Browser-based AI intelligent interactive software upgrading method

Abstract

The invention relates to the technical field of computer software, in particular to a browser-based AI intelligent interactive software upgrading method, and provides an upgrading method for solving the problem of how to autonomously improve the interactive experience of a user according to the use record of the user. Comprising the steps of setting a simulation account of target software, simulating login operation of a user through the simulation account, and extracting an operation interface of the target software; determining a page position and a page range of the operation interface in the detail interface according to the page label; labeling the detail interface in a mode of combining manual labeling and automatic identification, and inputting the detail interface into an interactive interface; obtaining an operation record executed by a user through the interactive interface, and judging whether the page position and the page range need to be adjusted or not according to the operation record.

Description

Technical Field

[0001] This invention relates to the field of computer software technology, and more specifically, to a method for upgrading browser-based AI intelligent interactive software. Background Technology

[0002] With the rapid development of artificial intelligence technology, users' demands for software interaction experiences are increasing. Traditional browser-based software systems mostly adopt traditional human-computer interaction methods, such as using mouse clicks and keyboard input to complete various functions. However, this interaction method is gradually revealing many drawbacks when facing increasingly complex business scenarios and diverse user needs.

[0003] On the one hand, traditional interaction methods are cumbersome. Users need to search through numerous menu options and buttons to find the function they need. For users who are not familiar with software operation, finding and using a specific function may take a lot of time. For example, in some enterprise resource planning (ERP) software, if a user wants to execute a specific business process, they may need to search through multiple levels of menus, which is complicated and reduces work efficiency.

[0004] On the other hand, traditional interaction methods lack flexibility and intelligence. They cannot understand users' natural language expressions and cannot make intelligent recommendations or automatically execute related operations based on users' intentions. For example, when a user wants to query sales data for a certain period of time, they need to manually enter the specific time range, query conditions, and other information into the software's query interface. If the user expresses "view last month's sales situation," traditional software cannot recognize and process such a natural language instruction.

[0005] In terms of interaction methods, the software not only needs to be able to recognize the user's intent, but also needs to be able to automatically upgrade itself based on the user's usage patterns, so that the software layout is as perfect as possible and the user's interactive experience is improved. The user's usage records are obviously an important source of data for providing upgrade directions. Therefore, how to automatically improve the user's interactive experience based on the user's usage records is one of the problems that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] The problem this invention addresses is: how to proactively improve the user's interactive experience based on the user's usage history.

[0007] To address the aforementioned issues, this invention provides an upgrade method for browser-based AI intelligent interactive software. The upgrade method includes: setting up a simulated account for the target software; simulating user login operations using the simulated account; extracting the target software's interface; simplifying each interface and determining its position and extent within the details interface based on page tags; annotating the details interface using a combination of manual annotation and automated recognition, and inputting the details interface into the interactive interface; obtaining the user's identity information; hiding some information from the details interface based on the identity information; adjusting the remaining page extent of the interface based on the hidden portion; and obtaining the user's operation records through the interactive interface, determining whether the page position and extent need adjustment based on the operation records.

[0008] Compared with existing technologies, the technical effects achieved by this solution are as follows: By setting up simulated accounts and simulating login operations to extract the target software's interface, the system can accurately acquire all interface information of traditional software. By optimizing the interface structure, the details interface focuses more on core content, and key information areas are accurately identified through page tags, providing clear semantics and spatial positioning for AI intelligent interaction, thus improving interaction efficiency. Each details interface is assigned page tags and business-related information, enabling the AI ​​model to perform intent matching and information retrieval based on this labeled data. The setting of hiding some content based on identity information protects data privacy, while intelligently adjusting the page layout ensures the continuity and aesthetics of the user experience, avoiding the interface fragmentation caused by information hiding in traditional solutions. Finally, by analyzing user behavior data in real time, the rationality of the interface layout is evaluated, potential interaction obstacles are promptly identified, and optimization suggestions are proactively proposed, allowing the software to better adapt to user habits and improve the convenience and accuracy of interaction.

[0009] In one embodiment of the present invention, a simulated account for the target software is set up, and the user login operation is simulated through the simulated account to extract the operation interface of the target software. Specifically, this includes: setting different permission levels according to the application content of the target software; preparing multiple accounts with different permission levels, and obtaining the corresponding permission interface of the target software under different permission levels through simulated login operations.

[0010] Compared to existing technologies, the technical benefits of this solution include: the ability to comprehensively acquire all operational interfaces of the target software under various permission levels. This ensures that the data foundation for subsequent page simplification and annotation is complete and representative, thereby significantly improving the intelligence and applicability of the upgraded AI intelligent interaction software.

[0011] In one embodiment of the present invention, each operation interface is simplified, and the page position and page range of the operation interface in the details interface are determined according to the page tags. Specifically, this includes: obtaining the user's dwell time on each operation interface; determining the usage ratio of each navigation module in the navigation menu based on the usage records of the navigation menu within the dwell time; determining the page tags of the operation interface based on the specific content of the navigation modules; filtering the navigation modules according to the page tags and usage ratios to obtain the retained modules; removing the navigation menu from the operation interface to obtain the details interface, and determining the page position and page range of the retained modules in the details interface.

[0012] Compared to existing technologies, this technical solution achieves the following effects: By acquiring the dwell time on each interface, a basic indicator of user attention to that interface is obtained. Analyzing user usage records of the navigation menu within the dwell time reveals the actual usage frequency and importance of each navigation module, thus determining the usage proportion of each module. The selection of retained modules not only ensures the retention of core functional modules but also effectively eliminates low-frequency or redundant modules. The setting of page position and page scope enables more intelligent interface optimization based on the importance and functional relevance of the retained modules.

