information SVI and data information DTI based on the analysis result of the given initiation opportunity, based on the specified service information SVI Accessing the service and obtaining data specific to the data information DTI from the service; this is used in the process of performing the processing for the initiation trigger based on the obtained data.

[0013] According to the data structure of the data described in this embodiment, by selecting the best service from the various services utilized by the AI, the quality or speed of processing the initiation trigger such as the user's prompt can be improved.

[0014] Here, the so-called "AI agent" refers to a system that acts as a human agent, has the ability to learn and make judgments on its own, and integrates various AI technologies to solve complex problems.

[0015] The AI ​​agent performs processing based on an initiation trigger given from the outside (e.g., a question (prompt) from the user, signal input from a sensor), or based on an initiation trigger given from the inside (e.g., initiation caused by an internal scheduler, occurrence of an anomaly).

[0016] For example, As an external trigger (event), once a question (prompt) is given, it will perform tasks such as retrieving or generating an answer to the question, and output the answer to the question.Furthermore, for example, as an external trigger (event), -4- HK 20137565 A 5 10 15 20 25, once given a topic (prompt), will perform tasks such as retrieving solutions to the topic or generating solutions, and output the solutions to the topic.

[0017] Furthermore, the so-called "prompt" refers to the instruction or question given to the system by the user, including, for example, AI prompts given to generative AI, or command prompt characters given through command sentences (instructions).

[0018] In addition, the "trigger" is not limited to a trigger given based on an external event (such as a question (prompt) from the user, or a signal input from a sensor), but can also be a trigger given based on an internal event (such as a start-up caused by an internal scheduler, or the occurrence of an anomaly).

[0019] Various processes performed by AI agents can also be performed by "proxy AI".

[0020] Here, "proxy AI" refers to a system in which multiple AI agents coordinate their actions. In proxy AI, in order to solve complex problems that a single AI agent cannot solve, each AI agent plays a specialized role, exchanging information with each other while outputting answers to questions or solutions to problems.

[0021] Furthermore, "service" refers to infrastructure environments, platforms, etc., such as software, databases, or storage provided from the outside, as "services", such as SaaS (Software as a Service), databases, data warehouses, storage services, search engines, etc.

[0022] "Data used by AI" includes both data stored in an external terminal that is different from the terminal where the AI ​​agent is installed, and data stored in the memory of the terminal where the AI ​​agent is installed.Furthermore, as "data utilized by AI", examples include: data referenced or read / written when AI utilizes external services, or documents created by an individual using their own terminal (e.g., text data of meeting materials). "Data information" can also be, -5- HK 20137565 A 5 10 15 20 25, dynamically assigning tags to and managing personalized data (e.g., text data of meeting materials) stored on an individual's terminal.

[0023] In addition, the data information DTI contains a service information ID representing the service information SVI. When the agent 12 needs to specify the combination of the service information SVI and the data information DTI, the data information DTI is specified based on the parsing result of the activation trigger, and the service information SVI is specified based on the service information ID contained in the specified data information DTI; this can also be used in this process.

[0024] If configured in this way, when the agent performs processing for a launch event, after specifying the data associated with the launch event, it can specify the service to which that data is stored, thereby improving the quality or speed of processing launch events such as user prompts.

[0025] Furthermore, it may also include data structure information (DSI) representing the structure of the data, and for the data information (DTI), data structure information is associated.

[0026] If configured in this way, the information representing the structure of the data can be bound to the data, so when the agent performs processing for a launch event, processing suitable for the data structure can be performed, thereby improving the quality or speed of processing launch events such as user prompts.

[0027] Alternatively, the pre-registration agent may, based on the analysis result of the pre-registration initiation trigger and in a specific case, combine the pre-registration service information and pre-registration data information, vectorize the data contained in the analysis result of the pre-registration initiation trigger, and obtain the pre-registration service information and pre-registration data information from the data stored in the vector database.

[0028] If configured in this way, the quality or speed of service information and data information retrieval can be improved, and the quality or speed of processing initiation triggers such as user prompts can be improved.

[0029] -6- HK 20137565 A 5 10 15 20 25 Furthermore, Agent 12 can also be an AI agent or a proxy AI, and the service can also include SaaS, database, data warehouse, and storage services.

[0030] If configured in this way, by utilizing a variety of external services through an AI agent or a proxy AI, the quality or speed of processing triggers such as prompts for users can be improved.

