Information analysis and interaction system and method

Abstract

The application relates to the technical field of information analysis and interaction, and discloses an information analysis and interaction system and method. The system comprises an instance management module, which is used for configuring independent system instances for different users based on historical data of the users. The system instance comprises a user feature database, a multi-modal input module and a control module. The multi-modal input module is used for receiving original input information input by a user, obtaining structured input data based on the original input information, and inputting the structured input data to the control module. The control module is used for performing semantic analysis based on the structured input data by using a preset large language model, and obtaining user demand according to a semantic analysis result and the user feature database. The control module is also used for determining corresponding function instructions and function modules in a client based on the user demand, and sending the function instructions to the function modules, so that the function modules perform corresponding operations according to the function instructions.

Description

Technical Field

[0001] This application relates to the field of information analysis and interaction technology, and in particular to an information analysis and interaction system and method. Background Technology

[0002] Currently, with the popularization of artificial intelligence and mobile computing technologies, various intelligent assistant systems have been widely integrated into user terminal devices, providing users with services such as voice interaction, information query and application control.

[0003] Existing technologies typically employ a centralized service architecture, where all users share the same logical processing engine. User commands are parsed using a general semantic understanding model, and corresponding operations are executed by calling predefined application programming interfaces (APIs) of the terminal system. This lack of an independent instantiation and configuration mechanism for individual users leads to a homogenization of service experiences and makes it impossible to achieve deep customization and data isolation. Summary of the Invention

[0004] The main purpose of this application is to provide an information analysis and interaction system and method, which aims to solve the existing technical problem of lacking an independent configuration mechanism for individual users.

[0005] To achieve the above objectives, this application provides an information analysis and interaction system, which includes an instance management module and at least one client. Different clients correspond to different users. The instance management module is used to configure an independent system instance for each user based on the user's historical data. The system instance is deployed on the client of the corresponding user, and the system instance includes: a user feature database, a multimodal input module, and a control module; The multimodal input module is connected to the control module, and the control module is connected to the functional modules in the client. A multimodal input module is used to receive raw input information from the user, obtain structured input data based on the raw input information, and input the structured input data to the control module; The control module is used to perform semantic analysis based on the structured input data using a preset large language model, and to obtain user needs based on the semantic analysis results and the user feature database. The control module is further configured to determine the corresponding function instructions and the function modules in the client based on the user requirements, and send the function instructions to the function modules so that the function modules perform the corresponding operations according to the function instructions.

[0006] In one embodiment, the system instance further includes: a communication proxy module connected to the control module, and the functional module includes a communication module connected to the communication proxy module; The communication proxy module is used to obtain the communication content and communication object corresponding to the communication request when the communication module detects that it has received a communication request. The control module is also used to perform real-time semantic analysis on the communication content, and based on the real-time semantic analysis results, the communication object, and the user feature database, to obtain a proxy processing method for the communication request.

[0007] In one embodiment, the functional module includes a prompting module; The control module is also used to determine the object category corresponding to the communication object based on the communication object and the user feature database; The control module is also used to prompt the user with the communication request through the prompting module when the object category is an important customer.

[0008] In one embodiment, the functional module further includes a status acquisition module; The control module is also used to obtain the user's current working status through the status acquisition module when the object category is a general customer; The control module is also used to prompt the user with the communication request through the prompting module when the current working state is an idle state; The control module is also used to generate a corresponding response statement based on the real-time semantic analysis result and the user feature database when the current working state is the working state; The communication proxy module is also used to reply to the communication object according to the response statement through the communication module.

[0009] In one embodiment, the functional module further includes a task list module; The control module is further configured to convert the real-time semantic analysis results into structured to-do items when the object is an unknown customer, and store the to-do items in the user feature database and the task list module.

[0010] In one embodiment, the system instance includes at least a first system instance and a second system instance, wherein the first system instance is deployed on a first client of a first user, and the second system instance is deployed on a second client of a second user; The first system instance includes a first user feature database, a first multimodal input module, a first control module, and a first communication proxy module; the second system instance includes a second user feature database, a second multimodal input module, a second control module, and a second communication proxy module; the first communication proxy module and the second communication proxy module are connected through a communication interface. The first multimodal input module is further configured to determine the second user and the interaction content corresponding to the interaction signal when it receives the interaction signal sent by the first user; The first communication proxy module is further configured to send the interactive content to the second communication proxy module through the communication interface; The second communication proxy module is further configured to obtain first user information corresponding to the first user upon receiving the interaction content; The second control module is further configured to obtain an interaction strategy for the interaction content based on the first user information, the interaction content, and the second user feature database, and drive the corresponding functional module to interact based on the interaction content according to the interaction strategy.

[0011] In one embodiment, the second control module is further configured to query the second user feature database to obtain a preset authorization policy associated with the first user information, and determine whether the interactive content belongs to the preset authorization policy; The second control module is further configured to, when the interactive content belongs to the preset authorization policy, extract task parameters from the interactive content, call the functional module of the second client to perform the corresponding operation according to the task parameters, and generate a processing result; The second communication proxy module is further configured to return the processing result to the first communication proxy module through the communication interface.

[0012] In one embodiment, the second control module is further configured to query the second user feature database to see if historical interaction data corresponding to the first user information exists when the interaction content does not belong to the preset authorization policy; The second control module is further configured to perform semantic similarity matching based on the historical interaction data and the interaction content when the historical interaction data corresponding to the first user information exists in the second user feature database; The second control module is further configured to generate a response scheme based on the semantic similarity matching result, call the functional module of the second client to perform the corresponding operation according to the response scheme, and generate the processing result; The second communication proxy module is further configured to return the processing result to the first communication proxy module through the communication interface.

[0013] In one embodiment, the second control module is further configured to, when the historical interaction data corresponding to the first user information does not exist in the second user feature database, convert the first user information and the interaction content into a structured item to be interacted, and store the item to be interacted in the second user feature database.

[0014] Furthermore, to achieve the above objectives, this application also provides an information analysis and interaction method, which is applied to the information analysis and interaction system described above, and the method includes: Receive raw input information from the user and obtain structured input data based on the raw input information; Semantic analysis is performed on the structured input data using a pre-defined large language model, and user needs are obtained based on the semantic analysis results and the user feature database. Based on the user's needs, the corresponding function instructions and the function modules in the client are determined, and the function instructions are sent to the function modules so that the function modules can perform the corresponding operations according to the function instructions.

[0015] This application provides an information analysis and interaction system and method. The system includes an instance management module for configuring independent system instances for each user based on their historical data. Each system instance is deployed on the client of the corresponding user and includes a user feature database, a multimodal input module, and a control module. The multimodal input module is connected to the control module, and the control module is connected to a functional module in the client. The multimodal input module receives raw input information from the user, obtains structured input data based on the raw input information, and inputs the structured input data to the control module. The control module performs semantic analysis based on the structured input data using a preset large language model and obtains user needs based on the semantic analysis results and the user feature database. The control module also determines corresponding functional instructions and functional modules in the client based on the user needs and sends the functional instructions to the functional modules so that the functional modules perform corresponding operations according to the functional instructions.

[0016] This application's information analysis and interaction system and method can be configured with different system instances deployed on different user clients. Each system instance includes a user feature database, a multimodal input module, and a control module, allowing for the configuration of independent personalized interaction environments for different users. In actual operation, the multimodal input module receives raw user input and transforms it into structured data. The control module performs semantic analysis using a large language model and integrates personalized information from the user feature database to accurately identify user needs, thereby determining and calling the corresponding functional modules to execute instructions. Compared to existing interaction systems that adopt a generalized service architecture, which struggles to balance data isolation and personalized responses in multi-user scenarios and lacks effective utilization of user historical characteristics, this application, by deploying independent instances for each user and introducing collaborative analysis of the feature database and large language model, can achieve highly personalized intent understanding and function scheduling while ensuring data privacy. Consequently, users can directly obtain intelligent interactive services that accurately match their habits and needs when using the client, improving the system's adaptability and user experience. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a structural block diagram of the first embodiment of the information analysis and interaction system of this application; Figure 2 This is a structural block diagram of the second embodiment of the information analysis and interaction system of this application; Figure 3 This is a flowchart of the first embodiment of the information analysis and interaction method of this application.