[0013] In one embodiment of the present invention, obtaining user identity information, hiding part of the information on the details interface based on the identity information, and adjusting the page range of the remaining operation interface based on the hidden part, specifically includes: identifying the identity information of each user through image capture, determining the information display range based on the permission level corresponding to the identity information; determining the hidden part based on the information display range, and adjusting the page range of the operation interface based on the hidden area of ​​the hidden part.

[0014] Compared with existing technologies, the technical effects achieved by this solution are as follows: real-time acquisition of user identity information through image capture technology ensures the automation and accuracy of identity recognition; the scope of information accessible to users is defined by permission levels, ensuring the security of sensitive information and enabling personalized information display; and the page range of the remaining visible operation interface is adjusted by quantitatively considering the actual area of ​​the hidden part, ensuring the aesthetics of the interface and the smoothness of operation, and avoiding interface fragmentation or inconsistency caused by information hiding.

[0015] In one embodiment of the present invention, the identity information of each user is identified by image capture, and the information display range is determined according to the permission level corresponding to the identity information. Specifically, this includes: when all users have only one permission level, the information display range is directly determined according to the permission level; when all users have multiple permission levels, the preset information range corresponding to each permission level is obtained, and the information display range is selected from the preset information range according to different combinations of multiple permission levels.

[0016] Compared with existing technologies, the technical effects achieved by this solution are as follows: When all users have only one permission level, the system can directly and quickly determine the information display range, avoiding unnecessary complex judgments, thereby improving system response speed and resource utilization. When there are multiple user combinations with different permission levels, by obtaining the preset information range corresponding to each permission level and intelligently selecting information based on different combinations, it ensures that sensitive information is effectively prevented from being leaked to users with lower permissions while guaranteeing normal use for users with higher permissions.

[0017] In one embodiment of the present invention, when all users have multiple permission levels, a preset range of information corresponding to each permission level is obtained, and an information display range is selected from the preset range of information according to different combinations of multiple permission levels. Specifically, this includes: obtaining the weight value corresponding to the permission level and calculating the average weight of all users; obtaining the maximum weight value of all users, determining the level difference based on the difference between the average weight and the maximum weight value, and selecting the information display range based on the level difference.

[0018] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: the setting of selecting the information display range based on the level difference effectively solves the problem that the efficiency and accuracy of information display range selection are insufficient when users with multiple permission levels interact simultaneously, which may lead to inaccurate information display or poor user interaction experience. Under the premise of ensuring information security and privacy, it maximizes the provision of effective information.

[0019] In one embodiment of the present invention, obtaining the operation record performed by the user through the interactive interface, and determining whether the page position and page range need to be adjusted based on the operation record, specifically includes: obtaining the search duration of each time the user clicks on the details interface from the interactive interface; determining whether the user has made an erroneous click based on the user's dwell time on the details interface, and recording the number of erroneous entries for each details interface; obtaining the total number of clicks for each details interface, and calculating the total erroneous entry ratio based on the total number of clicks and the number of erroneous entries; when the total erroneous entry ratio is greater than or equal to a first threshold, the page position or page range of the details interface needs to be adjusted; when the total erroneous entry ratio is less than the first threshold, the page position and page range of the details interface do not need to be adjusted.

[0020] Compared to existing technologies, the technical benefits of this solution are as follows: By acquiring real-time records of user actions on the interactive interface and based on key indicators such as search duration, dwell time, and number of accidental entries, the system can promptly identify detail pages with interactive problems and make targeted adjustments to their page positions or scope. This makes the interface layout no longer static but adaptively respond to the user's actual usage. This significantly improves the efficiency of users finding target information, reduces erroneous clicks, and thus optimizes the overall user experience.

[0021] In one embodiment of the present invention, when the total false entry ratio is greater than or equal to a first threshold, the page position or page range of the details interface needs to be adjusted. Specifically, this includes: statistically analyzing the sub-false entry ratios of the details interface under different information display ranges; correcting the sub-false entry ratios based on historical access records to obtain a corrected ratio; recording the information display ranges with corrected ratios greater than or equal to the first threshold as abnormal display ranges; recording the combination of permission levels corresponding to the abnormal display ranges as abnormal combination methods; obtaining the correction target after a user makes an erroneous click under the abnormal display range; recording the details interface corresponding to the correction target as the interface to be modified; and determining the adjustment method of the interface to be modified based on the difference between the corrected ratio and the first threshold.

[0022] Compared to existing technologies, this technical solution achieves the following effects: By statistically analyzing the sub-mistake ratio, it obtains the frequency of users making erroneous clicks or entering specific details interfaces under different information display ranges. It then correlates page position updates with different permission levels, ensuring that subsequent page layout adjustments effectively guide users to achieve their intended goals, significantly reducing the user error rate in complex permission scenarios. Furthermore, by determining the adjustment method based on the difference between the sub-mistake ratio and a first threshold, the adjustment strategy can adaptively match the severity of the problem, achieving more intelligent and personalized interface optimization. Attached Figure Description

[0023] Figure 1 This is one of the flowcharts for the upgrade method of the AI ​​intelligent interaction software of the present invention; Figure 2 This is the second flowchart of the upgrade method for the AI ​​intelligent interaction software of the present invention; Figure 3 This is the third flowchart of the upgrade method for the AI ​​intelligent interaction software of the present invention. Detailed Implementation

[0024] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0025] [First Embodiment] See Figure 1 In one specific embodiment, this application provides an upgrade method for browser-based AI intelligent interaction software, the upgrade method including: S100. Set up a simulated account for the target software, simulate user login operations using the simulated account, and extract the target software's interface. S200. Simplify the page layout of each operation interface and determine the page position and page range of the operation interface in the details interface based on the page tags; S300: The details interface is annotated using a combination of manual annotation and automated recognition, and the details interface is then input into the interactive interface; S400: Obtain the user's identity information, hide part of the information on the details page based on the identity information, and adjust the page range of the remaining operation interface based on the hidden part; S500: Obtain the operation records performed by the user through the interactive interface, and determine whether the page position and page range need to be adjusted based on the operation records.