[0031] Here, "AI agent" refers to autonomous artificial intelligence (AI) that acts as an agent (agent) for humans. For example, once an AI agent is given a question (prompt) by a user, it will retrieve the answer to the question, generate the answer, and output the answer to the question. Furthermore, for example, once a task (prompt) is given to a user, solutions to the task will be retrieved, generated, or output.

[0032] Furthermore, the so-called "proxy AI" refers to a system in which multiple AI agents coordinate their actions. In order to solve complex problems that a single AI agent cannot solve, each AI agent plays a specialized role, exchanging information with each other while outputting answers to questions or solutions to tasks.

[0033] Furthermore, agent 12 can also learn the relationship between the combination of prompts, service information SVI, and data information DTI through machine learning.

[0034] If configured in this way, the best service can be quickly selected from the various services used by the AI, which can improve the quality or speed of processing prompts for users.

[0035] Furthermore, Agent 12 can also perform data processing on operational metrics or data processing on purchase history as processing for initiation opportunities.

[0036] -7- HK 20137565 A 5 10 15 20 25 If configured in this way, the quality or speed of processing can be improved when using AI to perform operational analysis or purchase analysis.

[0037] Here, “operational values” refers to values ​​that represent the operating status of an enterprise, and are values ​​used to evaluate the performance of an enterprise through operating indicators or financial indicators. Furthermore, "data on operational metrics" refers to data (numerical or string information, etc.) representing the operational status of an enterprise. This data includes, for example: (1) sales / profit related data (e.g., sales data, profit data); (2) sales / product management related data (e.g., product specifications, shelf configuration data, sales performance and promotional effectiveness); (3) inventory / purchase related data (e.g., inventory status, order / purchase data); (4) customer / marketing related data (e.g., customer ID data (ID-POS), marketing results); (5) store operation / personnel management related data (e.g., store performance, staff scheduling); (6) expense / financial related data (e.g., store / headquarters expenses, financial indicators); (7) external environment data (e.g., business district / population data, competitor store location / price information, weather / ...). (8) Information on regional activities / school routine activities; (9) Data related to instructions / communication (e.g., business instructions / policies from superiors, reports / proposals from the field, and cooperation history with headquarters).

[0038] Furthermore, in the "data on operating values", there may be cases where sales / product management related data (e.g., product master specifications, shelf configuration data, sales performance and promotion effects) are referred to as "data on promotion (sales promotion)"; in the "data on operating values", there may be cases where customer / marketing related data (e.g., customer ID data (ID-POS), marketing results) are referred to as "data on purchase history".

[0039] (System Configuration Example) Next, using FIG2, a configuration example of the information processing system 10 to which the data structure of the data described in this embodiment can be applied will be described. FIG2 is a system configuration diagram of an information processing system 10 to which the data structure of the data described in this embodiment can be applied.

[0040] -8-HK 20137565 A 5 10 15 20 25 The information processing system 10 may, for example, be configured to have: a system terminal 12 controlling the entire system, and multiple services 14 and external terminals 16 that are connected to the system terminal 12 through a network NW and can communicate with each other.

[0041] System terminal 12 is a terminal that constitutes the AI ​​agent and is the terminal that controls the entire information processing system 10. It is composed of previously known servers or personal computers. In addition, although system terminal 12 is constructed as one server in this example, it can also be constructed as multiple servers or personal computers. The hardware configuration of system terminal 12 and the processing performed by system terminal 12 will be described later.

[0042] Multiple services 14 (14a, 14b, 14c, ...) are external services that system terminal 12 can utilize. As mentioned above, "service" refers to the infrastructure environment, platform, etc., such as software, database, or storage provided from the outside, as a "service". For example, SaaS (Software as a Service), database, data warehouse, storage service, search engine, etc. are equivalent.

[0043] Furthermore, the "service" described in this invention is not limited to cloud services provided via the Internet, but may also be services other than (or alternative to) these services, such as services provided by terminals connected to system terminal 12 via a local area network (e.g., LAN or P2P) to communicate with each other.

[0044] External terminal 16 is a terminal used by users of information processing system 10 (individuals, or specific groups (teams, departments, enterprises, etc.) of multiple people), and is composed of personal computers, tablet computers, smartphones, etc. The type of external terminal 16 is not particularly limited; for example, smartphones, personal computers, tablet computers used by individuals, or smartphones, personal computers, tablet computers used by employees of enterprises or specific groups are equivalent.

