An application interaction method, device and storage medium

Abstract

The embodiment of the application provides an application interaction method and device and a storage medium, and relates to the technical field of computers. The method sets an AI interaction module in a target application, the AI interaction module perceives a user instruction on a first page of the target application, interacts with an AI large model of a server based on the user instruction, obtains intention information by implementing intention recognition through the AI large model, calls a page management large model located at the server, and provides page jump judgment and page information of a second page by the page management large model. The AI interaction module can execute page jump based on the page information of the second page, so that the user can directly reach a required page through natural language in a traditional application, the operation steps are reduced, the operation efficiency is improved, the difficulty of finding functions and pages is reduced, the page jump function is flexibly realized with the aid of AI technology, the system can quickly adapt to new page and function changes, and the flexibility and scalability of the system are enhanced.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to an application interaction method, device and storage medium. Background Technology

[0002] In traditional applications (APPs), user interaction is not convenient enough. For example, when a user wants to navigate from the current page to a target page, they need to go through a cumbersome process (such as clicking through menu items or browsing multi-level category lists). With the development of Artificial Intelligence (AI) technology, applications developed based on AI models (referred to as AI applications) are gradually emerging. These applications allow users to invoke AI models, which then provide services such as chat, question answering, and automated operations. However, traditional APPs typically undergo long-term iterations, resulting in numerous and complex functional modules, making it difficult to integrate AI functions into traditional APPs. Traditional APPs face the technical challenge of how to flexibly implement page navigation functionality using AI technology. Summary of the Invention

[0003] This application provides an application interaction method, device, and storage medium that can integrate AI functions with traditional apps, flexibly implement page navigation functions with the help of AI technology, and improve user experience.

[0004] In a first aspect, embodiments of this application provide an application interaction method applied to an AI interaction module in a target application. The AI ​​interaction module is used to perceive user commands in an active state and interact with a server-side AI big model based on the user commands, so as to complete page jumps adapted to user commands with the help of the AI ​​big model. The method includes: If a user command is detected on the first page of the target application, the user command is sent to the intent recognition model located on the server so that the intent recognition model can perform intent recognition on the user command and obtain at least one intent information. Based on the target intent information in at least one of the intent information, the page management big model located on the server is invoked so that the page management big model determines whether page redirection is needed based on the target intent information and obtains the page information of the second page to be redirected when page redirection is needed. The second page is another page of the target application. The receiving page management model returns the page information of the second page when it determines that a page jump is required based on the target intent information; Based on the page information of the second page, the user jumps from the first page to the second page. There is either no jump relationship defined in the program code of the target application between the first page and the second page, or there is a jump relationship defined in the program code of the target application.

[0005] Secondly, embodiments of this application also provide an electronic device, including: a memory and a processor; wherein, the memory stores executable code, and when the executable code is executed by the processor, the processor can at least implement the method described in the first aspect.

[0006] Thirdly, embodiments of the present invention provide a non-transitory machine-readable storage medium storing executable code, which, when executed by a processor of an electronic device, enables the processor to at least implement the method described in the first aspect.

[0007] Fourthly, embodiments of the present invention provide a computer program product, the computer program product including a computer program or instructions, which, when executed by a processor, cause the processor to implement the method described in the first aspect above.

[0008] The method provided in this application embodiment sets up an AI interaction module in the target application. The AI ​​interaction module perceives user commands on the first page of the target application, interacts with the server-side AI big model based on the user commands, and obtains at least one intent information through intent recognition by the AI ​​big model. Then, based on the target intent information in the at least one intent information, it calls the page management big model located on the server. The page management big model provides page jump judgment and page information of the second page. The AI ​​interaction module can perform page jump based on the page information of the second page, thereby enabling users to directly reach the required page through natural language in traditional applications, reducing operation steps, improving operation efficiency, and reducing the difficulty for users to find functions and pages.

[0009] Furthermore, the jump path between the first page and the second page may or may not exist in the target application's program code. Thus, even if the target application code does not define the jump logic between pages, the method provided in this application embodiment can flexibly implement page jump functionality using AI technology. Furthermore, since this application does not require the target application code to contain page jump logic, there is no need to modify the target application's program code when pages are added or modified. This supports real-time page addition and updates, enabling the system to quickly adapt to new page and functional changes, enhancing the system's flexibility and scalability. Attached Figure Description

[0010] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1This is a schematic diagram of an application interaction scenario provided in an embodiment of this application; Figure 2 This is a schematic diagram of the structure of an application interaction system provided in an embodiment of this application; Figure 3 A flowchart illustrating an application interaction method provided in an embodiment of this application; Figure 4 A flowchart illustrating another application interaction method provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0011] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0012] It should be noted that, in the cases involving user information in the embodiments of this application, the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the embodiments of this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse. In addition, the various models involved in this application (including but not limited to language models or large models) comply with relevant laws and standards.