[0026] In step S100, the target software refers to an existing browser-based software system that requires an AI intelligent interaction upgrade. It typically includes multiple functional modules and operation pages, accessible and operated by users through a browser, such as internal enterprise management software. The simulated account is a virtual user credential set up to simulate real user behavior. Through this account, users can log in to the target software in a non-human manner and execute preset operation processes to obtain the software's interface information. The operation interface refers to the various pages presented by the target software when a user logs in and performs operations. These pages may include elements such as navigation menus, function buttons, data display areas, and input forms.

[0027] By pre-configuring one or more generic simulated accounts, these accounts are granted basic access permissions to the target software. Then, using automated scripting tools such as Selenium or Puppeteer, the user login process in a browser environment is simulated, including entering account credentials and clicking the login button. After successful login, the script is configured to traverse all accessible links and functional modules of the target software, taking screenshots or parsing the HTML structure of each page to extract the corresponding user interface data. For example, in an enterprise internal management system, a simulated account for a regular employee can be set up. This account logs into the system and automatically accesses all report pages and data entry pages within its authorized scope, extracting the visual presentation and underlying structural information of these pages.

[0028] In step S200, the user interface refers to the various pages presented by the target software when a user logs in and performs operations. These pages may contain elements such as navigation menus, function buttons, data display areas, and input forms, and are simplified using preset methods. For example, by analyzing the common features of a large number of user interfaces, non-core elements such as fixed navigation bars, advertising areas, or copyright information, which are usually located at the top, bottom, or side of the page, are identified. These non-core elements are marked and removed from the user interface, resulting in a more concise details interface. Subsequently, page tags can be generated for the details interface based on keywords or predefined layout templates. For example, a page displaying "sales data" can have its page tag set to "sales report." Based on these page tags, the system can identify the main content areas in the details interface and determine their specific positions and display areas within the entire page.

[0029] In step S300, automated tools are first used to perform preliminary identification and annotation of the details interface. For example, optical character recognition (OCR) technology is used to extract text information on the interface, or image recognition technology is used to identify common UI components (such as buttons and input boxes). The results of the automated identification are then submitted to human reviewers for proofreading and supplementation. Human reviewers can correct parts that the automated identification failed to handle accurately based on business logic and user needs, and add deeper semantic information, such as labeling a data table as "monthly sales statistics". This manually confirmed and supplemented annotation information, along with the structural data of the details interface, is integrated and input into the backend database of the AI ​​intelligent interaction system, serving as the basis for the AI ​​model to understand and respond to user commands.

[0030] In step S400, when a user opens the target software, the system obtains the current user's identity. For mobile software, camera permissions can be bound when the target software is opened, and identity information can be obtained through the phone's camera. For computer software, a facial recognition device can be set up next to the computer to identify the user in real time. Based on different identity information, the system queries the preset access control policy to determine whether the user has the right to view all content in the details interface. For example, for a details interface containing sensitive financial data, if the current user is identified as a non-finance department employee, sensitive information such as specific monetary amounts or customer names in the interface will be automatically hidden by the system. The hiding operation can be achieved by dynamically removing or obscuring the corresponding elements on the front-end page. After some information is hidden, in order to maintain the aesthetics and usability of the interface, the system will automatically adjust the layout of the remaining visible content according to the area and position of the hidden part. For example, the space originally occupied by the hidden content may be folded or redistributed to other visible elements to ensure the rationality of the interface layout.

[0031] In step S500, the system continuously records user behavior on the interactive interface, such as which areas the user clicked, how long they stayed on a particular details page, and whether they frequently returned to the previous page. These operation records are collected and preliminarily analyzed. For example, if the system detects that a large number of users frequently click the back button or switch to other unrelated pages within a short period of time after visiting a specific details page, this may indicate that the page location of the details page is not intuitive enough, or that there are problems with the information layout within its page range, making it difficult for users to quickly find the content they need. In this case, the system will generate an adjustment suggestion, prompting the administrator to review and optimize the page location or page range of the details page.

[0032] For example, when user A accesses the CRM system through the newly developed AI-powered intelligent interactive interface, the system obtains user A's identity information. Assume user A is a regular sales representative whose permissions are set to only view information about clients they are responsible for, and not to view clients' sensitive financial data. When user A attempts to view client B's details via voice command, the system processes the client B details interface based on user A's identity information. Specifically, client B's financial statement area and senior management annotation area are automatically hidden. Simultaneously, to maintain visual balance, the space occupied by the hidden areas is automatically adjusted; for example, the remaining client contact record area is expanded to fill the blank space left by the hidden parts, thus ensuring that the interface layout seen by user A remains logical and easy to read.

[0033] Finally, the system continuously collects operation records from User A and other users through the AI-powered intelligent interactive interface. For example, the system records whether User A frequently clicks the "back" button after viewing the customer details interface, or spends too little time in a specific area. If the system finds that users generally return frequently or switch quickly on a particular details interface, this may indicate that the page location of that details interface is not intuitive enough, or that there are problems with the information layout within its page range. For example, if the "Customer Contact Records" area is placed in an inconspicuous location, causing User A to spend a long time finding it, the system will determine, based on these operation records, that the page location or page range of that details interface may need adjustment, and will issue optimization suggestions to the system administrator, such as suggesting that the "Customer Contact Records" area be moved to a more prominent location or that its display range be expanded to improve the user experience.