[0045] The network NW is a line that allows the system terminal 12, service 14, and external terminal 16 to communicate with each other. Typically, it is composed of a WAN (Wide Area Network), also known as the Internet. -9- HK 20137565 A 5 10 15 20 25 In addition, the network NW can be wired / wireless, or it can be a LAN (Local Area Network), a VPN (Virtual Private Network), or a combination of these lines.

[0046] (System Terminal / Hardware Configuration Example) The hardware configuration example of the system terminal 12 will be described next.

[0047] The system terminal 12, as shown in FIG2, is configured to include, for example, a CPU 21, and ROM 22, RAM 23, external memory drive 25, memory device 26, input device 27, display device 28, communication unit 29, etc., which are connected to the CPU 21 via a bus.

[0048] The CPU 21 is a control means for controlling the entire system terminal 12, and performs processes such as executing applications or operating systems (OS) stored in ROM 22 or memory device 26, or storing data or files necessary for program execution into RAM 23 or memory device 26.

[0049] ROM 22 is a memory means for storing basic I / O programs or various data, and is configured using, for example, PROM or flash memory. RAM 23 is a memory means for temporarily storing data, and is configured using, for example, SDRAM or DRAM. The external memory drive 25 is a control means for reading and writing data to the recording medium 24, such as magnetic tape or DVD, and is configured using, for example, a magnetic tape storage device or a DVD drive.

[0050] The memory device 26 is a memory means for storing applications, OS, control programs, associated programs, various information, etc., and is configured using, for example, a hard disk drive (HDD) or a solid-state drive (SDD). The input device 27 is a device for inputting commands (instructions) to the system terminal 12, and is configured using, for example, a keyboard, a pointing device (mouse, etc.), or a touch panel.

[0051] The display device 28 displays commands input via the input device 27, responses from the system terminal 12 to those commands, or various other displays. It is configured using, for example, a liquid crystal display (LCD), a plasma display, or an organic EL. The communication unit 29 is a control means for controlling communication with the service 14 or external terminal 16 via the network NW. It is configured using, for example, a communication card.

[0052] (Data Structure of System Terminal / Data) Next, the data structure of the data stored in the database 12c will be explained using FIGS. 1 to 3. FIG. 3 is an illustration of an example of the use of the data described in this embodiment.

[0053] As shown in FIG1, the database 12c of the system terminal 12 stores the following data: Service Information (SVI), Data Information (DTI), and Data Structure Information (DSI). Furthermore, the database 12c can also be connected to the system terminal 12 via the Internet, or via a local area network (e.g., LAN or P2P), or it can be the memory device 26 shown in FIG2.

[0054] (System Terminal / Data Structure / Service Information) Next, the service information SVI of the data will be explained.

[0055] The service information SVI is information that can identify the services used by the system terminal 12 (AI agent). As shown in FIG1, in this example, the service information SVI contains: the identification element required to identify the service information, i.e., the service information ID, and the name associated with the service information ID to represent the name of the service, and the category representing the type of the service, etc.

[0056] For example, the service information SVI14a of service 14a contains: service information ID=0001, name=service 14a, category=data warehouse, etc.; the service information SVI14b of service 14b contains: service information ID=0002, name=service 14b, category=database, etc.

[0057] Also, as shown in FIG3, the service information SVI14b of service 14c contains: service information ID=0002, name=service 14b, category=database, etc. The service information SVI 14c is remembered as follows: Service Information ID=0003, Name=Service 14c, Type=SaaS, etc.

[0058] -11- HK 20137565 A 5 10 15 20 25 《System Terminal / Data Structure / Data Information》 Next, the data information DTI will be explained.

[0059] The data information DTI is information that can be used to identify the data stored (remembered) in the service. As shown in Figure 1, in this example, the data information DTI is remembered as follows: the identification element required to identify the data information, namely the data information ID, and the name associated with the data information ID to represent the name of the dataset (data), and the service information ID of the service information SVI associated with the data information DTI, and the data indicating the location where the data is stored, etc.

[0060] Furthermore, since the data information DTI contains the service information ID of the service information SVI associated with the data information DTI, the name or type of the service to which the dataset (data) of the data information DTI is stored can be obtained based on the data information DTI.