[0013] Traditional applications (APPs) refer to applications developed based on traditional programming logic and algorithms before the widespread application of artificial intelligence technology. These applications do not integrate AI models or corresponding API interfaces. In traditional applications, designs such as nested multi-level menus or hidden entry points make user interaction with the APP inconvenient. For example, when a user wants to jump from the current page to a target page, they need to go through a cumbersome process. For instance, such as... Figure 1 As shown, when using the 58.com app, if a user wants to access the whole-unit rental page from the main page, they need to first go to the main page, click "Rent a Room," then click "Rent a Whole Room," and finally click "Entire Unit." If the user wants to go to other pages, they have to return to the main page and repeat the process multiple times, making it rather cumbersome.

[0014] With the development of artificial intelligence (AI) technology, applications developed based on AI models (referred to as AI applications) are gradually emerging. These applications allow users to invoke AI models, which then provide services such as chat, question answering, and automated operation.

[0015] However, traditional apps, after long-term iterations, generally suffer from numerous functional modules, complex code, and mixed development using multiple languages. This results in high coupling between modules and complex API call chains. Integrating AI functions into traditional apps involves not only model training and deployment but also complex front-end and back-end logic restructuring and data integration, requiring significant time and financial investment to resolve the issues involved. Furthermore, inconsistent data formats and naming conventions across functional modules create data silos that are difficult to integrate and share. Accumulated data may also contain missing, erroneous, or duplicate data, necessitating substantial resources for data cleaning and preprocessing during model training. Moreover, the front-end and back-end logic restructuring requires redesigning the interaction logic for each page, establishing a mechanism for interfacing with AI functions, and ensuring consistent interactive experience and data processing standards across different pages. This full-stack transformation requires significant development resources for page adaptation and AI protocol development, as well as building complex semantic understanding models to accurately interpret user input. These issues make integrating AI functions into traditional apps challenging, and traditional apps face the technical challenge of flexibly implementing page navigation using AI technology.

[0016] In view of this, embodiments of this application provide an application interaction system, such as... Figure 2 As shown, the application interaction system includes a terminal device 21 and a server 22. The terminal device 21 has a target application installed. The target application is a traditional application. In this embodiment, a functional module called an AI interaction module is added to the traditional application to extend its AI functionality. Specifically, the AI ​​interaction module can be embedded into the target application by updating the traditional application, or by patching or adding a plugin. The server 22 deploys a large AI model and dataset (also known as a question-and-answer knowledge base). This large AI model may include an intent recognition model and a page management model. The AI ​​interaction module can interact with the server 22 to utilize the large AI model deployed on the server 22. Optionally, the AI ​​interaction module can interact with the server 22 to call the large AI model through the server 22, or it can directly call the large AI model by integrating its API. The functionality of the AI ​​interaction module is described in the following method embodiments and will not be detailed here.

[0017] The AI ​​interaction module exists in two states: active and dormant. The active state, also known as the working state, allows the AI ​​interaction module to sense user commands and interact with the server-side AI model based on those commands, enabling page navigation tailored to the user's instructions. The dormant state, also known as the waiting-to-be-wake-up state or non-working state, allows the user to use the target application based on its existing functions. The target application navigates between pages according to the page navigation logic defined in its program code.

[0018] For example, in the active state, the AI ​​interaction module can be visualized as a floating ball, which is a system-level window independent of page activity and can be displayed on any page of the target application. When the user is inactive for a preset time or actively sends a sleep command, the AI ​​interaction module can automatically switch to sleep mode. In sleep mode, the AI ​​interaction module can automatically hide to free up screen space, and the user can reactivate the AI ​​interaction module through a voice wake-up word or physical button operation.

[0019] In this embodiment, the terminal device 21 can be a mobile phone, tablet computer, wearable device, in-vehicle device, augmented reality (AR) / virtual reality (VR) device, laptop computer, ultra-mobile personal computer (UMPC), netbook, personal digital assistant (PDA), etc., and this embodiment does not impose any limitations on it. This embodiment does not specifically limit the type of electronic device.

[0020] The server 22 can be deployed on a local device or server, or on a server in the cloud. Furthermore, the server can be configured as a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server or cloud server cluster that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud storage, and big data and artificial intelligence platforms.

[0021] Figure 3 This illustration shows a flowchart of an application interaction method provided in an embodiment of this application. The method is applied to an AI interaction module in a target application, such as... Figure 3 As shown, the method includes the following steps S301-S304.

[0022] S301, if a user instruction is detected on the first page of the target application, the user instruction is sent to the intent recognition model located on the server so that the intent recognition model can perform intent recognition on the user instruction and obtain at least one intent information.

[0023] The target application can display the first page, and the AI ​​interaction module, when active, can be displayed on top of the first page. For example... Figure 4 As shown, users can input commands via voice or touch (e.g., on a secondhand housing search page). The AI ​​interaction module can sense user commands; if the command is a voice command, the AI ​​interaction module can use a voice recognition interface to convert the audio into text. Then, the AI ​​interaction module can send the user command to the intent recognition model. The "large model" refers to a deep learning model trained on massive amounts of data, with parameters reaching hundreds of billions or even trillions. By using the large model to perform semantic analysis and intent recognition on user questions, the accuracy of user question recognition can be improved.