[0034] By setting up simulated accounts and simulating login operations to extract the target software's interface, the system can accurately acquire all interface information of traditional software. Optimizing the interface structure allows the details interface to focus more on core content, and page tags precisely identify key information areas, providing clear semantic and spatial positioning for AI intelligent interaction, thus improving interaction efficiency. Assigning page tags and business-related information to each details interface enables the AI ​​model to perform intent matching and information retrieval based on this labeled data. The setting of hiding some content based on identity information protects data privacy while intelligently adjusting the page layout ensures a consistent and aesthetically pleasing user experience, avoiding the interface fragmentation caused by information hiding in traditional solutions. Finally, by analyzing user behavior data in real time, the system evaluates the rationality of the interface layout, promptly identifies potential interaction obstacles, and proactively proposes optimization suggestions, enabling the software to better adapt to user habits and improve the convenience and accuracy of interaction.

[0035] [Second Embodiment] In one specific embodiment, a simulated account for the target software is set up. This simulates a user's login process and extracts the target software's user interface. Specifically, this includes: S110. Set different permission levels according to the application content of the target software; S120. Prepare multiple accounts with different permission levels, and obtain the corresponding permission interfaces of the target software under different permission levels by simulating login operations.

[0036] In steps S110 to S120, the traditional method of setting up a simulated account to simulate user login operations and extract the operation interface may not be able to fully cover the differences in the interface of the software under different user permissions, resulting in incomplete or unrepresentative interfaces. This affects the accuracy of subsequent page simplification and annotation, thereby reducing the intelligence and applicability of the upgraded AI interaction. Therefore, it is necessary to conduct an in-depth analysis of the application content of the target software.

[0037] First, based on business logic and functional modules, various permission levels were set. For example, in an Enterprise Resource Planning (ERP) system, different permission levels could be defined according to business modules (such as finance, purchasing, sales, and inventory) and user roles (such as administrators, department managers, and ordinary employees). Alternatively, by analyzing the target software's database structure or functional module configuration, permission points related to data access and functional operations could be identified and combined into different permission levels. Based on this, corresponding mock accounts were prepared for each defined permission level. These accounts serve as credentials for simulating user login operations, ensuring that the target software's interface behavior is covered across all permission levels. For example, a series of test accounts could be manually created, each assigned a specific permission level. Alternatively, in some automated testing frameworks, mock accounts with different permission combinations could be dynamically generated to meet testing needs. Subsequently, by simulating login operations, these accounts with specific permissions were used to log into the target software one by one, and the interface presented at different permission levels was captured. Simulated login operations can be implemented using various technologies, such as using automated testing tools like Selenium and Puppeteer to simulate browser behavior, entering usernames and passwords, and clicking the login button. Alternatively, you can directly analyze the login requests of the target software, construct HTTP requests to log in, and use web crawling technologies (such as Scrapy and BeautifulSoup) to parse the page content after logging in, thereby obtaining the permission interface under different permission levels.

[0038] For example, consider upgrading an internal OA (Office Automation) system. First, based on the OA system's functional modules, such as "Administrative Management," "Financial Reimbursement," "Project Approval," and "Personal Schedule," and combined with user roles, such as "System Administrator," "Department Head," and "Regular Employee," set different permission levels. For instance, you could define "Administrator Permissions," "Department Manager Permissions," and "Regular Employee Permissions." Next, prepare multiple mock accounts. For example, create an account named "admin_test" with "Administrator Permissions," an account named "manager_test_hr" with "Department Manager Permissions (Human Resources Department)," and an account named "staff_test_dev" with "Regular Employee Permissions (R&D Department)." Then, use automation scripting tools, such as Python combined with the Selenium library, to simulate browser behavior. The script will sequentially use these mock accounts to log into the OA system. After each successful login, the script iterates through all accessible menus and links in the OA system, taking screenshots or scraping HTML content from each page to obtain all interfaces visible under "administrator privileges," "department manager privileges," and "regular employee privileges." For example, under "administrator privileges," interfaces such as system configuration and user management are accessible; under "department manager privileges," interfaces such as department approvals and subordinate performance management are accessible; while under "regular employee privileges," only interfaces such as personal attendance and leave applications are accessible.

[0039] Setting different permission levels allows for comprehensive access to all operational interfaces of the target software under various permission levels. This ensures that the data foundation for subsequent page simplification and annotation is complete and representative, thereby significantly improving the intelligence and applicability of the upgraded AI intelligent interaction software.

[0040] [Third Embodiment] See Figure 2 In one specific embodiment, each operation interface is simplified, and the page position and page range of the operation interface in the details interface are determined according to the page tags, specifically including: S210. Obtain the duration of the user's stay on each operation interface, and determine the usage ratio of each navigation module in the navigation menu based on the usage records of the navigation menu within the stay duration; S220. Determine the page labels of the operation interface based on the specific content of the navigation module, and filter the navigation modules according to the page labels and usage ratio to obtain the retained modules; S230. Remove the navigation menu from the operation interface to obtain the details interface, and determine the page position and page range of the retained module within the details interface.

[0041] In step S210, the dwell time can be captured in real time by deploying JavaScript code on the browser side to listen for the page's `focus` and `blur` events, as well as mouse and keyboard activity events. Timing starts when the page is active and there is user interaction, and pauses when the page loses focus or there is no interaction for a certain threshold, and the duration data is reported to the backend server. Alternatively, server logs or gateway logs can be analyzed to record the start and end timestamps of user requests for specific page resources, and combined with session management mechanisms, the approximate dwell time of the user on each operation interface can be estimated.

[0042] By analyzing log data of user interactions with various modules in the navigation menu during their dwell time, such as clicks, hovers, or activations, the frequency of each navigation module's access or attention can be calculated, thus determining its relative importance or usage proportion within the entire navigation menu. This can be achieved by capturing user interactions with navigation modules on the front end using event listeners and recording this data along with timestamps. The backend service receives this data and can aggregate and statistically analyze interaction events within a specific time period to calculate the number of clicks or activation duration for each navigation module, thereby determining its usage proportion. Alternatively, by analyzing page element click heatmaps or event tracking data collected by user behavior analysis tools, the access frequency and user paths of navigation modules can be extracted, and usage proportions can be determined through data mining and statistical analysis.