[0020] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0022] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0023] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of this application are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0024] Furthermore, the use of terms such as "first" and "second" in this application is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. Additionally, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. When the combination of technical solutions is contradictory or impossible to implement, such a combination of technical solutions should be considered non-existent and not within the scope of protection claimed in this application.

[0025] Understandably, with the popularization of artificial intelligence and mobile computing technologies, various intelligent assistant systems have been widely integrated into user terminal devices to provide users with services such as voice interaction, information query and application control.

[0026] Existing technologies typically employ a centralized service architecture, where all users share the same logical processing engine. User commands are parsed using a general semantic understanding model, and corresponding operations are executed by calling predefined application programming interfaces (APIs) of the terminal system. This lack of an independent instantiation and configuration mechanism for individual users leads to a homogenization of service experiences and makes it impossible to achieve deep customization and data isolation.

[0027] To address the aforementioned shortcomings, this embodiment of the information analysis and interaction system and method can be configured with different system instances deployed on different user clients. Each system instance includes a user feature database, a multimodal input module, and a control module, allowing for the configuration of independent personalized interaction environments for different users. In actual operation, the multimodal input module receives raw user input and transforms it into structured data. The control module performs semantic analysis using a large language model and integrates personalized information from the user feature database to accurately identify user needs, thereby determining and calling the corresponding functional modules to execute instructions. Compared to existing interaction systems that employ a generalized service architecture, which struggles to balance data isolation and personalized responses in multi-user scenarios and lacks effective utilization of user historical characteristics, this embodiment, by deploying independent instances for each user and introducing collaborative analysis of the feature database and the large language model, can achieve highly personalized intent understanding and function scheduling while ensuring data privacy. Consequently, users can directly obtain intelligent interactive services that accurately match their habits and needs when using the client, improving the system's adaptability and user experience.

[0028] For ease of understanding, the following is combined with Figures 1 to 3 The information analysis and interaction system provided in the embodiments of this application will be described in detail.

[0029] Reference Figure 1 , Figure 1 This is a structural block diagram of the first embodiment of the information analysis and interaction system of this application. Figure 1 As shown, in this embodiment, the system includes an instance management module and at least one client. Different clients correspond to different users. The instance management module is used to configure an independent system instance for each user based on the user's historical data.

[0030] It should be noted that the aforementioned instance management module can be a server-side program component or a client-side management module that creates and maintains independent system instances for different users. These different users can be natural persons or organizational entities with independent identities who require personalized services, such as project manager Zhang San or R&D engineer Li Si in a company, or virtual user roles corresponding to a department as a whole business unit. The aforementioned historical data can be behavioral records and content data generated by users during past interactions that can be collected and parsed by the system. Examples include voice commands transcribed into text via the multimodal input module over the past three months, call logs and email exchanges with the communication agent module, and task logs and decision results generated by the system during the execution of user requests.

[0031] The aforementioned system instance can be a personalized system runtime carrier that has exclusive computing resources, storage space, and runtime context, and is logically and data isolated from other instances. For example, the client program entity deployed on Wang Wu's personal workstation contains a user feature database, control module instance, and communication agent module instance that belong only to Wang Wu. Its internal state and data assets are not shared with other users' system instances.

[0032] In its implementation, the instance management module first obtains the identity identifiers of different users and their corresponding historical data. Then, it allocates dedicated storage resources and runtime environments for each user. Next, the instance management module preprocesses, extracts features, and vectorizes the historical data of each user before writing it into the user feature database of the newly created system instance, completing the initial configuration of that instance. At this point, each user obtains an independent system instance bound to them, based on their historical data, and completely isolated from other users.

[0033] The system instance is deployed on the client of the corresponding user, and the system instance includes: a user feature database, a multimodal input module, and a control module; The multimodal input module is connected to the control module, and the control module is connected to the functional modules in the client. A multimodal input module is used to receive raw input information from the user, obtain structured input data based on the raw input information, and input the structured input data to the control module; The control module is used to perform semantic analysis based on the structured input data using a preset large language model, and to obtain user needs based on the semantic analysis results and the user feature database. The control module is further configured to determine the corresponding function instructions and the function modules in the client based on the user requirements, and send the function instructions to the function modules so that the function modules perform the corresponding operations according to the function instructions.

[0034] It should be noted that the aforementioned client can be the terminal device and its operating system environment that hosts the system instance, such as Zhao Liu's personal laptop, smartphone, or workstation. The aforementioned user feature database can be a non-relational data storage component dedicated to the current system instance, used to persistently store the user's personalized feature information, such as an embedded database based on a vectorization engine. This database stores Zhao Liu's historical interaction records, document semantic slices, and preference configurations in high-dimensional vector form. The aforementioned multimodal input module can be a front-end processing component integrated into the system instance, used to receive and transform the user's raw input information in multiple modalities, such as a microphone interface supporting voice stream acquisition and endpoint detection, an OCR recognition unit supporting image file parsing, or a media processing plugin supporting video keyframe extraction. The aforementioned control module can be the core decision-making and scheduling component of the system instance. This component maintains communication connections with the large language model service, the user feature database, and the client functional modules, such as a large model inference engine scheduler deployed locally on Zhao Liu's client, which is responsible for performing semantic analysis, task planning, and instruction routing.

[0035] Furthermore, it should be noted that the functional modules in the aforementioned client can be applications or system service units with specific business capabilities that are registered with or dynamically invoked by the client operating system, such as calendar applications, email clients, instant messaging tools, task management software, file managers, or system settings services. The aforementioned raw input information can be the initial form of information directly entered by the user through the human-computer interaction interface without system semantic parsing, such as Zhao Liu speaking into the microphone, "Please book a meeting room with Zhang Hui next Tuesday at 3 PM," or a photo taken by Zhao Liu using his mobile phone containing handwritten meeting notes, uploaded through the file selector.

[0036] It is understandable that the aforementioned structured input data can be text or key-value pair data generated by the multimodal input module after converting the original input information, conforming to a predefined format, and directly parsable by the control module. For example, the text string "Help me book a meeting with Zhang Hui next Tuesday at 3 PM" generated after the aforementioned speech stream undergoes speech recognition, or a JSON data structure containing "time, location, and attendees" fields output after the aforementioned handwritten record image undergoes OCR recognition. The aforementioned preset large language model can be a deep neural network model pre-trained on a large-scale corpus, possessing general semantic understanding and text generation capabilities. The aforementioned semantic analysis can be the calculation process by which the control module calls the large language model to perform intent recognition, entity extraction, and context understanding on the structured input data. For example, parsing "Help me book a meeting with Zhang Hui next Tuesday at 3 PM" into a structured representation of [intent: meeting room reservation], [entity: time = next Tuesday 15:00], and [entity: attendee = Zhang Hui]. The semantic analysis results mentioned above can be intermediate products output by a large language model after semantic analysis, containing user intent categories and key information entities, such as the aforementioned set of parsed intent tags and entity key-value pairs.

[0037] It should also be understood that the aforementioned user requirements can be user goal descriptions generated by the control module after fusing semantic analysis results with personalized information stored in the user feature database, and can be further transformed into specific tasks. For example, after querying Zhao Liu's feature database, the control module finds that "Zhang Hui" belongs to the "Product Review Committee" and that next Tuesday is a project milestone. The resulting requirement is "Reserve a projection room that can accommodate eight people for the product review meeting and automatically add review committee members as optional attendees." The aforementioned functional instructions can be executable command sequences generated by the control module based on user requirements and directed to specific functional modules. For example, the instruction "Create Event: Title = Product Review Meeting, Time = 2026-02-17, 15:00-16:00, Location = Pending" generated for a calendar application, and the instruction "Send meeting invitations to Zhang Hui and other review committee members" generated for an email client. The corresponding operations can be atomic business actions executed by the functional module after receiving the functional instructions, producing specific effects locally on the client. For example, the calendar application creates a new meeting event entry in Zhao Liu's schedule, or the email client sends a meeting invitation email to a designated contact.