[0024] The intent recognition model performs semantic understanding of user commands and handles ambiguous expressions such as polysemous words and homophones through semantic disambiguation. Then, it performs entity recognition, extracting core business entities (e.g., "second-hand housing") and operation types (e.g., "query"). Further, it uses a multi-head attention mechanism to decompose multiple intents; for example, "query" might correspond to a property search or a price search. Finally, it uses a pre-trained real estate knowledge graph for intent classification, mapping standardized entities to dimensions such as business type (new / second-hand / rental) and keywords (query / post / compare), outputting at least one intent piece of information. For example, the identified user intent is {business type: second-hand housing; keyword: query}.

[0025] S302, based on the target intent information in at least one intent information, the page management big model located on the server is invoked, so that the page management big model determines whether page redirection is needed based on the target intent information and obtains the page information of the second page to be redirected when page redirection is needed, the second page being another page of the target application.

[0026] The AI ​​interaction module can receive at least one intent information returned by the intent recognition model. If the at least one intent information corresponds to a unique intent, then that intent information is the target intent information. If the intent information corresponds to multiple intents or ambiguous intents, the intent confirmation module in the AI ​​interaction module can display at least one intent on the first page, allowing the user to select the target intent information from among the at least one intents.

[0027] Furthermore, the AI ​​interaction module can invoke the page management model located on the server based on the target intent information. After receiving the target intent information, the page management model can intelligently analyze and determine whether a page jump is necessary. For example, the page management model determines the necessity of a jump by calculating the semantic relevance between the intent information and the page. First, the page management model can extract the core semantic features of the target intent and perform vectorized matching with the functional tags and metadata descriptions of potential pages, calculating the cosine similarity between the target intent and the page. Based on the cosine similarity, a decision is output. For example, when the semantic coverage of a page with the target intent is greater than or equal to a threshold (e.g., 0.92), it is determined that a page jump is necessary, and this page is the second page to be jumped to. Further, the page management model can obtain the page information of the second page from the page metadata database stored on the server and return it to the AI ​​interaction module. When the semantic coverage of a page with the target intent is less than the threshold, it is determined that a page jump is not necessary. Further, the page management model can search for the answer information corresponding to the target intent information from the question-and-answer knowledge base based on the target intent information and return it to the AI ​​interaction module. Furthermore, when matching page information based on intent, the large-scale page management model can incorporate historical matching results as prior experience to improve matching accuracy.

[0028] For example, the server can maintain a set of intents, which includes multiple intents. Each intent is marked with whether a page redirect is required, and if so, the page to be redirected to is also marked. The page management model matches the processing flow corresponding to the target intent information from the intent set. When the target intent information matches an intent in the intent set, if the intent is marked as requiring a page redirect, it is determined that the target intent requires a page redirect. The page management model further obtains the page information to be redirected to marked by the intent, which is used as the page information for the second page. If the intent is marked as not requiring a page redirect, it is determined that the target intent does not require a page redirect. The page management model further searches for the answer information corresponding to the target intent information from a question-and-answer knowledge base.

[0029] In some implementations, the AI ​​interaction module can generate prompts for the page management model based on the target intent information, call the page management model based on the prompts, and then, guided by the prompts, determine whether a page jump is needed and, if a page jump is needed, obtain the page information of the second page to be jumped to.

[0030] Optionally, a prompt template can be pre-designed. This template includes task description information, which describes the role of the page management model, the functions it needs to perform, and the constraints that must be met to perform those functions. The prompt template also includes a populated area, into which the target intent information is filled to obtain the prompt. The second prompt includes both the task description information and the target intent information.

[0031] S303, Receive the page information of the second page returned by the page management model when it is determined that a page jump is required based on the target intent information.

[0032] When the page management big model determines that a page jump is needed based on the target intent information, the AI ​​interaction module can receive the page information of the second page returned by the page management big model, and then jump from the first page to the second page based on the page information of the second page.

[0033] The page information includes, but is not limited to, page title (title), page category (cateid), page redirect address (url), and page input parameters. The page title is a text tag describing the core content of the page, containing key business elements and strongly related to the user's intent. For example, when the user's instruction is "fully furnished loft for rent," the page title would be "For Rent." Alternatively, the page title could be "Loan Calculation," "Property Listing," etc. The page category is a unique code identifying the business type of the target page; for example, the business type could be "used housing," "new housing," etc. The page redirect address is the target address to which the page redirects. Page input parameters are the data set passed to the target page, ensuring that the target page can dynamically load content. For example, input parameters can include the user's input budget range, geographical location, etc. These parameters must be aligned with the pre-filling logic of the target page's form fields.

[0034] If the page management big model determines that no page jump is needed based on the target intent information, the AI ​​interaction module can receive the answer information corresponding to the target intent information returned by the page management big model and display the answer information corresponding to the target intent information on the first page for the user to view.