[0043] In step S220, determining that page tags can categorize navigation modules with similar functions or related content can be achieved through Natural Language Processing (NLP) technology, which involves keyword extraction, topic modeling, or text classification of the navigation module's text content to automatically generate or recommend page tags. For example, modules such as "Financial Statements" and "Income and Expense Details" can be automatically tagged with "Financial Management." Alternatively, through manual review and configuration, domain experts or operations personnel can manually assign one or more predefined page tags to each navigation module or its associated interface based on the actual function and business logic of the navigation module.

[0044] Select the modules that are more important, frequently used, or core to users from the original navigation menu and keep them. At the same time, for modules with the same page tags, even if the usage of a single module is not high, if the overall usage of its page tag is high, it can be considered to integrate or retain it.

[0045] In step S230, by removing the navigation menu, the user's attention is focused on the core content of the page, reducing visual interference. It should be noted that the navigation function of the details interface is not missing compared to the operation interface; it is just more streamlined and highlights the navigation modules with higher usage.

[0046] For example, when a user visits the "Sales Order Management" interface in an Enterprise Resource Planning (ERP) system, the system continuously records the user's activity time on that interface through front-end tracking scripts. Timers run continuously as the user clicks, scrolls, or types on the interface. They pause or reset when the user switches to another tab, minimizes the browser, or remains inactive for an extended period. The total time the user spends on the interface is then reported to the backend user behavior analytics service. Simultaneously, the system records every click or activation of navigation modules within the "Sales Order Management" interface, such as "Create Order," "View Order List," "Order Approval," "Customer Management," and "Product Inventory," during the user's stay. For instance, a user might frequently click "View Order List" and "Create Order," but less frequently "Order Approval" or "Customer Management." The backend service calculates the usage percentage of each navigation module based on these records; for example, "View Order List" might have a usage percentage of 40%, "Create Order" 30%, "Order Approval" 15%, "Customer Management" 10%, and "Product Inventory" 5%. Next, the system automatically or semi-automatically determines the page tags for each navigation module based on its text content. For example, "Create Order," "View Order List," and "Order Approval" can be tagged as "Order Management," "Customer Management" as "Customer Relationships," and "Product Inventory" as "Inventory Management." Then, the system filters the navigation modules based on these page tags and their usage percentage. Typically, modules with a usage percentage below 10% and not part of the core business process are removed. For the "Order Management" tag, even though the "Order Approval" module has a relatively low usage percentage, it still needs to be retained due to its criticality in the business process. After filtering, the "Create Order," "View Order List," and "Order Approval" modules are retained. Subsequently, when rendering the "Sales Order Management" interface, the system will no longer display the original top or side navigation menu, resulting in a details interface containing only core business content. In this details interface, the system dynamically adjusts the layout of the retained modules based on their usage percentage. "View Order List" and "Create Order" are used more frequently and may be placed in a prominent position at the top or left of the details page, occupying a large portion of the page, and presented as buttons or cards; while "Order Approval" may be placed on the right or bottom of the details page as a smaller icon or secondary link, occupying a smaller portion of the page.

[0047] By capturing the dwell time on each interface, we obtain basic metrics of user attention to that interface. Analyzing user usage records of the navigation menus within that dwell time reveals the actual usage frequency and importance of each navigation module, thus determining the usage proportion of each module. The selection of retained modules not only ensures the retention of core functional modules but also effectively eliminates low-frequency or redundant modules. Setting page positions and page scope allows for more intelligent interface optimization based on the importance and functional relevance of the retained modules.

[0048] [Fourth Embodiment] In one specific embodiment, the user's identity information is obtained, a portion of the information on the details screen is hidden based on the identity information, and the remaining page range of the operation interface is adjusted based on the hidden portion. Specifically, this includes: S410. Identify each user's identity information through image capture, and determine the information display range based on the permission level corresponding to the identity information; S420. Determine the hidden portion based on the information display range, and adjust the page range of the operation interface based on the hidden area of ​​the hidden portion.

[0049] In steps S410 to S420, the identity information can be obtained by capturing the user's facial image through a camera module integrated into the browser environment, and then comparing it with a pre-stored user facial feature database using facial recognition technology, thereby identifying the user's identity. After identifying a user, the system queries the corresponding permission level through a pre-defined permission level mapping table to determine the user's information display scope. This is achieved by comparing all displayable information elements in the current details interface with the user's information display scope. Any information element whose content or associated data exceeds the user's permission scope is marked as hidden. Alternatively, during interface rendering, interface elements can be dynamically generated or loaded based on preset interface templates and permission rules. Modules or data that do not fall within the user's information display scope will not be rendered or loaded, thus achieving hiding.

[0050] By acquiring user identity information in real time through image capture technology, the automation and accuracy of identity recognition are ensured. The scope of information that users can access is defined by permission levels, ensuring the security of sensitive information and enabling personalized information display. By quantitatively considering the actual area of ​​the hidden parts, the page range of the remaining visible operation interface is adjusted, ensuring the aesthetics of the interface and the smoothness of operation, and avoiding interface fragmentation or inconsistency caused by information hiding.

[0051] [Fifth Embodiment] In one specific embodiment, the identity information of each user is identified through image capture, and the information display range is determined according to the permission level corresponding to the identity information, specifically including: S411. When all users have only one permission level, the information display range is directly determined according to the permission level. S412. When all users have multiple permission levels, obtain the information preset range corresponding to each permission level, and select the information display range from the information preset range according to different combinations of multiple permission levels.

[0052] In step S411, when existing methods capture and identify identity information through image capture and determine the information display range based on permission levels to hide some information and adjust the page range, the determination of the information display range may be inaccurate or inefficient when the user has multiple permission levels, resulting in a mismatch between the displayed content and the user's permission combination, which affects the interactive experience.