[0038] In practical applications, firstly, the system instance deployed on the user client continuously monitors its multimodal input module. When the user inputs raw information via any modality such as voice, text, or image, the multimodal input module immediately initiates a conversion process, parsing the raw information into standardized structured input data. After conversion, the multimodal input module pushes the structured input data to the control module directly connected to it.

[0039] Subsequently, upon receiving the structured input data, the control module immediately invokes a pre-defined large language model service to perform deep semantic analysis on the data. After the large language model completes the intent recognition and entity extraction of the user's command, it returns the semantic analysis results to the control module. The control module then uses the key information in the semantic analysis results as query conditions to actively retrieve and read the user feature database specific to the current system instance, obtaining the user's historical preferences, contextual constraints, and personalized configurations related to the current task. The control module fuses and infers the semantic analysis results with the database query results, ultimately generating a precise description of the user's needs.

[0040] Finally, the control module parses the user's requirements, breaking them down into a series of atomic executable steps, and matching each step with a registered functional module in the client that possesses the corresponding capabilities. The control module then generates corresponding functional instructions for each module. Next, the control module distributes these instructions to the corresponding functional modules via inter-process communication or system service calls. Upon receiving the instructions, each functional module executes the specific operations locally on the client according to the instruction parameters, completing the interaction with the user's requirements.

[0041] In this embodiment, the user feature database is also used to store the user's digital asset information, which includes: personal information, enterprise information, asset information, and e-commerce information.

[0042] The control module is also used to perform asset management operations on the digital asset information, including: asset operation, asset management, asset encryption, transfer of asset ownership, asset recall, asset destruction, and asset registration.

[0043] The control module uses the privacy module and the compliance module to encrypt and verify the asset management operations.

[0044] It should be noted that the aforementioned digital asset information can be various electronic property records owned by individuals or organizations in the digital space, such as scanned copies of a user's ID card, bank account information, digital certificates, cryptocurrency wallet private keys, electronic contracts, intellectual property registration certificates, domain name holding records, game virtual equipment lists, and e-commerce platform membership certificates. The aforementioned asset management operations can be the full lifecycle management functions of digital assets provided by the control module for users. For example, based on the user's voice command "transfer my domain name to the company's name," the control module can automatically generate the authorization documents required for domain name transfer, call the domain registrar's API to initiate an ownership transfer application, and file the transfer record in the asset filing sub-database of the user's characteristic database; or, after detecting the loss of a user's device, the control module can automatically send an asset recall command to the lost device based on the user's command to "recall all personal assets" issued through other devices, triggering the privacy module to erase all locally stored sensitive asset information on the device and transfer the encrypted backup asset data to a new device designated by the user.

[0045] For ease of understanding, the following example illustrates the concept, but does not impose specific limitations on this embodiment. Product Manager Zhao Liu opens the system client on his personal workstation. Zhao Liu speaks into the microphone: "Help me check the processing progress of the user feedback from last week regarding the issue of 'exporting reports too slowly'." The microphone array on Zhao Liu's workstation, acting as a multimodal input module, captures this speech stream and immediately calls the local speech recognition engine to convert the audio into the text string "Help me check the processing progress of the user feedback from last week regarding the issue of 'exporting reports too slowly'." The multimodal input module encapsulates this text string into a standard input data structure and pushes it to the control module. Upon receiving the text command, the control module initiates a semantic analysis request through the enterprise's internally deployed large language model API. The parsing results returned by the large model show: [Intent: Query task progress], [Object: User feedback], [Topic: Export report performance], [Time constraint: Last week]. The control module then queries the user feature database exclusive to Zhao Liu's instance using "export report," "performance optimization," and "last week" as keywords. The database returned three related records: Last week, the product team converted the feedback into a Jira ticket with the number PERF-2026-0210. The ticket's current status is "Under Development and Testing". The assigned backend engineer, Wang Hai, is expected to complete the testing today.

[0046] The control module integrates the large model parsing results with the database query results to generate a complete user requirement: "Zhao Liu needs to query the current processing progress and estimated completion time of performance optimization work order PERF-2026-0210." Next, the control module parses this requirement and identifies the need to call two functional modules: a Jira client to obtain the real-time status of the work order, and a client to initiate a lightweight confirmation to Wang Hai. The control module sends a function command to the Jira client: "Query work order PERF-2026-0210, return the status field, assignee field, and update log." The Jira client executes the query and returns the results to the control module. The control module then sends a function command to the client: "Send a message to contact 'Wang Hai' with the content 'Mr. Wang, can the PERF-2026-0210 report export optimization work order be tested today?'"

[0047] The client sends a message, and Wang Hai replies, "Testing is expected before 4 PM." The client then sends its reply back to the control module. The control module summarizes the Jira work order status and Wang Hai's reply, generating a complete conclusion. At this point, Zhao Liu's original voice command has been fully executed by the system, and each functional module has completed its corresponding operation according to the control module's instructions.

[0048] It should be noted that, in this embodiment, the system instance also includes a privacy module and a compliance module.

[0049] The privacy module and the compliance module are interconnected, and both are connected to the user feature database, the control module, and the functional modules in the client.

[0050] The privacy module is used to perform full lifecycle management of user privacy data in the system instance. The full lifecycle management includes: data auditing, data encryption, data distribution, data retrieval, and data destruction. When the privacy module communicates with other modules in the system instance and external systems, it uniformly follows the preset encryption, distribution, and retrieval protocols.

[0051] The compliance module is used to perform compliance verification on the data processing behavior in the system instance. The compliance verification includes: data collection legality verification, data processing permission verification, data export compliance verification, and data retention period verification. The compliance module monitors the decision-making behavior of the control module and the operation behavior of the functional module in real time based on the compliance policies stored in the user feature database and the compliance protocol synchronized by the external public information management and control module.

[0052] It should be noted that the aforementioned privacy module can be a software component embedded within the control module, specifically responsible for the security management of user privacy data. For example, it could be an encryption engine responsible for AES-256 encryption of sensitive data such as user biometric information, contacts, and location data, or an anonymization processor responsible for de-identifying data packets before cross-instance data transmission. The aforementioned encryption, distribution, and recall protocols can be unified data security standards that all privacy modules of all system instances adhere to. For example, they could stipulate that all cross-instance transmissions involving user identity information must use end-to-end encrypted tunnels, and that the data recipient must delete a copy of the data within a specified time after use, with an automatic recall mechanism.

[0053] In addition, the aforementioned compliance module can be a decision support component within the control module specifically designed for ensuring legal compliance. For example, before the control module calls the large language model to process user instructions, the compliance module first performs sensitive word filtering and minor protection verification on the original input data; or before the control module shares data from the user feature database with third-party applications, the compliance module performs cross-validation based on the authorization records of the Personal Information Protection Law stored in the user feature database and the scope of authorization explicitly agreed upon by the user, and only allows the operation to proceed if it is completely within the scope of authorization.

[0054] Furthermore, in this embodiment, the multimodal input module is also used to acquire sensor data in real time, and the sensor data includes at least one of real-time video data, real-time audio data, speed data, gravity data, pressure data, temperature data, and position data.

[0055] The control module is also used to actively generate feedback information based on the sensor data, the tasks stored in the task list module, and the historical data stored in the user feature database, and push the feedback information to the functional module in the client, so as to provide timely feedback on task progress or push network information streams to the user through the functional module.

[0056] It should be noted that the aforementioned sensor data can be real-time environmental parameters and user status parameters collected by the client device where the system instance is located through its hardware sensors. Examples include: user gaze direction and facial expression video streams collected by the front-facing camera of a laptop; ambient noise decibel values ​​and user voice activity detection data collected by the microphone array; device motion trajectory and jitter frequency collected by the accelerometer; latitude and longitude coordinates and movement speed collected by the GPS module; and CPU core temperature and ambient temperature collected by the temperature sensor. The aforementioned proactively generated feedback information can be prompts or operations proactively triggered by the control module based on context awareness without receiving explicit user instructions. For example, when the control module detects that the user's location data is continuously moving towards the airport and that there is a "business trip luggage checklist" to be done in the task list, it proactively pushes a prompt card saying, "You are about to go to the airport, would you like to check your luggage checklist?"; or when the control module detects that the user is driving and the vehicle speed exceeds 60 km / h, it automatically transfers incoming calls to voicemail and sends an automatic response message saying, "Driving, will reply later."