[0035] Optionally, if the page management model determines that page redirection is unnecessary based on the target intent information, it can return a target field, such as "false," indicating that page redirection is not required. The AI ​​interaction module can receive the target field returned by the page management model. Furthermore, the AI ​​interaction module can search for the corresponding answer information from the question-and-answer knowledge base and display the answer information on the first page for the user to view.

[0036] S304, based on the page information of the second page, jump from the first page to the second page, wherein there is no jump relationship defined in the program code of the target application between the first page and the second page, or there is a jump relationship defined in the program code of the target application.

[0037] In this embodiment, the navigation relationships between pages are divided into two types: one is where there is no navigation relationship defined in the target application's program code; in other words, the target application's program code does not define the navigation relationship between pages. The other is where there is a navigation relationship defined in the target application's program code; in other words, the target application's program code defines the navigation relationship between pages. It should be noted that defining the navigation relationship between pages in the target application's program code can be either static or dynamic. Static definition means that the navigation relationship between pages is hard-coded in the target application's program code, while dynamic definition means that the target application's program code has dynamic routing capabilities. The route determination occurs during the user's use of the application (runtime). Based on dynamic routing capabilities, the navigation relationship between pages can be dynamically determined according to the application's runtime state. The runtime state of the application includes, but is not limited to: component / page level state: such as whether a form has been filled out or a list item has been selected; global application state: such as whether a user is logged in, user role / permission, current theme, and language; data state: such as the content of API response data and whether data loading has been completed; route parameters themselves: such as the id parsed from the URL; location state: such as the query parameter ?search=term, etc.

[0038] When determining whether to redirect, the page management model can identify the second page and obtain its information, regardless of whether the target application's program code specifies the redirection relationship between the first and second pages, and then return the second page's information to the AI ​​interaction module.

[0039] In one implementation, when navigating from the first page to the second page based on the page information of the second page, the AI ​​interaction module can search for a jump path from the first page to the second page within the page navigation relationships defined in the target application's program code, based on the page information of the second page. If no such path exists, the AI ​​interaction module can create a jump path from the first page to the second page based on the page information of the first page and the second page. Then, the user can navigate from the first page to the second page based on this jump path.

[0040] For example, when creating a jump path from the first page to the second page, the AI ​​interaction module can extract information from the first page (such as context parameters, page identifiers, operable components, etc.), and then use the page information of the second page as the target endpoint. Based on the context of the first page, it dynamically constructs routing bridging logic, generates a path configuration in memory containing the first page identifier, target endpoint, and parameter mapping relationships, and registers this path configuration to the runtime routing table by calling the system-level routing API. Furthermore, based on the dynamically generated path configuration, it calls the jump method of the underlying framework to execute the jump, realizing the jump from the first page to the second page. For example, if the user command is "return to shopping cart", the AI ​​interaction module can realize a direct jump from the current first page to the shopping cart page (even if the code does not preset a path). This embodiment of the application, through this dynamic path creation mechanism, can realize direct jumps between any pages, support any combination of pages, break the static constraints of traditional hard-coded jumps, and is more flexible.

[0041] For example, if the page navigation relationships defined in the target application's program code do not include a direct path from the first page to the second page, the AI ​​interaction module can invoke a path generation model located on the server. This model then creates a direct path from the first page to the second page based on the page information of both pages and returns it to the AI ​​interaction module. The AI ​​interaction module then navigates from the first page to the second page based on this direct path. Specifically, the path generation model can extract page functional features through semantic analysis based on the page information of both pages, generate candidate paths based on a page relationship graph and a pre-defined rule base, evaluate path feasibility using reinforcement learning, and finally output the optimal direct path.

[0042] Furthermore, the AI ​​interaction module can update the jump path from the first page to the second page in the page jump relationship defined in the target application's program code.

[0043] Optionally, if the page navigation relationship defined in the target application's program code contains a navigation path from the first page to the second page, the AI ​​interaction module can navigate from the first page to the second page based on this existing navigation path.

[0044] In another implementation, when navigating from the first page to the second page based on the page information of the second page, the AI ​​interaction module can navigate from the first page to the second page without going through intermediate pages. In this way, even if the navigation path between the first and second pages specified in the target application's code includes at least one intermediate page, the AI ​​interaction module can directly reach the desired page using natural language, bypassing the intermediate pages specified in the target application's code, achieving a "one-step" effect and improving the user experience.

[0045] For example, after obtaining the information of the second page, the AI ​​interaction module can trigger an interceptor (handler) to detect the information. Upon successful detection, the AI ​​interaction module calls the corresponding page redirection protocol, dynamically filling the second page's information into a predefined redirection protocol template, assembling a complete redirection protocol string. Then, a system-level framework (such as Android's Intent) triggers protocol parsing and executes the redirection action. Specifically, the system-level framework encapsulates the redirection protocol string into a platform standard object, binds the protocol to an explicit intent using the intent setting data method (Intent.setData(Uri.parse())), then matches the second page Activity based on the registered intent filter (IntentFilter) and routes it to the corresponding Activity class. The second page Activity, in its onCreate() lifecycle, calls the getIntent().getData().getQueryParameter() method to extract key fields such as meta_url, and completes page rendering using the native rendering engine.