[0053] When the system recognizes that all currently active users have the same permission level, it will directly load the preset display range for that permission level. For example, if all currently active users have "ordinary employee" permissions, the system will directly load the preset display range corresponding to "ordinary employee".

[0054] In step S412, the system has a mapping table between permissions and display ranges, which defines in detail the information range that each permission level can view or operate. When all users have multiple permission levels, the system first obtains the preset information range corresponding to each permission level. Based on different combinations of multiple permission levels, the system selects the final information display range from the obtained preset information ranges. Specifically, the system identifies the set of permission levels for all current users and calculates or selects the most suitable information display range according to a preset combination strategy.

[0055] For example, permission levels include "administrator", "department manager" and "ordinary employee". When an "administrator" and an "ordinary employee" are operating or demonstrating on the same interactive interface at the same time, the system can prioritize the information preset range of the "ordinary employee" according to the preset demonstration mode combination strategy to avoid the leakage of sensitive information, but at the same time allow the "administrator" to perform higher-level operations without displaying them to the "ordinary employee".

[0056] When all users have only one permission level, the system can directly and quickly determine the information display range, avoiding unnecessary complex judgments and thus improving system response speed and resource utilization. When there are multiple user combinations with different permission levels, the system obtains the preset information range corresponding to each permission level and intelligently selects information based on different combinations. This ensures that while allowing users with higher permissions to use the system normally, it effectively prevents sensitive information from being leaked to users with lower permissions.

[0057] [Sixth Embodiment] In a specific embodiment, when all users have multiple permission levels, a preset information range corresponding to each permission level is obtained. Based on different combinations of multiple permission levels, an information display range is selected from the preset information range. Specifically, this includes: S421a. Obtain the weight value corresponding to the permission level and calculate the average weight of all users; S421b: Obtain the maximum weight value of all users, determine the level difference based on the difference between the average weight and the maximum weight value, and select the information display range based on the level difference.

[0058] In step S421a, each permission level corresponds to a weight value. The weight value quantifies the importance of the information access capabilities or operational permissions represented by different permission levels. The higher the permission level, the greater the corresponding weight value. The administrator's weight value is greater than that of the department manager, and the department manager's weight value is greater than that of the ordinary employee. In the current interaction session, the system identifies all users participating in the interaction and obtains the weight value corresponding to their respective permission levels. Subsequently, the system accumulates these weight values ​​and divides them by the total number of participating users to obtain the average weight of all users.

[0059] In step 421b, the system iterates through the weight values ​​corresponding to the permission levels of all users in the current interactive session and identifies the highest weight value, which is the maximum weight value for all users. The maximum weight value represents the upper limit of information that the user with the highest permission in the current user group can access, serving as the potential maximum boundary of the information display range. After obtaining the average weight and the maximum weight value, the system subtracts the average weight from the maximum weight value to obtain the level difference. The larger the difference, the more uneven the distribution of user permissions. The level difference is normalized and mapped to a preset range. The system then selects information from the preset information display range based on the calculated level difference.

[0060] Assuming an administrator's weight is 100, a department manager's weight is 70, a regular employee's weight is 30, and an external visitor's weight is 10, the system presets three information display ranges based on these three permission levels: Scope A (Widespread): Displays 90% of information visible to administrators, suitable for scenarios with highly centralized permissions.

[0061] Range B (Medium): Displays 60% of the information visible to administrators, suitable for scenarios where permissions differ but still need to accommodate higher permissions.

[0062] Range C (narrowest): Displays 30% of the information visible to administrators, suitable for scenarios with large differences in permissions or where strict confidentiality is required.

[0063] The rules for selecting the information display range based on the level difference are as follows: When the grade difference is less than 20, select range A; When 20 ≤ grade difference < 50, select range B; When the grade difference is ≥50, select range C.

[0064] For example, in a project presentation, one administrator, two department managers, and one regular employee are online simultaneously. The calculated average weight of all users is 67.5, the maximum weight of all users is 100, and the level difference is 32.5. According to the rules for selecting the information display range, the system will select range B, that is, display 60% of the information visible to the administrator. This allows the administrator and department managers to see most of the necessary information while hiding some sensitive content from the regular employee.

[0065] The setting of selecting the information display range based on the level difference effectively solves the problem that the efficiency and accuracy of selecting the information display range are insufficient when users with multiple permission levels interact at the same time, which may lead to inaccurate information display or poor user interaction experience. It maximizes the provision of effective information while ensuring information security and privacy.

[0066] [Seventh Embodiment] See Figure 3 In one specific embodiment, the user's operation record through the interactive interface is obtained, and the page position and page range are determined based on the operation record to see if they need to be adjusted. Specifically, this includes: S510. Obtain the search duration for each user clicking on the details page from the interactive interface; S520. Determine whether the user made an erroneous click based on the duration of the user's stay on the details page, and record the number of times each details page is mistakenly entered; S530. Obtain the total number of clicks for each details page, and calculate the total false entry ratio based on the total number of clicks and the number of false entries; S540. When the total false entry ratio is greater than or equal to the first threshold, the page position or page range of the details page needs to be adjusted. When the total false entry ratio is less than the first threshold, the page position and page range of the details page do not need to be adjusted.

[0067] In step S510, an excessively long search time indicates that the interface layout is unreasonable or the information presentation is unclear. Typically, a start timestamp is recorded when the user clicks to enter the details interface, and an end timestamp is recorded when the user has their first effective interaction on the details interface. The difference between the two times is the search time, or it can be defined as the time from when the user enters the details interface to when they leave the details interface without any effective interaction. This is used to measure whether the user has quickly found the information they need.