[0057] Furthermore, the system instance also includes: a communication proxy module, which is connected to the control module; the functional module includes a communication module, which is connected to the communication proxy module. The communication proxy module is used to obtain the communication content and communication object corresponding to the communication request when the communication module detects that it has received a communication request. The control module is also used to perform real-time semantic analysis on the communication content, and based on the real-time semantic analysis results, the communication object, and the user feature database, to obtain a proxy processing method for the communication request.

[0058] It should be understood that the aforementioned communication proxy module can be an edge processing unit within the system instance dedicated to communication channel access, communication request interception, and communication data transmission and reception. For example, it could be a softphone switch component integrated into Zhao Liu's system instance. This component is bound to the operating system's telephone service registry and can capture all VoIP call signaling incoming to Zhao Liu's work number. The aforementioned communication module can also be a low-level hardware driver or system service interface within the client operating system responsible for physical communication link management and data transmission and reception. Examples include the cellular network adapter driver for Zhao Liu's laptop, the PSTN voice gateway driver, or the messaging service API of an instant messaging client.

[0059] It is important to emphasize that the aforementioned communication requests can be inbound signaling or message carriers delivered to the client via a communication module, intending to establish an information exchange link with the current user. Examples include voice calls originating from Zhao Liu's publicly known mobile number, SMTP email messages sent to Zhao Liu's corporate email address, or instant message notifications pushed through the WeChat Work server. The aforementioned communication content can be a valid information payload attached to the communication request, carrying the initiator's intended message. Examples include a voice stream of the caller saying "Hello Mr. Zhang, this is a sales representative from [Company Name]" after the call is connected, text descriptions of project cooperation terms in the email body, or text statements like "Have you confirmed this version of the plan?" in an instant message. The aforementioned communication object can be the addressing identifier or identity representation of the initiator of the communication request in the communication network, such as the caller ID carried in the incoming call signal, the sender's address in the email header field, or the sender's WeChat Work ID attached to an instant message.

[0060] It is also important to emphasize that the aforementioned real-time semantic analysis can be an online computational process in which the control module synchronously calls a large language model to extract intent, recognize sentiment, and locate key information from the streaming text of the communication content during the lifespan of the communication request. For example, it can parse "Hello Mr. / Ms. Zhang, I am a sales representative from [Company Name], and I would like to schedule a presentation with you next Wednesday" into a structured representation of [Initiator Identity: Sales Representative from [Company Name]], [Intent: Meeting Invitation], and [Entity: Time = Next Wednesday]. The results of the aforementioned real-time semantic analysis can be an instant parsing product output by the control module after completing the real-time semantic analysis, containing the core intent and information elements of the communication request. The aforementioned proxy processing method can be an automated action category for the current communication request decision made by the control module based on the real-time semantic analysis results, the identity of the communication object, and historical interaction strategies and user preferences stored in the user feature database. For example, the decision could be "immediately transfer the user," "have the system respond and schedule a time," or "convert to a structured message and store it in the to-do list."

[0061] In the specific implementation, firstly, the communication proxy module deployed on the user client continuously monitors the communication module directly connected to it. When the communication module detects an inbound communication request, it immediately forwards the event notification and signaling metadata of the request to the communication proxy module. Upon receiving the notification, the communication proxy module then retrieves the communication content data stream corresponding to the communication request and the identity identifier of the communication object from the communication module.

[0062] Next, the communication proxy module pushes the acquired communication content and communication object information to the connected control module. Upon receiving the data, the control module immediately initiates a real-time semantic analysis process for the communication content. The control module invokes a preset large language model service to perform streaming parsing of the communication content, continuously generating real-time semantic analysis results.

[0063] Meanwhile, the control module actively retrieves the user feature database specific to the current system instance, using the communication object as the query condition, to obtain historical interaction records associated with that communication object, user-preset priority tags, and past processing strategies. The control module jointly decides based on the real-time semantic analysis results, the communication object's identity information, and the query results returned by the user feature database, ultimately generating a proxy processing method for the current communication request.

[0064] Specifically, the functional module includes a prompting module; The control module is also used to determine the object category corresponding to the communication object based on the communication object and the user feature database; The control module is also used to prompt the user with the communication request through the prompting module when the object category is an important customer.

[0065] It should be noted that the aforementioned notification module can be a system service or application component in the client responsible for outputting visual, auditory, or tactile notification information to the user, such as the operating system notification center, the message pop-up service of WeChat Enterprise Edition, the banner reminder plugin of an email client, or the breathing light strip driver of a laptop. The aforementioned object category can be a hierarchical classification label output by the control module after dividing the relationship between the communication initiator and the user based on the identity identifier of the communication object and the tag rules stored in the user characteristic database. For example, the communication object can be classified as "Important Customer," "General Customer," "Partner," "Unknown Contact," or "Harassment Interception." The aforementioned "Important Customer" can be a specific tag value in the object category, indicating that the communication object has a high-priority business relationship or strong social trust association with the user, requiring immediate attention allocation from the user. For example, contacts in the user characteristic database marked as "Annual Strategic Partner," "Board Member," or "Direct Supervisor."

[0066] In practical use, when executing the proxy processing decision flow for communication requests, the control module actively retrieves the user feature database specific to the current system instance, using the communication object as the query condition. The control module reads metadata such as tag fields, historical interaction frequency, and session value score associated with the communication object from the user feature database. Subsequently, based on the aforementioned metadata and preset classification mapping rules, the control module calculates and determines the object category to which the communication object belongs. The control module temporarily stores the object category determination result in the context data structure of the current communication request. Next, the control module performs conditional judgment on the object category. Upon determining that the object category matches the predefined "Important Customer" tag, the control module immediately generates a prompt instruction. The control module sends the prompt instruction to the prompt module directly connected to the client operating system. Based on the instruction parameters, the prompt module outputs a high-priority notification signal at the user interface layer, notifying the user of a communication request from an important customer through visual pop-ups, voice announcements, or vibration.

[0067] Correspondingly, the functional module also includes a status acquisition module; The control module is also used to obtain the user's current working status through the status acquisition module when the object category is a general customer; The control module is also used to prompt the user with the communication request through the prompting module when the current working state is an idle state; The control module is also used to generate a corresponding response statement based on the real-time semantic analysis result and the user feature database when the current working state is the working state; The communication proxy module is also used to reply to the communication object according to the response statement through the communication module.

[0068] It should be noted that the aforementioned status acquisition module can be a service unit or application interface in the client responsible for collecting, aggregating, and parsing the user's current activity status and system context information. Examples include the Focus Assistant status query interface in Windows, the Do Not Disturb mode detection service in macOS, the busy / idle status reading plugin in Outlook Calendar, or a face detection component that analyzes camera frames to determine if the user is at their workstation. The aforementioned "general customer" can be a specific tag value within an object category. This tag indicates that the communication object has a regular business relationship with the user but does not reach a high-priority intervention level. Examples include contacts in the user feature database marked as "supplier ordinary employee," "cross-departmental collaboration colleague," or "historical single-inquiry customer." The aforementioned current work status can be a contextual identifier returned in real-time by the status acquisition module, reflecting the user's current attention resource allocation and task immersion level, such as "idle state," "busy state," "in a meeting," "Do Not Disturb mode is on," or "locked screen state."

[0069] It should also be noted that the aforementioned idle state can be a specific enumeration value within the current working state. This value indicates that the user is not currently immersed in a continuous work task and has the ability to respond to external communications immediately. For example, the operating system detects that the keyboard and mouse have been idle for more than three minutes and the foreground application is a non-productivity tool. The aforementioned working state can also be a specific enumeration value within the current working state. This value indicates that the user is currently immersed in a continuous work task and does not wish to be interrupted. For example, the operating system detects that the user is running video conferencing software in full-screen mode, the IDE is in a debug breakpoint paused state, or the user has manually switched to "Do Not Disturb" mode. The aforementioned response statement can be a dynamically generated, anthropomorphic natural language reply text based on the communication intent extracted by the control module based on real-time semantic analysis results and user preferences, response templates, and historical speech styles stored in the user feature database. For example, a response to a meeting invitation request might be: "Hello, thank you for your invitation. Manager Zhao is currently handling other matters and will review your meeting proposal later and confirm the time as soon as possible."

