A method and system for recommending service information by integrating user tag data

Abstract

This invention discloses a service information recommendation method and system that integrates user tag data, comprising: retrieving users' historical access data to generate user tag data; performing preliminary analysis of the user tag data using an NLP model incorporating the XGBoost attention mechanism to generate initial weight ratios and service information links, and displaying the service information links prominently on the user interface; obtaining user feedback information on the service information links, analyzing and processing the feedback information based on the Q-learning framework, and adjusting the initial weight ratios; combining the optimized weight ratios and feedback information with the user tag data for optimization calculation, and displaying the optimized service information links. By dynamically adjusting recommendation weights and combining them with real-time user feedback, this method solves the accuracy problem caused by static weights in traditional recommendation systems, and has the advantages of dynamically optimizing recommendation weights and improving the accuracy of personalized recommendations.

Description

Technical Field

[0001] This invention relates to the field of information service technology, and specifically to a service information recommendation platform and system that integrates user tag data. Background Technology

[0002] In today's internet and mobile application ecosystem, conventional information service pushes are based on fixed information content and are displayed according to set trigger conditions. They combine user historical data to build user profiles or tag systems to carry out personalized information service pushes in a fixed pattern. Although this method achieves a certain degree of personalization, the lack of dynamic scheduling leads to a serious lag in information service updates. Recommended content is difficult to effectively optimize based on real-time user feedback, causing the system to fail to keenly capture short-term changes in user interests or immediate intentions. This results in poor information service recommendation effects and a decline in user experience. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art. This invention provides a service information recommendation method and system that integrates user tag data. By acquiring user feedback data on information service links in real time and optimizing the keyword weights of user tag data information, the invention achieves dynamic adjustment and optimization of information service recommendation effects, thereby improving the accuracy of information service recommendations and the user experience.

[0004] This invention provides a service information recommendation method that integrates user tag data, the recommendation method comprising:

[0005] S11: Detect user account login status, retrieve user historical access data, and generate user tag data;

[0006] S12: The NLP model with XGBoost attention mechanism is used to perform preliminary analysis on user tag data, generate initial weight ratios, calculate and generate service information links based on the initial weight ratios, and display the service information links on the top of the user interface.

[0007] S13: Obtain user operation feedback information on the service information link, analyze and process the operation feedback information based on the Q-learning framework, adjust the initial weight ratio, and generate an optimized weight ratio;

[0008] S14: Combine the optimization weight ratio and operation feedback information with user tag data to perform optimization calculations, and display the optimized service information link on the user interface.

[0009] Furthermore, step S11 includes:

[0010] Get the login request signal of the user account, retrieve the historical access data of the corresponding user account from the database based on the user account information, and set the historical access data as user tag data;

[0011] During the use of a user account, the access information of the user account is monitored in real time, and the user tag data is updated in real time based on the access information.

[0012] Furthermore, step S12 includes:

[0013] S121: Extract several keywords from user tag data based on an NLP semantic recognition model;

[0014] S122: Perform word frequency analysis on several keywords using the XGBoost attention mechanism algorithm, and set the initial weight ratio of several keywords based on the word frequency analysis data;

[0015] S123: Extract one or more keywords from a number of keywords based on the initial weight ratio, and generate service information links based on the extracted keywords.

[0016] Furthermore, step S121 includes:

[0017] Semantic analysis of user tag data is performed using a semantic model, and the data content of the user tag data is divided into several semantic categories.

[0018] The data content within the user tag data is categorized into several semantic categories.

[0019] Furthermore, the initial weight calculation formula for each keyword is as follows:

[0020] ;

[0021] in, The weight of the i-th keyword. The original score for the i-th keyword. The original score for the j-th keyword. This is for positive operations.

[0022] Furthermore, step S123 includes:

[0023] Based on the semantic categories of the extracted keywords, the keywords are arranged and combined according to their semantic categories to generate several service information links.

[0024] The priority of each service information link is calculated based on the weight ratio of the keywords within each link, and the links are sorted according to the priority.

[0025] Furthermore, step S13 includes:

[0026] Detect user actions related to service information links and generate feedback data.

[0027] The Q-learning framework is used to perform weight optimization calculations on the feedback information data and output the optimized weight ratios.