[0068] In step S520, the dwell time refers to the time a user continuously browses or interacts with a certain details page. If a user's dwell time on a certain details page is extremely short and no effective interaction is performed, it may indicate that the user did not intentionally enter the page, i.e., a mistaken click occurred, and this situation is counted as the number of accidental entries.

[0069] In steps S530 to S540, the total number of clicks refers to the total number of times a user enters a specific details page through various methods (such as navigation, search, recommendations, etc.). The total false entry ratio is calculated based on the total number of clicks and the number of false entries. This ratio is the ratio between the number of false entries and the total number of clicks; the higher the ratio, the more easily the page is mistakenly clicked or misunderstood by users. Adjustments are typically made by changing the display position of the details page within its parent page to make it more prominent, or by adjusting the size of its display area within the current page to highlight key points or reduce distractions. The first threshold can be determined based on experience or through testing, and is typically set to 0.1. When the total false entry ratio is less than the first threshold, it indicates that the user interaction experience of the page is good, and no intervention is required.

[0070] For example, when a user enters a "Sales Report" details page via voice command or by clicking the navigation menu, the system records information such as the user ID, entry time, and entry path. Simultaneously, the system starts a timer to record the time from when the user enters the "Sales Report" page to when they perform their first valid action, as the search duration. User M enters the "Sales Report" page at 9:00 AM, but at 9:02 AM, the system detects that User M clicked the back button without performing any valid action, and the dwell time was only 2 minutes. At this point, the system will determine that User M may have made a mistake based on a preset dwell time threshold and increment the count of erroneous entries to the "Sales Report" details page by 1. The system will continuously count the total number of clicks on the "Sales Report" details page.

[0071] Within a day, the "Sales Report" interface was clicked 100 times, with 15 clicks identified as accidental clicks. The system calculates an accidental click ratio of 0.15. Subsequently, the system compares this total accidental click ratio of 0.15 with a preset first threshold. If the first threshold is set to 0.10, the system determines that the page position or page area of ​​the "Sales Report" details interface needs adjustment. The system will automatically move the "Sales Report" position in the navigation menu downwards or reduce its display area on the main page to decrease the probability of accidental clicks.

[0072] By acquiring real-time records of user actions on the interactive interface and analyzing key metrics such as search duration, dwell time, and number of accidental entries, the system can promptly identify detail pages with interactive issues and make targeted adjustments to their page positions or scope. This transforms the interface layout from static to adaptive, responding to actual user behavior. This significantly improves the efficiency of users finding target information, reduces erroneous clicks, and thus optimizes the overall user experience.

[0073] [Eighth Embodiment] In one specific embodiment, when the total false entry ratio is greater than or equal to a first threshold, the page position or page range of the details page needs to be adjusted, specifically including: S541. Calculate the sub-mis-entry ratio of the details interface under different information display ranges, and correct the sub-mis-entry ratio based on historical access records to obtain the corrected ratio; S542. The information display range with a correction ratio greater than or equal to the first threshold is recorded as the abnormal display range, and the combination of the permission levels corresponding to the abnormal display range is recorded as the abnormal combination method. S543. Obtain the correction target after the user makes an erroneous click in the abnormal display range, and record the details interface corresponding to the correction target as the interface to be modified. S544. Determine the adjustment method of the interface to be modified based on the difference between the correction ratio and the first threshold.

[0074] In steps S541 to S544, since different information display ranges are provided based on the permission level for different user combinations, in order to ensure that other combination methods are not affected when the interface is modified, the rationality of each information display range is judged by the sub-mis-entry ratio.

[0075] In reality, the users of certain browser interfaces vary at different times. For computers that are publicly accessible, such as those in the company lobby used for demonstrations to visiting clients, browser usage is often more casual. Therefore, this data needs to be filtered out to obtain the true usage situation. Within the company, if a large number of new employees join, their initial use of the interface often contributes a lot of inaccurate data. This data needs to be corrected using historical access records.

[0076] When the system detects that a user has accidentally clicked on a details page within an abnormal display range, it immediately records the next details page the user visits. This page, which the user actively visited and which usually better reflects the user's true intention, is considered the user's "correction target" by the system, and its corresponding details page is marked as "page to be modified." Furthermore, the system can also combine the user's historical operation records, search keywords, or analyze the user's voice commands using natural language processing models to predict the user's true intention after the accidental click, and use the prediction result as the correction target.

[0077] The system calculates the difference between the correction ratio of the details page within the abnormal display range and a preset first threshold. If the difference is large, it indicates a more serious accidental click problem, and the system can directly adjust the page position of the interface to be modified or significantly expand its page range. If the difference is small, it can adjust the spacing of page elements, font size, color contrast, or add additional navigation prompts. When the difference is between (0, 0.1], the spacing of page elements is adjusted; when the difference is between (0.1, 0.2], the page position is adjusted; when the difference is greater than 0.2, the page range of the details page causing the accidental click within the abnormal display range is reduced.

[0078] For example, when the system detects that the total error rate of a financial report interface exceeds a preset first threshold, the system will calculate the correction ratio for that financial report interface under different information display ranges. The calculation shows that under the "Administrator + Finance Manager" permission combination, the correction ratio for this interface is 0.02. However, under the "Regular Employee + Sales Manager" permission combination, due to the hiding of some sensitive data causing changes in the interface layout, users may frequently click on the wrong area, resulting in a correction ratio of 0.15. If the first threshold is set to 0.10, then the system will mark the information display range under the "Regular Employee + Sales Manager" permission combination as an abnormal display range, and record the corresponding permission level combination as an abnormal combination. Next, the system will analyze the user's behavior after making an erroneous click under this abnormal display range. It finds that most users, after mistakenly clicking on a chart on the financial report interface, immediately return and click on the "Sales Data Analysis" interface. At this point, the "Sales Data Analysis" interface is identified as the user's correction target and marked as an interface to be modified. Finally, the system determines the adjustment method for the interface to be modified based on the difference between the correction ratio of 0.15 and the first threshold of 0.10.