[0070] In its implementation, the control module, during the communication request proxy processing flow, has already determined through the aforementioned steps that the current communication object belongs to the "general customer" category. In response to this category determination, the control module immediately generates a status query command and sends it to the status acquisition module connected to it on the client. Upon receiving the command, the status acquisition module immediately invokes the operating system's underlying user activity detection interface, schedule status reading service, and foreground process analysis tools to collect the user's multi-dimensional activity context at the current moment. The status acquisition module aggregates the collected raw data into standardized work status enumeration values ​​and returns these values ​​to the control module. Upon receiving the current work status value, the control module performs conditional branch judgments on that value. If the control module determines that the current work status is "idle," it decides to adopt an immediate prompting strategy. The control module generates a prompt command and sends it to the prompting module, which, based on the command parameters, outputs the communication request notification information to the user through visual or auditory channels. In response to determining the current working state as "working state," the control module adopts an automatic response strategy. This involves using the intent category, key entities, and sentiment trends from real-time semantic analysis as the generation context, and simultaneously retrieving the user feature database using the communication object identifier as the query condition to obtain historical response templates, user-preferred response styles, and taboo phrases matching this scenario. Based on the above input, the control module invokes a large language model service to dynamically generate a response statement that conforms to the current context. Subsequently, the control module encapsulates this response statement into a reply instruction and sends it to the communication proxy module. The communication proxy module then transmits the response statement to the communication object through the communication module, completing an immediate automatic response to the communication request.

[0071] In addition, the functional modules also include a task list module; The control module is further configured to convert the real-time semantic analysis results into structured to-do items when the object is an unknown customer, and store the to-do items in the user feature database and the task list module.

[0072] It should be noted that the aforementioned task list module can be a service unit or application component in the client responsible for creating, storing, tracking the status of, and reminding users of due dates for to-do items. Examples include Microsoft To-Do's local data service interface, TickTick's task dashboard engine, Outlook's task panel creation and synchronization plugin, or the to-do item generator in WeChat Work's calendar module. The aforementioned unknown customer can be a specific tag value within an object category. This tag represents the identity of the communication object, indicating that there are no historical interaction records, associated tags, or preset strategies in the user characteristic database of the current system instance. Examples include an unfamiliar phone number making its first call, an unfamiliar email address sending its first email, or an instant messaging account that has never appeared in the enterprise address book. The aforementioned real-time semantic analysis results can be structured data products output by the control module after calling a large language model to perform streaming parsing of the communication content. These data products contain user intent categories, key information entities, sentiment tendencies, and the urgency of the request. The structured to-do items mentioned above can be task record entries generated by the control module after extracting fields, standardizing formats, and semantically reorganizing unstructured text information from real-time semantic analysis results according to a predefined data model. These entries should possess attributes such as sortability, filterability, assignable responsibility, and deadline. For example, they could be JSON objects containing six fields: "Source of Task," "Creation Time," "Description of Task," "Deadline," "Priority," and "Associated Contact Person." These to-do items can also be structured and persistently stored task record instances, awaiting user review, confirmation, execution, or closure at a future point in time. The user feature database mentioned above can be a data storage component within the system instance dedicated to vectorized storage of personalized user feature information. This database also serves as an archiver for associating to-do item metadata with the original communication content.

[0073] In practical use, when executing the communication request proxy processing flow, the control module has already determined the category of the current communication object to be "unknown customer" through the aforementioned steps. Subsequently, the control module calls the built-in task structuring engine and, based on the predefined to-do item data model, sequentially extracts the item source field, item description text, request time constraint, contact person identification, and urgency score from the real-time semantic analysis results. Next, the control module reassembles and serializes the extracted fields according to a standardized format, generating a complete structured to-do item record with a unique identifier. The control module adds a current timestamp and a default status value of "pending" to this to-do item record. Finally, the control module writes the generated structured to-do item record into the user feature database dedicated to the current system instance, establishing an association index with the original communication content, communication object identification, and timestamp of this communication request. Simultaneously, the control module pushes this to-do item record to the task list module connected to it in the client. After receiving the record, the task list module renders a new task entry in the user's task dashboard interface and sets an due date reminder, priority color label, and quick operation button according to preset rules.

[0074] In another example, the task list module stores tasks including: to-do tasks and instruction tasks; the to-do tasks are information record tasks that users can view, confirm, execute, or close; the instruction tasks are executable operation sequences that can be directly parsed by the control module and executed by the functional module. The control module is further configured to read the instruction-type task from the task list module, parse the instruction-type task into a corresponding function instruction, and send the function instruction to the corresponding function module so that the function module performs the corresponding operation according to the function instruction.

[0075] It should be noted that the aforementioned instruction-type tasks can be automated task objects that encapsulate specific device control logic, such as a timed switch instruction like "turn off the office lights at 9:00 AM every Monday," a conditional trigger instruction like "turn on the air conditioner when the indoor temperature is above 28°C," or a communication proxy instruction like "automatically reply with a preset text when a communication request with a specific keyword is received." The aforementioned executable operation sequence can be a structured instruction set generated after the instruction-type task is parsed, consisting of multiple atomic operations combined in sequence. For example, the "turn off the lights" instruction corresponds to the single operation of "sending a UDP packet to the smart home gateway with the packet content {device: 'light', action: 'off'}", while the "meeting preparation" instruction can correspond to the sequence of operations of "closing the curtains, turning on the projector, dimming the lights, and sending a meeting reminder message."

[0076] For example, when a system instance receives a user's command via the multimodal input module to "set up an automatic shutdown command for my computer every Friday at 6 PM," the control module calls the large language model for semantic analysis, parsing out the following parameters: [Intent: Create a scheduled task], [Task type: Command], [Execution time: Every Friday at 6 PM], [Target: Computer], and [Action: Shut down]. The control module structures these parameters into a command-type task record and stores it in the task list module. When the system time reaches 6 PM every Friday, the control module reads the command-type task from the task list module, parses it into the corresponding operating system shutdown command, and executes the shutdown operation through the system service interface in the functional module.

[0077] Reference Figure 2 , Figure 2 This is a structural block diagram of the second embodiment of the information analysis and interaction system of this application. Based on the first embodiment described above, the second embodiment of this application is proposed.

[0078] like Figure 2 As shown, in this embodiment, in order to facilitate interaction between different users in the system, the system instance includes at least a first system instance and a second system instance. The first system instance is deployed on the first client of the first user, and the second system instance is deployed on the second client of the second user. The first system instance includes a first user feature database, a first multimodal input module, a first control module, and a first communication proxy module; the second system instance includes a second user feature database, a second multimodal input module, a second control module, and a second communication proxy module; the first communication proxy module and the second communication proxy module are connected through a communication interface. The first multimodal input module is further configured to determine the second user and the interaction content corresponding to the interaction signal when it receives the interaction signal sent by the first user; The first communication proxy module is further configured to send the interactive content to the second communication proxy module through the communication interface; The second communication proxy module is further configured to obtain first user information corresponding to the first user upon receiving the interaction content; The second control module is further configured to obtain an interaction strategy for the interaction content based on the first user information, the interaction content, and the second user feature database, and drive the corresponding functional module to interact based on the interaction content according to the interaction strategy.

[0079] It is understandable that the aforementioned first user can be a natural person bound to the first system instance and having full control over that instance, such as product manager Zhao Liu. The aforementioned second user can be a natural person bound to the second system instance and having full control over that instance, such as R&D engineer Li Si. The aforementioned first system instance can be an independent, personalized system running entity created and maintained by the instance management module for the first user and deployed on the first user's client, such as the dedicated intelligent agent client instance running on product manager Zhao Liu's workstation. The aforementioned second system instance can be an independent, personalized system running entity created and maintained by the instance management module for the second user and deployed on the second user's client, such as the dedicated intelligent agent client instance running on R&D engineer Li Si's workstation.