[0061] For example, the data information DTIa of data a stored in service 14a contains: data information ID=0001, name=data a, service information ID=0001 (=service information ID of service information SVI14a of service 14a), location=service14a: / / x / / a.csv, etc.; the data information DTIa of data a is associated with the service information SVI14a of service 14a.

[0062] Furthermore, the data information DTIb of data b stored in service 14b is stored as follows: Data information ID=0002. Name: data b, Service information ID=0002 (service information ID of service information SVI14b of service 14b), Location=service 14b: / / y / / b.csv, etc.; The data information DTIb of data b is associated with the service information SV114b of service 14b.

[0063] Furthermore, as shown in FIG3, the data information DTIc of data c stored in service 14c is stored as follows: data information ID=0003, name=data c (e.g., budget data), service information IDR003 (service information ID of service information SVI14c of service 14c), location=service 14c: / / z / / c.csv, etc.; the data information DTIc of data c is associated with the service information SVI14c of service 14c. -12- HK 20137565 A 5 10 15 20 25

[0064] Also, as shown in Figure 3, the data information DTId of the data d stored in service 14c is stored as follows: data information ID=0004, name=data d (e.g., performance data), service information ID=0003 (=service information ID of service information SVI14c of service 14c), location=service 14c: / / z / / d.csv, etc.; the data information DTId of data d is associated with the service information SVI14c of service 14c.

[0065] 《System Terminal / Data Data Structure / Data Structure Information》 Next, the data structure information DSI of the data will be explained.

[0066] Data structure information DSI is information that represents the structure of the data. As shown in Figure 1, in this example, the data structure information DSI contains: the identification element required to identify the data structure information, namely the data structure information ID, and the data information ID of the data information DTI associated with the data structure information DSI, which is associated with the data structure information ID, and the name representing the field name, and the data type representing the data type, etc.

[0067] For example, the data structure information DSIa of data a stored in service 14a contains: data structure information ID=0001, data information IDR001 (= data information ID of data information DTIa of data a stored in service 14a), name=aaa, type=int, etc.; the data information DTIa of data a is associated with the data structure information DSIa.

[0068] Furthermore, the data structure information DSI1b of data b stored in service 14b is stored as follows: data structure information ID=0002, data information ID=0002 (the data information ID of the data information DTIb of data b stored in service 14b), name=bbb, type=strings, etc.; the data information DTIb of data b and the data structure information DSIb are associated.

[0069] Furthermore, the data structure information DSIc of the data c stored in service 14c is stored as follows: data structure information ID=0003, data information ID=0003 (the data information ID of the data information DTIc of the data c stored in service 14c), name=ccc, type=timestamp, etc.; the data information DTIc of data c is associated with the data structure information DSIc.

[0070] Furthermore, the data structure information DSId of the data d stored in service 14c is stored as follows: data structure information ID=0004, data information ID=0003 (data information ID of the data information DTId of the data d stored in service 14c), name=ddd, type=int, etc.; the data information DTId of data d and the data structure information DSId are associated.

[0071] 《System Terminal / Software Configuration Example》 Next, the software configuration example of system terminal 12 will be described.

[0072] As shown in FIG3, system terminal 12 is configured with: a startup trigger parsing unit 12a that processes the prompt input by the user, a processing execution unit 12b that uses AI to perform processing for the prompt, and a database 12c (vector database) used by the processing execution unit 12b, etc. Furthermore, database 12c is not limited to vector databases, but can also be a relational database, etc.

[0073] The trigger analysis unit 12a uses an external terminal 16 to obtain the prompts input by the user of the information processing system 10 (individual, or a specific group (team, department, enterprise, etc.) of multiple people), and associates the obtained prompts with the user's ID and stores them in the memory device 26.

[0074] Here, "prompt" refers to the instructions or questions (data) input by the user of the information processing system 10 using the input means (keyboard, indicator device (mouse, etc.), touch panel, microphone, etc.) of the external terminal 16. The data format of "prompt PRM" is not particularly limited, and can be, for example, any of the following: text data, voice data, still image data, video data, documents, etc.

[0075] Examples of "prompt PRM" include: text or voice data such as instructions or questions to the information processing system 10, files uploaded by the user, still image data, or video data.

[0076] Next, the trigger analysis unit 12a is activated to analyze the prompts stored in the memory device 26, generate the analysis result, and associate the analysis result with the user's ID and store it in the memory device 26.