[0046] Interceptors are used to check whether user permissions or business conditions meet the redirection requirements before executing the redirection, ensuring the legality of the redirection. For example, users can only access the personal settings page if they are logged in.

[0047] Among them, the jump protocol is a protocol specification in an app for switching between different pages. It is used to implement in-app directional navigation, and its core lies in triggering protocol class path parsing and executing jump actions through a predefined protocol format. For example, a jump protocol format can use "openxxxxx: / / " as the protocol header, followed by the path and detailed jump parameters contained in params to form a complete jump instruction.

[0048] An example of a specific jump protocol is as follows: openxxxxx: / / jump / house / list?params={"meta_url": "xxx.x / zufang / app / rentlist / api_rentlist_data?page_type=zufang","action":"pagetrans","actiontype":"","tradeline":"house","title":"chuzu","is_backtomain":false,"pagetype":"list","list_name":"chuzu","cateid":"37031","logParam":"","params":{"nsource":"house_sou","key":"Jiuxianqiao North Road No.5 Courtyard"},"filterParams":{},"use_cache":false}. Wherein, the path part / house / list indicates jumping to the housing listing page. Detailed jump parameters include meta_url specifies the data interface address (rental housing listing), title defines the page title ("chuzu" represents the rental channel), cateid marks the service category (37031 corresponds to second-hand housing), params.key delivers the search keyword (specific community Jiuxianqiao North Road No.5 Courtyard).

[0049] In some cases, after the page management large model determines that page jumping is required but does not return the page information of the second page, at this time, the AI interaction module can determine the third page having a jump relationship with the first page from the jump relationship between pages defined in the program code of the target application, and acquire the jump path from the first page to the third page; and jump from the first page to the third page based on the jump path from the first page to the third page.

[0050] For example, the AI interaction module can parse the routing configuration file of the application to construct a complete page jump relationship graph, determine a plurality of candidate pages having a jump relationship with the first page, then perform semantic matching on the plurality of candidate pages according to the target intention information, screen out the candidate page matching the target intention information as the third page, then acquire the jump path from the first page to this candidate page as the jump path from the first page to the third page; and then jump from the first page to the third page based on this jump path.

[0051] Furthermore, the AI interaction module can extract the page information of the third page and send the page information of the third page to the server for the server to store the page information of the third page into the intention set.

[0052] As Figure 4 As shown, the AI ​​interaction module can extract page information from the third page according to the jump protocol, add the extracted page information to the first page change list, and mark the page information with the target intent information. Then, the AI ​​interaction module reports the first page change list to the intent registry center on the server in real time or periodically, and the intent registry center on the server updates the page change list to the intent set on the server.

[0053] The third page can be the same as or different from the second page. If the target application's code specifies the navigation relationship between pages, the third page can be the same as the second page. For example, if the target application's code specifies the navigation relationship between page A and page B, but not between page A and page C, then when the page to be navigated to is page B, page B can be either the second or third page, in which case the third page is the same as the second page. When the page to be navigated to is page C, page C is the second page.

[0054] The method provided in this application embodiment does not require updating all existing pages of the target application to the intent set at once. Instead, it can be dynamically updated as the target application is used. This reduces the workload of manually organizing pages, the workload of manual updates and maintenance, and lowers the cost of transforming traditional APPs into AI applications.

[0055] Furthermore, the target application uses the first protocol when developing its pages, and extracts page information based on the first protocol. The page management model supports the second protocol. When the first and second protocols are the same, the first page change list can be directly updated to the server's intent set.

[0056] When the first and second protocols differ, the AI ​​interaction module extracts the page information based on the first protocol, adds it to the first page change list, and reports it to the server's intent registry. The intent registry then calls the target toolset to convert the page information based on the first protocol into page information based on the second protocol, obtaining a second page change list that includes the page information based on the second protocol. Finally, the second page change list is updated in the server's intent set.

[0057] For example, the first protocol can be Transmission Control Protocol (TCP), Internet Protocol (IP), and Hypertext Transfer Protocol (HTTP), etc., and the second protocol can be Model Context Protocol (MCP). The target toolset can be the MCP TOOL toolset. The intent registry can call the MCP TOOL toolset through an event-driven mechanism to convert page information based on the first protocol into a standard format conforming to the MCP protocol. The MCP protocol aims to unify the communication protocol between large models and external data sources and tools, enabling the standardized connection of various data sources, tools, and functions to large AI models. The MCP TOOL toolset is a standardized development kit based on the MCP protocol, using pre-built adapters, type converters, and security sandboxes to convert page information based on the first protocol into an MCP-compatible standard format.