[0080] The aforementioned first user feature database can be a data storage component in the first system instance, dedicated to the first user and used to vectorize and store their personalized feature information. The aforementioned second user feature database can be a data storage component in the second system instance, dedicated to the second user and used to vectorize and store their personalized feature information. The aforementioned first multimodal input module can be a front-end processing component in the first system instance responsible for receiving and transforming the original input information of the first user. The aforementioned second multimodal input module can be a front-end processing component in the second system instance responsible for receiving and transforming the original input information of the second user. The aforementioned first control module can be a core decision-making and scheduling component of the first system instance, maintaining communication connections with the large language model service, the first user feature database, and the first client functional module. The aforementioned second control module can be a core decision-making and scheduling component of the second system instance, maintaining communication connections with the large language model service, the second user feature database, and the second client functional module. The aforementioned first communication proxy module can be an edge processing unit in the first system instance dedicated to communication channel access, communication request sending and receiving, and cross-instance data exchange. The aforementioned second communication proxy module can be an edge processing unit in the second system instance dedicated to communication channel access, communication request sending and receiving, and cross-instance data exchange.

[0081] The aforementioned communication interface can be a communication proxy module connecting different system instances, a logical link or physical channel supporting cross-client data exchange and signaling transmission, such as a message middleware service based on the MQTT protocol, a gRPC bidirectional streaming channel in an enterprise intranet, or an end-to-end encrypted tunnel established via WebSocket. The aforementioned interaction signal can be a carrier of instructions actively issued by the first user through the first multimodal input module, intending to establish a task collaboration relationship with the second user, such as Zhao Liu speaking into the microphone, "Please ask Li Si if the work order for exporting report performance optimization can be submitted for testing today." In the context of the interaction signal, the aforementioned second user specifically refers to the identity identifier of the target object that the first user intends to interact with. The aforementioned interaction content can be an information payload attached to the interaction signal, carrying the first user's intention to transmit to the second user, such as the text string "Can the work order for exporting report performance optimization be submitted for testing today?" (The first part of the text refers to the second user in the context of the interaction signal.) The aforementioned first user information may be a set of attributes extracted by the second communication agent module from the communication protocol stack or message envelope after receiving the interaction content, used to identify the identity of the first user and the source of the first system instance, such as the employee number of the first user, the device fingerprint of the first system instance, or the routing address of the first user in the cross-instance communication network.

[0082] In its implementation, the first multimodal input module of the first system instance continuously monitors the input channel of the first user. When the first user sends an interaction signal intending to collaborate with the second user via voice, text, or image modality, the first multimodal input module immediately initiates a conversion process, parsing the original interaction signal into structured interaction content and identifying the target object of this interaction, i.e., the identity of the second user, from the signal context or user specification. After parsing, the first multimodal input module pushes the second user's identity and interaction content to the first control module. Upon receiving the data, the first control module generates a cross-instance communication request and forwards the request and interaction content to the first communication proxy module. Upon receiving the forwarded data, the first communication proxy module parses the routing address of the second system instance in the network communication layer based on the second user's identity. Subsequently, the first communication proxy module encapsulates the interaction content into a cross-instance message packet through a pre-established communication interface and sends it to the second communication proxy module of the second system instance. The second communication proxy module continuously monitors the communication interface and immediately performs a parsing operation upon receiving a message packet from the first communication proxy module. The second communication proxy module extracts the identity identifier of the first user and the source information of the first system instance from the message header or message envelope, and aggregates them to generate the first user information object.

[0083] Next, the second communication proxy module pushes the received interaction content and the extracted first user information to the second control module. Upon receiving this data, the second control module actively retrieves the second user feature database, which is exclusive to the second system instance, using the first user information and key entities in the interaction content as composite query conditions. This retrieves historical interaction records associated with the first user, the second user's preset authorization policy for that first user, and processing preferences from past collaboration scenarios. The second control module then performs joint reasoning based on the real-time semantic analysis results of the interaction content, the first user information, and the query results returned by the second user feature database to generate an interaction strategy for this cross-instance interaction request. Following the decision content of the interaction strategy, the second control module drives the corresponding functional modules in the second client to execute specific interaction operations.

[0084] Furthermore, the second control module is also used to query the second user feature database to obtain a preset authorization policy associated with the first user information, and to determine whether the interactive content belongs to the preset authorization policy; The second control module is further configured to, when the interactive content belongs to the preset authorization policy, extract task parameters from the interactive content, call the functional module of the second client to perform the corresponding operation according to the task parameters, and generate a processing result; The second communication proxy module is further configured to return the processing result to the first communication proxy module through the communication interface.

[0085] It should be noted that the aforementioned preset authorization strategy may be a set of rules pre-stored in the second user characteristic database, associated with a specific first user identity or the organization to which the first user belongs, used to define what interactive content the second control module can directly process and its processing boundaries without real-time intervention from the second user. For example, Li Si might configure policy entries such as "work order progress query requests initiated by Zhao Liu can be automatically responded to," "meeting time inquiry requests initiated by Zhao Liu only return busy / idle status and cannot be automatically scheduled," or "any data query request initiated by Zhao Liu's product department must be accompanied by an operation log for future reference." The aforementioned interactive content may be an information payload generated by the first multimodal input module after parsing the first user's interactive signals, and transmitted to the second communication proxy module via the communication interface. This payload contains natural language instructions or structured task descriptions that the first user intends to convey to the second user. The aforementioned task parameters can be a set of key fields extracted in an orderly manner from the semantic parsing results of the interaction content by the second control module after determining that the interaction content falls within the scope of the preset authorization policy. These fields are used to drive the second client functional modules to perform specific operations. For example, the work order identifier "PERF-2026-0210" and the query intent "test submission time" extracted from "Can the PERF-2026-0210 work order be submitted for testing today?" can be extracted. The aforementioned second client can be the terminal device and its operating system environment that hosts the second system instance, such as the personal workstation used by Li Si daily. The aforementioned functional modules can be applications or system service units with specific business capabilities that are registered or dynamically invoked in the second client's operating system, such as the local client interface of the Jira work order system, the busy / idle status query service of Outlook calendar, or the message sending plugin of WeChat Work. The aforementioned corresponding operations can be an atomic sequence of actions executed by the functional module after receiving the task parameters, which produces specific business effects locally on the second client. For example, the Jira client queries the current status field of a specified work order based on the work order identifier and returns the result. The above processing result can be a structured response object that encapsulates the task execution status and returned data after the functional module performs the corresponding operation. For example, it can be a JSON data packet containing the work order status "under testing", the estimated test submission time "before 16:00 today", the last updater "Li Si" and the update timestamp.

[0086] In its implementation, the second control module, upon receiving the first user information and interaction content pushed by the second communication proxy module, immediately uses the first user information as the primary key to actively retrieve all preset authorization policy entries associated with the first user's identity from the second user feature database dedicated to the second system instance. After loading the policy entries, the second control module performs real-time semantic analysis on the interaction content, parsing out the intent category, operation object type, and specific operation instructions of this request. Next, the second control module matches the semantic analysis results with each loaded preset authorization policy set to verify whether the request intent and operation scope expressed by the interaction content completely fall within the processing boundaries allowed by one or more authorization policies. In response to the second control module's determination that the interaction content fully falls within the authorization scope of the preset authorization policies, the second control module immediately terminates the process branch of initiating an interactive request to the second user and instead starts the automated processing flow. The second control module systematically extracts the key fields required to execute this request from the semantic parsing results of the interaction content and assembles them into a standardized task parameter structure.

[0087] Subsequently, the second control module, based on the operation type identifier in the task parameters, dynamically parses and locates the specific functional module with corresponding processing capabilities in the functional module service directory registered on the second client. The second control module encapsulates the task parameters into an instruction format recognizable by the functional module via inter-process communication or service call interfaces and synchronously sends it to that functional module. Upon receiving the instruction, the functional module strictly follows the operation instructions and field values ​​carried in the task parameters, executing the corresponding business operation locally on the second client. After completing the operation, the functional module encapsulates the status codes, return data, and exception information generated during execution into a structured response object, i.e., the processing result, and sends this result back to the second control module. Upon receiving the processing result, the second control module performs format standardization and necessary content summary extraction. Then, the second control module forwards the processing result to the second communication proxy module, attaching the target address of this response, i.e., the routing identifier of the first communication proxy module of the first system instance. Upon receiving the processing result and the target address, the second communication proxy module immediately connects through the existing communication interface, encapsulates the processing result into a cross-instance message packet, and delivers it to the message queue or endpoint channel monitored by the first communication proxy module. The first communication proxy module completes message parsing at the receiving end and pushes the processing result to the first control module, so that the first control module can provide feedback on the processing result to the first user through the prompt module, thus completing the information interaction.