[0079] The subsequent adjustments to the system include, but are not limited to: adjusting the page position by moving the entry point of the "Sales Data Analysis" interface to a higher position within the current abnormal display range, making it more prominent near the financial statement interface; adjusting the page range by reducing the page range of the area in the financial statement interface that caused the accidental clicks, and adding a shortcut to the "Sales Data Analysis" interface next to or below it; or embedding part of the preview content of the "Sales Data Analysis" interface into the blank area of ​​the financial statement interface, and allocating its page range in the financial statement interface proportionally based on the frequency of user clicks on the "Sales Data Analysis" interface.

[0080] By statistically analyzing the sub-mistaken entry ratio, the frequency of users making erroneous clicks or entering specific details pages under different information display ranges was obtained. Page position updates were correlated with different permission levels to ensure that subsequent page layout adjustments effectively guided users to achieve their intended goals, significantly reducing the user error rate in complex permission scenarios. The adjustment method was determined based on the difference between the sub-mistaken entry ratio and a first threshold, allowing the adjustment strategy to adaptively match the severity of the problem, achieving more intelligent and personalized interface optimization.

[0081] While the present invention has been disclosed above, it is not limited thereto. Any person skilled in the art can make various modifications and alterations without departing from the spirit and scope of the invention; therefore, the scope of protection of the present invention should be determined by the scope defined in the claims.

Claims

1. A method for upgrading browser-based AI intelligent interactive software, characterized in that, The upgrade method includes: Set up a simulated account for the target software, simulate user login operations using the simulated account, and extract the user interface of the target software; Each of the aforementioned operation interfaces is simplified, and the page position and page range of the operation interface in the details interface are determined according to the page tags; The details interface is annotated using a combination of manual annotation and automated recognition, and then the details interface is input into the interactive interface. Obtain the user's identity information, hide part of the information on the details interface based on the identity information, and adjust the remaining page range of the operation interface based on the hidden part; Obtain the operation record performed by the user through the interactive interface, and determine whether the page position and page range need to be adjusted based on the operation record.

2. The upgrade method according to claim 1, characterized in that, Set up a simulated account for the target software, simulate user login operations using the simulated account, and extract the user interface of the target software, specifically including: Different permission levels are set according to the application content of the target software; Prepare multiple accounts with different permission levels, and obtain the corresponding permission interfaces of the target software under different permission levels by simulating login operations.

3. The upgrade method according to claim 2, characterized in that, The step of simplifying each of the operation interfaces and determining the page position and page range of each operation interface in the details interface based on page tags specifically includes: The user's dwell time on each of the operation interfaces is obtained, and the usage ratio of each navigation module in the navigation menu is determined based on the usage record of the navigation menu within the dwell time. The page tags of the operation interface are determined based on the specific content of the navigation module, and the navigation modules are filtered based on the page tags and the usage ratio to obtain the retained modules; Remove the navigation menu from the operation interface to obtain the details interface, and determine the page position and page range of the retained module within the details interface.

4. The upgrade method according to claim 3, characterized in that, The steps of obtaining the user's identity information, hiding a portion of the information on the details interface based on the identity information, and adjusting the remaining page range of the operation interface based on the hidden portion specifically include: The identity information of each user is identified by image capture, and the information display range is determined according to the permission level corresponding to the identity information. The hidden portion is determined based on the information display range, and the page range of the operation interface is adjusted based on the hidden area of ​​the hidden portion.

5. The upgrade method according to claim 4, characterized in that, The step of identifying each user's identity information through image capture and determining the information display range based on the permission level corresponding to the identity information specifically includes: When all users have only one of the aforementioned permission levels, the information display range is directly determined based on the permission level. When all users have multiple permission levels, obtain the information preset range corresponding to each permission level, and select the information display range from the information preset range according to different combinations of the multiple permission levels.

6. The upgrade method according to claim 5, characterized in that, When all users have multiple permission levels, the step of obtaining a preset information range corresponding to each permission level, and selecting the information display range from the preset information range according to different combinations of the multiple permission levels, specifically includes: Obtain the weight value corresponding to the permission level, and calculate the average weight of all users; Obtain the maximum weight value for all users, determine the level difference based on the difference between the average weight and the maximum weight value, and select the information display range based on the level difference.

7. The upgrade method according to claim 6, characterized in that, The step of obtaining the user's operation records through the interactive interface, and determining whether the page position and page range need to be adjusted based on the operation records, specifically includes: The duration of each user's search click on the details page from the interactive interface is obtained; Determine whether the user made an erroneous click based on the length of time the user stays on the details page, and record the number of times each details page is mistakenly entered; Obtain the total number of clicks for each of the details pages, and calculate the total false entry ratio based on the total number of clicks and the number of false entries; When the total false entry ratio is greater than or equal to the first threshold, the page position or page range of the details interface needs to be adjusted. When the total false entry ratio is less than the first threshold, the page position and page range of the details interface do not need to be adjusted.

8. The upgrade method according to claim 7, characterized in that, When the total false entry ratio is greater than or equal to the first threshold, the page position or page range of the details interface needs to be adjusted, specifically including: The percentage of false entry into the details interface is calculated under different information display ranges. The percentage of false entry into the details interface is then corrected based on historical access records to obtain a corrected percentage. The information display range with a correction ratio greater than or equal to the first threshold is recorded as the abnormal display range, and the combination of the permission levels corresponding to the abnormal display range is recorded as the abnormal combination method. Obtain the correction target after the user makes the erroneous click within the abnormal display range, and record the details interface corresponding to the correction target as the interface to be modified. The adjustment method of the interface to be modified is determined based on the difference between the correction ratio and the first threshold.

Images