[0077] For example, as shown in FIG3, if we consider a case where a user is given the prompt "I want to see the budget and performance data for this period", the trigger analysis unit 12a uses the text data contained in the prompt to specify the content of the prompt (in this example, obtaining and comparing the budget and performance for this period, etc.), and uses the specified content as the result of the prompt analysis, and associates it with the user's ID and stores it in the memory device 26.

[0078] Next, the processing execution unit 12b refers to the result of the prompt analysis stored in the memory device 26 by the trigger analysis unit 12a to perform the processing for the prompt.

[0079] For example, if the memory of the prompt is "to obtain and compare the current period's budget and actual results", then the processing execution unit 12b specifies the combination of service information SVI and data information DTI based on the prompt analysis result and past machine learning.

[0080] Specifically, since the prompt contains the text data "the current period's budget", the text data is vectorized, and data information DTIc (data c (the current period's budget data)) is specified (retrieved) from the database 12c. Based on the service information ID (0003 in this example) contained in the specified data information DTIc, service information SVI14c (service 14c) is specified.

[0081] Meanwhile, the processing execution unit 12b learns the relationship between the combination of the prompt (budget for this period), service information SVI14c (service 14c), and data information DTIc (data c (budget data for this period)) using machine learning.

[0082] -15- HK 20137565 A 5 10 15 20 25 Furthermore, since the prompt contains the text data "performance for this period", the text data is vectorized, and data information DTId (data d (performance data for this period)) is specifically (retrieved) from the database 12c. Based on the service information ID (0003 in this example) contained in the specific data information DTId, the service information SVI14c (service 14c) is specified.

[0083] Meanwhile, the processing execution unit 12b learns the relationship between the combination of the prompt (performance data for the current period), the service information SVI14c (service 14c), and the data information DTId (data d (performance data for the current period)) through machine learning.

[0084] Through such processing, the processing execution unit 12b, based on the parsing results of the prompt (obtaining and comparing the current period's budget and actual performance), specifies the combination of service information SVI14c (service 14c) and data information DTIc (data c (the current period's budget data)) and the combination of service information SVI14c (service 14c) and data information DTId (data d (the current period's actual performance data)).

[0085] Next, the processing execution unit 12b accesses the service 14c (in this example, SaaS) based on the specified service information SVI14c, and from the service 14c, obtains the data c (the current period's budget data) specified in the storage location based on the data information DTIc, and obtains the data d (the current period's actual performance data) specified in the storage location based on the data information DTId.

[0086] Next, the processing execution unit 12b, based on the specific data c (budget data for this period) and data d (performance data for this period), performs prompt processing, generates a response such as "The budget for this period is ** dollars, the performance is ** dollars, and the ratio of performance to budget is **%", and sends it to the external terminal 16 used by the user.

[0087] According to this example, by selecting the best service from the various services used by the AI, the quality or speed of prompt processing for the user can be improved.

[0088] 《Data Structure / Summary》 -16- HK 20137565 A 5 10 15 20 25 As explained above, the data structure of the data described in this embodiment (e.g., the data shown in Figure 1) is a data structure for data used by an AI agent (e.g., AI agent 12 shown in Figure 1). It is characterized by containing: service information (e.g., service information SVI shown in Figure 1) that can identify the services utilized by the previous AI (e.g., service 14 shown in Figure 1); and data information (e.g., data information DTI shown in Figure 1) that can identify the data utilized by the previous AI (used in the previous service). The previous agent, based on the parsing result of the given initiation trigger (e.g., prompt), specifies the combination of previous service information and previous data information, and accesses the previous service based on the specified previous service information. And obtain the prior data specific based on the prior data information from the service; it will be used in the process of performing processing for prior initiation opportunities based on the obtained prior data.

[0089] According to the data structure of the data described in this embodiment, by selecting the best service from the various services used by the AI, the quality or speed of processing for initiation opportunities such as prompts for users can be improved.

[0090] Alternatively, the prior data information may contain a service information ID representing prior service information. When the prior agent wants to specify the combination of prior service information and prior data information, the prior data information is specified based on the parsing result of the prior initiation opportunity, and the prior service information is specified based on the prior service information ID contained in the specified data information.

[0091] If configured in this way, when the agent performs processing for the activation trigger, after specifying the data associated with the activation trigger, it can specify the service to which the data is stored, thereby improving the quality or speed of processing activation triggers such as prompts for users.