[0088] Accordingly, the second control module is also used to query the second user feature database to see if there is historical interaction data corresponding to the first user information when the interaction content does not belong to the preset authorization policy; The second control module is further configured to perform semantic similarity matching based on the historical interaction data and the interaction content when the historical interaction data corresponding to the first user information exists in the second user feature database; The second control module is further configured to generate a response scheme based on the semantic similarity matching result, call the functional module of the second client to perform the corresponding operation according to the response scheme, and generate the processing result; The second communication proxy module is further configured to return the processing result to the first communication proxy module through the inter-system communication interface.

[0089] It should be noted that the aforementioned historical interaction data can be past cross-instance communication records and their processing contexts stored in the second user feature database and associated with the first user's identity. For example, the second user feature database of Li Si may contain the original text of all collaborative requests initiated by Zhao Liu in the past three months, the interaction strategy decided by the second control module at that time, the final feedback content of the second user to the request, and the vectorized semantic representation persisted after the interaction. The aforementioned semantic similarity matching can be the process by which the second control module calculates the distance between the vectorized representation of the current interaction content and the vectorized representation of each record in the historical interaction data set, in order to recall the historical interaction sample that is semantically closest. For example, the cosine similarity calculation is performed between "Has the PERF-2026-0210 work order test passed?" and the historical request "Can the PERF-2026-0209 work order be submitted for testing today?" to determine that the intent of both requests is to query the work order status and the object is a performance optimization type work order.

[0090] It is important to emphasize that the semantic similarity matching result mentioned above can be a structured output from the second control module after completing the similarity calculation, containing the identifier of the historical interaction record most similar to the current interaction content and its similarity score. The response scheme mentioned above can be a dynamic response strategy and corresponding reply script generated by the second control module based on the historical response content, processing methods, and user feedback tendencies extracted from the historical interaction records recalled by the semantic similarity matching result, combined with the real-time semantic parsing result of the current interaction content. For example, if the historical record "Zhao Liu inquired about the progress of work order A, and Li Si replied at the time 'It is expected to be tested before the end of today' and was accepted by Zhao Liu'" is recalled, then the response scheme generated for the current work order B progress inquiry is "Use the historical similar reply template to generate 'It is expected to be tested before the end of today'".

[0091] In its implementation, the second control module first completes the matching and determination of the preset authorization policies, and the determination result is that the current interaction content does not fall within the authorization scope of any preset authorization policy. In response to this negative determination result, the second control module immediately initiates the alternative decision-making process. Using the first user information as the query primary key, the second control module actively retrieves all historical interaction data records associated with the first user's identity identifier from the second user feature database, which is exclusive to the second system instance. This retrieval scope includes the original text of past cross-instance communication requests, the request time, the second control module's processing decision at that time, the second user's actual response content, and the vectorized semantic vector persisted after the interaction.

[0092] Subsequently, the second control module preprocesses the retrieved historical interaction data record set, extracting the request text vector and response text vector stored in each record. Simultaneously, the second control module invokes a vectorization service to convert the current interaction content into a semantic vector of the same dimension in real time. Next, using the semantic vector of the current interaction content as the query center, the second control module performs a nearest neighbor search on the set of request text vectors in the historical interaction data record set, calculating the semantic similarity score between each historical record and the current interaction content. The second control module sorts the historical records from high to low based on the similarity score and selects several optimal matching records with scores exceeding a preset threshold, generating semantic similarity matching results. Based on the historical response content, historical processing methods, and the second user's subsequent feedback tendency for this interaction corresponding to the highest-scoring record in the semantic similarity matching results, combined with the key entity differences in the real-time semantic parsing results of the current interaction content, the second control module adaptively adjusts and reconstructs the response scheme for this request. The response scheme includes the type of operation to be performed, the identifier of the functional module to be invoked, the task parameters to be passed to the functional module, and the natural language response text to be replied to the first user.

[0093] After the response scheme is generated, the second control module locates the corresponding functional module in the service directory of the functional modules registered by the second client, based on the operation type identifier in the scheme. The second control module encapsulates the task parameters in the response scheme into an instruction format via the service call interface and sends it to the functional module. Upon receiving the instruction, the functional module executes the corresponding operation according to the task parameters and encapsulates the status codes and returned data generated during execution into a structured processing result, which is then sent back to the second control module. Upon receiving the processing result, the second control module merges and encapsulates the data returned by the functional module with the pre-generated natural language response text in the response scheme to form a complete processing result object. Subsequently, the second control module forwards the processing result to the second communication proxy module, attaching the routing identifier of the first communication proxy module of the first system instance as the response target address. The second communication proxy module connects through the existing inter-system communication interface, encapsulates the processing result into a cross-instance message packet, and delivers it to the message channel monitored by the first communication proxy module.

[0094] Meanwhile, the second control module is also used to store the first user information and the interaction content as a structured item to be interacted with in the second user feature database when the historical interaction data corresponding to the first user information does not exist in the second user feature database.

[0095] It is understandable that the aforementioned items to be interacted with can be task record entries generated by the second control module after extracting fields, standardizing formats, and semantically reorganizing key elements from the first user information and interaction content according to a predefined data model. These entries are awaiting the second user's review, confirmation, response, or handling at a future point in time. For example, they could be structured JSON objects containing six fields: "item source, initiator identity, request time, item description, urgency, and processing status." These structured items to be interacted with can be instances of items to be interacted with that have already undergone field extraction and format standardization.

[0096] In its implementation, the second control module first completes the matching and determination of the preset authorization policy and the existence retrieval of historical interaction data. The determination result is that the current interaction content does not belong to any preset authorization policy, and at the same time, there are no historical interaction data records associated with the first user's identity in the second user feature database. In response to this double negative determination result, namely "no authorization policy" and "no historical interaction", the second control module immediately terminates the automated processing attempt for the current interaction request and initiates an asynchronous processing procedure.

[0097] Subsequently, the second control module obtains the parsed first user information object and the structured data of the interaction content that has undergone semantic analysis from the current interaction context. The second control module calls the built-in event structuring engine, and according to the predefined data model of the event to be interacted with, extracts the initiator's identity identifier, initiator's name, and initiator's system instance routing address from the first user information in sequence, and extracts the event description text, request intent tags, key entity fields, and the urgency score automatically assessed by the system from the semantic analysis results of the interaction content.

[0098] Next, the second control module reassembles and serializes the extracted fields according to a standardized format, attaching a globally unique identifier, a current timestamp, and a default status value of "pending processing" to the current item, generating a complete structured record of items to be interacted with. The second control module performs a persistence operation, writing the structured record of items to be interacted with into the second user feature database dedicated to the second system instance. During the writing process, the second user feature database automatically associates the record with the first user information index, the full text of the original interaction content, and the receiving timestamp, and stores its semantic vectorized representation in the vector index library for semantic matching retrieval of similar requests in the future.

[0099] After storage is completed, the second control module may optionally send a non-intrusive reminder of the new task to the second user through the functional module of the second client, or render the task to be interacted with in the second user's task dashboard interface to generate a new task entry, waiting for the second user to review and handle it later.

[0100] In this embodiment, the system instance further includes a public information management and control module.

[0101] The public information management and control module is used to manage and maintain the public information of all system instances, including each user's external business information and public transaction information.

[0102] The public information management and control module is also used to formulate and unify the public information management protocol. The public information protocol is used to ensure that the exchange of public information and public transaction behavior between system instances comply with unified standards and legal norms.

[0103] The public information management and control module has a communication mechanism with the privacy module and the compliance module in each of the system instances, which is used to synchronize privacy policies and compliance protocols.