[0028] Furthermore, the formula for weight optimization calculation is as follows:

[0029] ;

[0030] in, The weight value for the (t+1)th iteration. The weight value at the t-th iteration. For learning rate, The feedback data at time t, Let be the feedback fluctuation at time t. It is the minimum value. For feedback impact coefficient, Let t be the number of clicks at time t. This represents the number of clicks at the initial moment.

[0031] Furthermore, step S14 includes:

[0032] Obtain the real-time access page layout of the user account, and adjust the display position of the service information links according to the access page layout.

[0033] This invention also provides a service information recommendation system that integrates user tag data, the recommendation system comprising:

[0034] Detection component: Used to detect user account login status, retrieve user historical access data, and generate user tag data;

[0035] Service Link Component: This component is used to perform preliminary analysis of user tag data using an NLP model that incorporates the XGBoost attention mechanism, generate initial weight ratios, calculate and generate service information links based on the initial weight ratios, and display the service information links at the top of the user interface.

[0036] Weight optimization component: used to obtain user operation feedback information on the service information link, analyze and process the operation feedback information based on the Q-learning framework, adjust the initial weight ratio, and generate an optimized weight ratio;

[0037] Link optimization component: Used to combine optimization weight ratios and operation feedback information with user tag data to perform optimization calculations and display the optimized service information links on the user interface.

[0038] This invention provides a service information recommendation method and system that integrates user tag data. By dynamically adjusting recommendation weights and combining them with real-time user feedback, it solves the problem of insufficient accuracy caused by static weights in traditional recommendation systems, and has the advantages of dynamically optimizing recommendation weights and improving the accuracy of personalized recommendations. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a flowchart of the service information recommendation method that integrates user tag data in an embodiment of the present invention;

[0041] Figure 2 This is a flowchart illustrating the specific operation of step S12 in an embodiment of the present invention;

[0042] Figure 3 This is a schematic diagram of a service information recommendation system that integrates user tag data in an embodiment of the present invention. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] Example 1:

[0045] Figure 1 This invention illustrates a flowchart of a service information recommendation method that integrates user tag data, as shown in an embodiment of the present invention. The service information recommendation method includes:

[0046] S11: Detect user account login status, retrieve user historical access data, and generate user tag data.

[0047] Detecting user account login status refers to real-time monitoring of user account login behavior, including: obtaining user account login request signals based on real-time system detection; retrieving historical access data of the corresponding user account from the database based on user account information and setting the historical access data as user tag data; scheduling the user account's historical access data within the system, including the user account's browsing history, search history, data reading history, etc., and using the historical access data as user tag data to provide data support for personalized service information recommendations for users.

[0048] After a user's login request signal is received, the user account information is matched with historical access data in the database using a unique identifier. This historical access data is then retrieved and used as the initial baseline for user tag data. During user account usage, access information content is monitored in real time, with a monitoring frequency that can be set to collect page dwell time, click events, or search keywords every 5 seconds. After the access information content is parsed into structured data, the user tag data is dynamically adjusted using an incremental update algorithm. For example, a sliding window mechanism can be used to retain behavioral data from the most recent 30 minutes and overwrite older data in the initial baseline that exceeds a set time threshold. The real-time updated user tag data is stored in an independent cache area, forming a hierarchical storage structure with the historical access data. The cache area data update cycle can be set to 10 seconds.

[0049] During the use of a user account, the access information of the user account is monitored in real time, and the user tag data is updated in real time based on the access information.

[0050] When a user account login request is triggered, the unique identifier in the user account information is used to query the database, retrieving the account's historical access records from the past 6 months as initial tag data. During subsequent user usage, the real-time detection module continuously monitors user behavior. For example, after semantic analysis of the currently viewed page content, keywords are extracted. When the frequency of a keyword exceeds a preset threshold, the weight of the corresponding interest tag in the user tag data is increased. When a user visits the same category more than 3 times consecutively, the priority of that category's tag is increased to 1.5 times the weight of the corresponding tag in the initial baseline data. The updated user tag data is synchronized to the recommendation model's input, enabling the model to calculate weights based on a dataset containing real-time behavioral features. By monitoring and updating user tag data in real time, changes in user behavior in the short term can be instantly fed back to the tag system for dynamic service information recommendation updates, thus prioritizing the display of relevant service information links in subsequent recommendation results.