[0092] Furthermore, it may also contain data structure information (e.g., data structure information DSI shown in FIG1) representing the structure of the preceding data (e.g., the data type shown in FIG1); the preceding data information is associated with the preceding data structure information.-17- HK 20137565 A 5 10 15 20 25

[0093] If configured in this way, the information representing the structure of the data can be bound to the data. Therefore, when the agent performs processing for the activation opportunity, processing suitable for the data structure can be performed, which can improve the quality or speed of processing activation opportunities such as prompts for users.

[0094] Furthermore, it can also be configured such that, based on the parsing result of the activation opportunity, the previous agent, when combining the previous service information and the previous data information in a specific way, vectorizes the data contained in the parsing result of the activation opportunity and obtains the previous service information and the previous data information from the data stored in the vector database (for example, the database 12c shown in FIG1).

[0095] If configured in this way, the quality or speed of retrieving service information and data information can be improved, and the quality or speed of processing activation opportunities such as prompts for users can be improved.

[0096] Furthermore, the pre-recorded agent can also be an AI agent or a proxy-type AI, and the pre-recorded service can also include SaaS, database, data warehouse, and storage services.

[0097] If configured in this way, the quality or speed of processing triggers such as prompts for users can be improved by utilizing a variety of external services.

[0098] Furthermore, the pre-recorded agent can also learn the relationship between the combination of pre-recorded triggers, pre-recorded service information, and pre-recorded data information through machine learning.

[0099] If configured in this way, the best service can be quickly selected from the various services utilized by the AI, thereby improving the quality or speed of processing triggers such as prompts for users.

[0100] In addition, the data structure of the data described in this invention is not limited to the above-described implementation form, and various modifications can be made without departing from the spirit of this invention. -18- HK 20137565 A 5 10 15 20

[0101] Therefore, for example, in the above-described embodiment, although the data is constructed as service information, data information, and data structure information, it can also be constructed as service information and data information. Furthermore, the agent using AI is not limited to an AI agent; it can also be a proxy-type AI. Also, the "data" described in this invention is not limited to the data used in the service; it can be any data utilized by the AI.

[0102] Furthermore, although an example has been shown where, in the case of specifying a combination of service information and data information, data information is specified based on the result of the prompting analysis, and service information is specified based on the service information ID contained in the specified data information, the present invention is not limited to this. Service information may be specified based on the result of the prompting analysis, data structure information may be specified based on the result of the prompting analysis, data information may be specified based on the data ID contained in the specified data structure information, and service information may be specified based on the service information ID contained in the specified data information.

[0103] Furthermore, the database for storing "data" described in the present invention is not limited to a vector database, and may also be a relational database, etc. Furthermore, although the AI ​​agent is configured with a trigger analysis unit and a processing execution unit, the configuration of the AI ​​agent is not limited to this example. [Industrial Applicability]

[0104] The data structure of the data described in the present invention is widely applicable to agents that use AI. -19- HK 20137565 A 5 10 15 20 25 Claim 1. A data structure for data used by an AI agent, characterized in that it contains: service information that identifies the services used by the AI ​​agent; and data information that identifies the data used by the AI ​​agent; the preceding data information contains: a service information ID representing the preceding service information; the AI ​​agent identifies the preceding data information based on the parsing result of a given prompt, and identifies the preceding service information based on the preceding service information ID contained in the identified data information; accesses the preceding service based on the identified preceding service information, and obtains the preceding data identified based on the preceding data information from the service; and is used in the processing of the AI ​​agent when initiating AI processing for the preceding prompt based on the obtained preceding data. 2. The data structure as described in claim 1, further comprising: data structure information representing the structure of the preceding data; and the preceding data information being associated with the preceding data structure information.3. The data structure as described in claim 1 or 2, wherein, when the pre-recording AI agent combines the pre-recording service information and pre-recording data information based on the parsing results of the pre-recording prompts, the data contained in the parsing results of the pre-recording prompts is vectorized, and the pre-recording service information and pre-recording data information are obtained from the data stored in the vector database. 4. The data structure as described in claim 1 or 2, wherein the pre-recording service includes: SaaS, database, data warehouse, and storage services. 5. An information processing system using the data structure as described in claim 1 or 2, wherein the pre-recording AI agent learns the relationship between the combination of pre-recording prompts and the pre-recording service information and the pre-recording data information through machine learning.