[0104] It should be noted that the aforementioned public information management and control module can be a centralized or distributed governance component independent of each user system instance. For example, it could be a public service node deployed on an enterprise intranet or consortium blockchain, maintaining the external product catalogs, public quotations, transaction term templates, and industry compliance whitelists of all member companies. The aforementioned public information protocol can be a set of data exchange formats, interface specifications, and business rules published by the public information management and control module and that all access system instances must adhere to. For example, it could stipulate that all inquiry requests must use JSON format, include standard fields {inquiry number, product code, quantity, expected delivery date}, and be accompanied by a digital signature compliant with the Electronic Signature Law.

[0105] The communication mechanism between the aforementioned public information management and control module and the privacy and compliance modules can be a periodic policy synchronization channel or an event-driven message push link. For example, after the public information management and control module detects that the revised version of the Data Security Law has come into effect, it automatically packages the new data classification and grading standards into a compliance policy update package and pushes it to the compliance modules of each system instance through an encrypted channel. The compliance modules then update their local compliance verification rules accordingly. Alternatively, when the privacy module of a system instance of an enterprise user publishes a new data sharing agreement, the privacy module synchronizes the agreement summary to the public information management and control module for filing. The public information management and control module then broadcasts the agreement update to other members in the alliance to ensure that each member can recognize and comply with the agreement in subsequent interactions.

[0106] Reference Figure 3 , Figure 3 This is a flowchart illustrating the first embodiment of the information analysis and interaction method of this application. Based on the first and second embodiments of the information analysis and interaction system described above, the first embodiment of the information analysis and interaction method of this application is proposed. This method is applied to the aforementioned information analysis and interaction system, such as... Figure 3 As shown, the method includes: Step S10: Receive raw input information from the user and obtain structured input data based on the raw input information; Step S20: Perform semantic analysis on the structured input data using a preset large language model, and obtain user needs based on the semantic analysis results and the user feature database; Step S30: Based on the user's needs, determine the corresponding function instructions and the function modules in the client, and send the function instructions to the function modules so that the function modules perform the corresponding operations according to the function instructions.

[0107] In its implementation, the system continuously monitors the user's input channels via a multimodal input interface. Upon detecting raw information input by the user via voice, text, or image modality, the system immediately invokes its built-in conversion engine to parse the raw information into standardized structured input data. After data conversion, the system pushes the structured input data to a locally deployed lightweight large language model inference service. The system performs deep semantic analysis on the structured input data using this large language model, identifying the intent category and key information entities in the user's command and generating semantic analysis results. Using the key entities in the semantic analysis results as search criteria, the system actively queries its proprietary user feature database to obtain the user's historical preferences, business constraints, and personalized configuration parameters associated with the command context. The system performs joint inference and information fusion between the semantic analysis results and the database query results to generate a user requirement object that accurately characterizes the user's true intent and execution boundaries. The system parses the user requirement object, decomposes it into several atomic executable steps, and matches each step with a specific functional module registered in the client operating system that possesses corresponding processing capabilities. The system generates structured functional instructions for each functional module. These instructions contain all the operational parameters and calling context required for the module to execute its task. The system distributes these instructions to the corresponding functional modules through the operating system's inter-process communication mechanism or system service scheduling interface. The system then drives each functional module to execute specific business operations locally on the client side according to the instruction parameters, completing the entire transformation chain from user input to terminal task execution.

[0108] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. An information analysis and interaction system, characterized in that, The system includes an instance management module and at least one client, with different clients corresponding to different users. The instance management module is used to configure an independent system instance for each user based on the user's historical data. The system instance is deployed on the client of the corresponding user, and the system instance includes: a user feature database, a multimodal input module, and a control module; The multimodal input module is connected to the control module, and the control module is connected to the functional modules in the client. A multimodal input module is used to receive raw input information from the user, obtain structured input data based on the raw input information, and input the structured input data to the control module; The control module is used to perform semantic analysis based on the structured input data using a preset large language model, and to obtain user needs based on the semantic analysis results and the user feature database. The control module is further configured to determine the corresponding function instructions and the function modules in the client based on the user requirements, and send the function instructions to the function modules so that the function modules perform the corresponding operations according to the function instructions.

2. The system as described in claim 1, characterized in that, The system instance further includes: a communication proxy module, which is connected to the control module; the functional module includes a communication module, which is connected to the communication proxy module. The communication proxy module is used to obtain the communication content and communication object corresponding to the communication request when the communication module detects that it has received a communication request. The control module is also used to perform real-time semantic analysis on the communication content, and based on the real-time semantic analysis results, the communication object, and the user feature database, to obtain a proxy processing method for the communication request.

3. The system as described in claim 2, characterized in that, The functional modules include a prompting module; The control module is also used to determine the object category corresponding to the communication object based on the communication object and the user feature database; The control module is also used to prompt the user with the communication request through the prompting module when the object category is an important customer.

4. The system as described in claim 3, characterized in that, The functional module also includes a status acquisition module; The control module is also used to obtain the user's current working status through the status acquisition module when the object category is a general customer; The control module is also used to prompt the user with the communication request through the prompting module when the current working state is an idle state; The control module is also used to generate a corresponding response statement based on the real-time semantic analysis result and the user feature database when the current working state is the working state; The communication proxy module is also used to reply to the communication object according to the response statement through the communication module.

5. The system as described in claim 3, characterized in that, The functional modules also include a task list module; The control module is further configured to convert the real-time semantic analysis results into structured to-do items when the object is an unknown customer, and store the to-do items in the user feature database and the task list module.

6. The system as described in claim 2, characterized in that, The system instance includes at least a first system instance and a second system instance, wherein the first system instance is deployed on a first client of a first user, and the second system instance is deployed on a second client of a second user; The first system instance includes a first user feature database, a first multimodal input module, a first control module, and a first communication proxy module; the second system instance includes a second user feature database, a second multimodal input module, a second control module, and a second communication proxy module; the first communication proxy module and the second communication proxy module are connected through a communication interface. The first multimodal input module is further configured to determine the second user and the interaction content corresponding to the interaction signal when it receives the interaction signal sent by the first user; The first communication proxy module is further configured to send the interactive content to the second communication proxy module through the communication interface; The second communication proxy module is further configured to obtain first user information corresponding to the first user upon receiving the interaction content; The second control module is further configured to obtain an interaction strategy for the interaction content based on the first user information, the interaction content, and the second user feature database, and drive the corresponding functional module to interact based on the interaction content according to the interaction strategy.

7. The system as described in claim 6, characterized in that, The second control module is further configured to query the second user feature database to obtain a preset authorization policy associated with the first user information, and determine whether the interactive content belongs to the preset authorization policy; The second control module is further configured to, when the interactive content belongs to the preset authorization policy, extract task parameters from the interactive content, call the functional module of the second client to perform the corresponding operation according to the task parameters, and generate a processing result; The second communication proxy module is further configured to return the processing result to the first communication proxy module through the inter-system communication interface.

8. The system as described in claim 7, characterized in that, The second control module is further configured to query the second user feature database to see if there is historical interaction data corresponding to the first user information when the interaction content does not belong to the preset authorization policy; The second control module is further configured to perform semantic similarity matching based on the historical interaction data and the interaction content when the historical interaction data corresponding to the first user information exists in the second user feature database; The second control module is further configured to generate a response scheme based on the semantic similarity matching result, call the functional module of the second client to perform the corresponding operation according to the response scheme, and generate the processing result; The second communication proxy module is further configured to return the processing result to the first communication proxy module through the inter-system communication interface.

9. The system as described in claim 8, characterized in that, The second control module is further configured to, when the historical interaction data corresponding to the first user information does not exist in the second user feature database, convert the first user information and the interaction content into a structured item to be interacted, and store the item to be interacted in the second user feature database.

10. An information analysis and interaction method, characterized in that, The method is applied to the system as described in any one of claims 1 to 9, the method comprising: Receive raw input information from the user and obtain structured input data based on the raw input information; Semantic analysis is performed on the structured input data using a pre-defined large language model, and user needs are obtained based on the semantic analysis results and the user feature database. Based on the user's needs, the corresponding function instructions and the function modules in the client are determined, and the function instructions are sent to the function modules so that the function modules can perform the corresponding operations according to the function instructions.

Images