[0058] The method provided in this application utilizes the MCP protocol and toolset to convert traditional APP pages into an AI-recognizable protocol format, reducing the workload of manual page protocol adaptation and lowering the cost of transforming traditional APPs into AI applications. Furthermore, through automated processing and standardized protocols, errors caused by manual operation can be reduced, improving the stability and reliability of the system.

[0059] The AI ​​interaction module can also identify whether the target application version has been updated through incremental code scanning; if the target application version has been updated, it extracts the updated page information from the updated target application; it sends the updated page information to the server so that the server can call the target toolset to convert the updated page information into page information based on the second protocol, and saves the updated page information based on the second protocol to the server.

[0060] For example, the AI ​​interaction module can monitor version updates of the target application. After each new version of the target application is integrated and released, it scans the page path through incremental code scanning to identify the updated (including added or modified) pages, extracts the updated page information and adds it to the first page change list. The AI ​​interaction module reports the first page change list to the intent registry center on the server. The intent registry center on the server adds intents to each page information in the page change list and then updates the intents and page information to the intent set on the server.

[0061] Optionally, when the target application needs to be integrated and released by a new version, incremental code scanning can also be performed on the server side. The server identifies the updated page, extracts the updated page information and adds it to the first page change list, and then the intent registry center performs subsequent processing.

[0062] In this way, when the APP version is updated, the updated page can be automatically added to the intent set, reducing the workload of manual updates and maintenance and improving development and maintenance efficiency. The method provided in this application embodiment, through the intent registry center and page change list, can realize centralized management and dynamic updates of page information, which helps to break down data silos, improve the efficiency of data integration and utilization, support the instant addition and updating of pages, and can quickly adapt to new page and function changes.

[0063] In summary, the method provided in this application embodiment sets up an AI interaction module in the target application. This module senses user commands on the first page of the target application, interacts with a server-side AI model based on the user commands, and obtains at least one intent information through intent recognition via the AI ​​model. Then, based on the target intent information from the at least one intent information, it calls a page management model located on the server. The page management model provides page jump judgment and page information for the second page. The AI ​​interaction module can then execute page jump based on the page information of the second page. This allows users to directly access the desired page through natural language in traditional applications, reducing operation steps, improving operational efficiency, and reducing the difficulty for users to find functions and pages. Furthermore, the jump path between the first and second pages may or may not exist in the target application's program code. Thus, even if the target application code does not define the jump logic between pages, the method provided in this application embodiment can flexibly implement page jump functionality using AI technology. Furthermore, since this application does not limit the target application code to include page navigation logic, it eliminates the need to modify the target application's code when pages are added or modified. This allows for real-time page addition and updates, enabling the system to quickly adapt to new page and functional changes, thus enhancing its flexibility and scalability. In summary, the method provided in this application's embodiments achieves efficient integration of AI and traditional apps, improves user experience, reduces development and maintenance costs, and enhances system flexibility and scalability, demonstrating significant commercial and technical value.

[0064] In some of the processes described in the above embodiments and accompanying drawings, multiple operations are included that appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or they may be executed in parallel. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the terms "first," "second," etc., used herein are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0065] Figure 5This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 5 As shown, the electronic device includes a memory 51 and a processor 52.

[0066] Memory 51 is used to store computer programs and can be configured to store various other data to support operation on the computing platform. Examples of this data include instructions for any application or method operating on the computing platform, data structures, contact data, phone book data, messages, pictures, videos, etc.

[0067] The processor 52, coupled to the memory 51, is used to execute the computer program in the memory 51 for the following purposes: if a user instruction is detected on the first page of the target application, the user instruction is sent to an intent recognition model located on the server, so that the intent recognition model can identify the user instruction and obtain at least one intent information; based on the target intent information in the at least one intent information, the page management model located on the server is invoked, so that the page management model determines whether a page jump is needed based on the target intent information and, if a page jump is needed, obtains the page information of the second page to which the user should jump, the second page being another page of the target application; the page management model receives the page information of the second page returned by the page management model when it determines that a page jump is needed based on the target intent information; based on the page information of the second page, the user jumps from the first page to the second page, wherein there is no jump relationship defined in the program code of the target application between the first page and the second page, or there is a jump relationship defined in the program code of the target application.

[0068] Optionally, the processor 52 is further configured to, based on the page information of the second page, search for a jump path from the first page to the second page in the jump relationship between pages defined in the program code of the target application; if it does not exist, create a jump path from the first page to the second page based on the page information of the first page and the page information of the second page; and jump from the first page to the second page based on the jump path from the first page to the second page.

[0069] Optionally, the processor 52 is also configured to update the jump path from the first page to the second page in the jump relationship between pages defined in the program code of the target application.

[0070] Optionally, the processor 52 is further configured to jump from the first page to the second page without going through an intermediate page, based on the page information of the second page, wherein the jump path between the first page and the second page as defined in the program code of the target application includes at least one intermediate page.