[0051] S12: The NLP model with XGBoost attention mechanism is used to perform preliminary analysis on user tag data, generate initial weight ratios, calculate and generate service information links based on the initial weight ratios, and display the service information links at the top of the user interface.

[0052] Specifically, Figure 2 The following is a flowchart illustrating the specific operation of step S12 in an embodiment of the present invention. Step S12 includes:

[0053] S121: Extract several keywords from user tag data based on an NLP semantic recognition model;

[0054] Specifically, NLP semantic recognition models can use pre-trained models based on the Transformer architecture to achieve semantic parsing, and use the BERT model to perform contextual encoding on user tag data to extract semantically related entity words and behavioral feature words as candidate keywords.

[0055] Furthermore, step S121 includes:

[0056] Semantic analysis of user tag data is performed using a semantic model, and the data content of the user tag data is divided into several semantic categories.

[0057] The data content within the user tag data is categorized into several semantic categories.

[0058] Semantic categories are classified based on a contextual similarity threshold, specifically by calculating the cosine similarity value. Two tags with a vector similarity exceeding 0.85 are grouped into the same semantic category. The correspondence between data content and semantic categories is maintained through a dynamic mapping table, which is updated hourly based on user behavior data, retaining the most recent 30 days of historical mapping records as training samples. The number of semantic categories is controlled within the range of 5-8. When the number of newly added semantic categories exceeds the threshold, a category merging operation based on a density clustering algorithm is triggered, with the merging radius parameter set to 0.3.

[0059] Furthermore, after semantic recognition, the keywords in the user tag data are semantically categorized to classify the keywords in the user tag data and to combine and arrange several keywords to construct service information links that conform to the user tag data.

[0060] Specifically, when user tag data contains discrete keywords such as "sofa" and "dining table," the semantic model first extracts the context vector of each keyword through a bidirectional encoder. Then, a hierarchical clustering algorithm is used to categorize keywords with a vector distance less than 0.2 into the semantic category of "furniture selection." During this process, the original location information of the keywords is preserved in the metadata field of the mapping table for temporal correlation analysis in the subsequent weight calculation stage. In the service information link generation stage, keywords within the same semantic category are forcibly combined into related phrases. For example, "decoration style" and "furniture selection" form a combined feature under the semantic category of "home design." The combined weight is calculated through the interaction term in the XGBoost attention mechanism, with a weight adjustment coefficient set to 0.2-0.5. The resulting semantic units participate in reward value allocation as independent decision nodes in the Q-learning optimization stage, enabling click-through rate feedback to accurately trace back to feature optimization in a specific semantic dimension. This solution increases the semantic coverage of recommended links to over 92% while reducing the mismatch rate to below 8%.

[0061] S122: Perform word frequency analysis on several keywords using the XGBoost attention mechanism algorithm, and set the initial weight ratio of several keywords based on the word frequency analysis data;

[0062] Specifically, the original score for each keyword is generated by the XGBoost attention mechanism algorithm. By inputting the original score into an exponential function, all negative scores are converted into positive numbers approaching zero, while positive scores grow exponentially. For example, when the original scores of two keywords are -0.5 and 1.2 respectively, after exponential operation, they are converted to approximately 0.61 and 3.32 respectively, increasing the difference from 1.7 times to 5.4 times. Normalization is then performed, dividing the exponential score of each keyword by the sum of the exponential scores of all keywords, so that the weight values ​​are distributed in the range of 0 to 1 and sum to 1. This weight value is used to calculate the priority of service information links. The XGBoost attention mechanism simplifies both calculations through weight mapping and positive conversion.

[0063] The XGBoost attention mechanism algorithm can construct a word frequency weight mapping relationship through an integrated decision tree model, where each decision tree corresponds to a word frequency distribution pattern. The weight allocation strategy is iteratively optimized through gradient boosting. For example, word frequency statistics and user historical click behavior data are jointly input into the model to generate dynamic weight values ​​that reflect the strength of user preferences.