[0071] Optionally, the processor 52 is further configured to determine target intent information from at least one type of intent information; based on the target intent information, invoke a large page management model located on the server so that the large page management model matches the processing flow corresponding to the target intent information from the intent set. The intent set includes multiple intents, each intent indicating whether to perform a page jump and indicating the page information to be jumped to when the page jump occurs.

[0072] Optionally, the processor 52 is further configured to receive at least one intent information returned by the intent recognition big model, and display at least one intent on the first page for the user to select target intent information from at least one intent; generate prompt words for the page management big model based on the target intent information, and call the page management big model based on the prompt words.

[0073] Optionally, the processor 52 is further configured to, if the page management big model does not return page information of the second page, determine the third page that has a jump relationship with the first page from the jump relationship between pages defined in the program code of the target application, and obtain the jump path from the first page to the third page; and jump from the first page to the third page based on the jump path from the first page to the third page.

[0074] Optionally, the processor 52 is further configured to extract the page information of the third page from the intent set maintained by the server in the case of the page information of the second page in the page management big model, and send the page information of the third page to the server so that the server can store the page information of the third page in the intent set.

[0075] Optionally, the processor 52 is further configured to extract page information based on the first protocol for the third page, send the page information based on the first protocol to the server, so that the server can call the target toolset to convert the page information based on the first protocol into page information based on the second protocol, and store the page information based on the second protocol. The first protocol is the protocol used when the target application develops the page, and the second protocol is the protocol supported by the page management big model. The first protocol and the second protocol are different.

[0076] Optionally, the processor 52 is further configured to identify whether the version of the target application has been updated through incremental code scanning; if the version of the target application has been updated, extract the updated page information from the updated target application; send the updated page information to the server so that the server can call the target toolset to convert the updated page information into page information based on the second protocol, and store the updated page information based on the second protocol.

[0077] Optionally, the processor 52 is further configured to receive the answer information corresponding to the target intent information returned by the page management big model when it is determined that no page jump is required based on the target intent information, wherein the answer information is retrieved from the question-and-answer knowledge base by the page management big model based on the target intent information and returned to the AI ​​interaction module; and display the answer information corresponding to the target intent information on the first page for the user to view.

[0078] Furthermore, such as Figure 5 As shown, the electronic device also includes other components such as a communication component 53, a display 54, a power supply component 55, and an audio component 56. Figure 5 The diagram only shows some components and does not mean that the electronic device includes only these components. Figure 5 The components shown. Additionally... Figure 5 The components within the dashed box are optional, not mandatory, and their specific requirements depend on the product form of the work node. In this embodiment, the work node can be a terminal device such as a desktop computer, laptop computer, smartphone, or IoT device, or a server-side device such as a conventional server, cloud server, or server array. If the work node in this embodiment is implemented as a terminal device such as a desktop computer, laptop computer, or smartphone, it may include... Figure 5 The components within the dashed box; if the working node in this embodiment is implemented as a server-side device such as a conventional server, cloud server, or server array, it may be omitted. Figure 5 The component within the dashed box.

[0079] The aforementioned memory can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random-Access Memory (SRAM), Electrically Erasable Programmable Read Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0080] The aforementioned communication component is configured to facilitate wired or wireless communication between the device containing the communication component and other devices. The device containing the communication component can access wireless networks based on communication standards, such as 2G, 3G, 4G / LTE, 5G, or combinations thereof. In one exemplary embodiment, the communication component receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel.

[0081] The aforementioned display includes a screen, which may include a Liquid Crystal Display (LCD) and a Touch Panel (TP). If the screen includes a Touch Panel, the screen can be implemented as a touchscreen to receive input signals from the user. The Touch Panel includes one or more touch sensors to sense touches, swipes, and gestures on the Touch Panel. The touch sensors can sense not only the boundaries of touch or swipe actions but also the duration and pressure associated with the touch or swipe operation.

[0082] The aforementioned power supply components provide power to various components within the device in which they reside. These power supply components may include a power management system, one or more power sources, and other components associated with generating, managing, and distributing power to the device in which they reside.

[0083] The aforementioned audio component can be configured to output and / or input audio signals. For example, the audio component includes a microphone (MIC) configured to receive external audio signals when the device containing the audio component is in an operating mode, such as call mode, recording mode, or voice recognition mode. The received audio signals can be further stored in memory or transmitted via a communication component. In some embodiments, the audio component also includes a speaker for outputting audio signals.

[0084] Accordingly, embodiments of this application also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to implement the steps in the above-described method embodiments. The computer-readable storage medium includes volatile or non-volatile components, or a combination thereof, and can be removable or non-removable. Examples of computer-readable storage media include, but are not limited to, phase-change random access memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random-access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), flash memory or other memory technologies, CD-ROM, Digital Video Disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium.

[0085] Accordingly, this application also provides a computer program product, which includes a computer program or instructions that, when executed by a processor, cause the processor to implement the steps in the above method embodiments. It should be understood that each step or combination of steps in the above method flow can be implemented by the computer program or instructions. Furthermore, these computer programs or instructions can be applied to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device, enabling the processor of the general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to function as an apparatus for implementing the corresponding functions in the above method embodiments.

[0086] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0087] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. An application interaction method, characterized in that, An AI interaction module applied in a target application, wherein the AI ​​interaction module is used to perceive user commands in an active state, and interact with a server-side AI model based on the user commands, so as to complete page jumps adapted to the user commands with the help of the AI ​​model, the method including: If a user instruction is detected on the first page of the target application, the user instruction is sent to the intent recognition model located on the server so that the intent recognition model can perform intent recognition on the user instruction and obtain at least one intent information. Based on the target intent information in the at least one intent information, the page management big model located on the server is invoked, so that the page management big model determines whether page redirection is needed based on the target intent information and obtains the page information of the second page to be redirected when page redirection is needed, the second page being another page of the target application; The page management model receives the page information of the second page when it determines that a page jump is required based on the target intent information; Based on the page information of the second page, a jump is made from the first page to the second page, wherein there is no jump relationship defined in the program code of the target application between the first page and the second page, or there is a jump relationship defined in the program code of the target application.

2. The method according to claim 1, characterized in that, Based on the page information of the second page, navigating from the first page to the second page includes: Based on the page information of the second page, search for a jump path from the first page to the second page in the page jump relationship defined in the program code of the target application; If it does not exist, create a jump path from the first page to the second page based on the page information of the first page and the page information of the second page; Based on the jump path from the first page to the second page, jump from the first page to the second page.

3. The method according to claim 2, characterized in that, Also includes: Update the jump path from the first page to the second page to the page jump relationship defined in the program code of the target application.

4. The method according to claim 1, characterized in that, Based on the page information of the second page, navigating from the first page to the second page includes: Based on the page information of the second page, the user jumps from the first page to the second page without going through an intermediate page, wherein the jump path between the first page and the second page defined in the program code of the target application includes at least one intermediate page.

5. The method according to claim 1, characterized in that, Based on the target intent information from the at least one intent information, a large-scale page management model located on the server is invoked, so that the large-scale page management model determines whether a page redirection is needed based on the target intent information, and obtains the page information of the second page to be redirected to when a page redirection is needed, including: The target intent information is determined from the at least one intent information; Based on the target intent information, the page management big model located on the server is invoked so that the page management big model matches the processing flow corresponding to the target intent information from the intent set. The intent set includes multiple intents, each intent is marked whether a page jump is to be performed, and when a page jump is performed, the page information to be jumped to is marked.

6. The method according to claim 5, characterized in that, Determining the target intent information from the at least one intent information includes: The method further includes receiving at least one intent information returned by the intent recognition model and displaying the at least one intent on the first page for the user to select the target intent information from the at least one intent; the method also includes: Based on the target intent information, a prompt word is generated for the page management model, and the page management model is invoked based on the prompt word.

7. The method according to any one of claims 1-6, characterized in that, The method further includes: If the page management model does not return the page information of the second page, determine the third page that has a jump relationship with the first page from the jump relationship between pages defined in the program code of the target application, and obtain the jump path from the first page to the third page; Based on the jump path from the first page to the third page, jump from the first page to the third page.

8. The method according to claim 7, characterized in that, Also includes: When the page management model obtains the page information of the second page from the intent set maintained by the server, it extracts the page information of the third page and sends the page information of the third page to the server so that the server can store the page information of the third page in the intent set.

9. The method according to claim 8, characterized in that, The step of extracting the page information of the third page and sending the page information of the third page to the server includes: For the third page, page information based on the first protocol is extracted and sent to the server so that the server can call the target toolset to convert the page information based on the first protocol into page information based on the second protocol and store the page information based on the second protocol. The first protocol is the protocol used when developing the target application page, and the second protocol is the protocol supported by the page management big model. The first protocol and the second protocol are different.

10. The method according to claim 9, characterized in that, Also includes: Incremental code scanning is used to identify whether the target application has been updated. If the target application is updated, extract the updated page information from the updated target application; The updated page information is sent to the server so that the server can call the target toolset to convert the updated page information into page information based on the second protocol and store the updated page information based on the second protocol.

11. The method according to claim 1, characterized in that, The method further includes: The page management model receives the answer information corresponding to the target intent information when it determines that no page jump is needed based on the target intent information. The answer information is retrieved from the question-and-answer knowledge base by the page management model and returned to the AI ​​interaction module based on the target intent information. The answer information corresponding to the target intent information is displayed on the first page for the user to view.

12. An electronic device, characterized in that, include: A memory and a processor; wherein the memory stores executable code, which, when executed by the processor, causes the processor to perform the method as described in any one of claims 1 to 11.

13. A non-transitory machine-readable storage medium, characterized in that, The non-transitory machine-readable storage medium stores executable code that, when executed by a processor of an electronic device, causes the processor to perform the method as described in any one of claims 1 to 11.

Images