[0064] The initial weight calculation formula for each keyword is as follows:

[0065] ;

[0066] in, The weight of the i-th keyword. The original score for the i-th keyword. The original score for the j-th keyword. This is for positive operations.

[0067] In the keyword weight calculation process, the original score of each keyword is first obtained. For example, for five keywords A, B, C, D, and E, the original scores are 2.5, -1.2, 3.8, 0.7, and 1.9, respectively. Next, an exponential operation is performed on each original score, resulting in exp(2.5)=12.18, exp(-1.2)=0.30, exp(3.8)=44.70, exp(0.7)=2.01, and exp(1.9)=6.69. Then, the sum of all the exponential operation results is calculated, i.e., 12.18+0.30+44.70+2.01+6.69=65.88.

[0068] Finally, the exponential calculation result of each keyword is divided by the sum to obtain the final weights: A's weight is 12.18 / 65.88 = 0.185, B's weight is 0.30 / 65.88 = 0.005, C's weight is 44.70 / 65.88 = 0.679, D's weight is 2.01 / 65.88 = 0.031, and E's weight is 6.69 / 65.88 = 0.102. The exponential function performs a non-linear transformation on the original scores, converting scores within any real number range into positive numbers, avoiding interference from negative scores in the weight calculation. Simultaneously, the exponential calculation amplifies the differences between scores, significantly increasing the weight of high-scoring keywords. Normalization constrains the weights within the range of 0 to 1, ensuring the total weight sum is 1, forming a stable and comparable weight ratio. This calculation method enhances the rationality of weight allocation, provides a mathematical basis for prioritizing service information links, and makes the recommendation results more closely match the actual needs of users.

[0069] S123: Extract one or more keywords from a number of keywords based on the initial weight ratio, and generate service information links based on the extracted keywords.

[0070] Specifically, step S123 includes:

[0071] Based on the semantic categories of the extracted keywords, the keywords are arranged and combined according to their semantic categories to generate several service information links.

[0072] The priority of each service information link is calculated based on the weight ratio of the keywords within each link, and the links are sorted according to the priority.

[0073] When generating service information links, the keywords are first categorized according to their semantic categories. For example, keywords may fall into three semantic categories: "house," "renovation," and "home furnishing." The "house" category may include keywords such as "school district housing" and "subway housing"; the "renovation" category may include keywords such as "fully renovated" and "basically renovated"; and the "home furnishing" category may include keywords such as "smart home appliances" and "Nordic style."

[0074] Next, keywords within the same semantic category are arranged and combined to generate multiple service information links. For example, links such as "school district housing + fully furnished", "subway-adjacent housing + basic renovation", and "school district housing + smart home appliances" can be generated. This combination method ensures that the keywords within the links have a clear semantic connection.

[0075] Then, the link priority is calculated based on the weight ratio of keywords within each service information link. For example, if "school district housing" has a weight of 0.4 and "fully furnished" has a weight of 0.3, then the link "school district housing + fully furnished" might have a priority of 0.7. In contrast, if "subway housing" has a weight of 0.2 and "basically furnished" has a weight of 0.1, then the link "subway housing + basically furnished" would have a priority of 0.3.

[0076] Finally, all generated service information links are sorted according to the calculated priority. Links with higher priority are displayed first to the user. This ensures both the semantic coherence of the links and reflects the user's interests and preferences.

[0077] S13: Obtain user feedback information on the service information link, analyze and process the feedback information based on the Q-learning framework, adjust the initial weight ratio, and generate an optimized weight ratio.

[0078] Specifically, step S13 includes:

[0079] The system detects user actions on service information links and generates feedback data, including click data and dwell time. Click data includes valid click data (the service information link is clicked to view) and invalid click data (the service information link is clicked to close). Dwell time is the duration the service information link remains on the user interface, i.e., the time from when the service information link is generated until it is clicked to view or closed.

[0080] The Q-learning framework is used to perform weight optimization calculations on the feedback information data and output the optimized weight ratios.

[0081] The formula for weight optimization calculation is as follows:

[0082] ;

[0083] in, The weight value for the (t+1)th iteration. The weight value at the t-th iteration. For learning rate, The feedback data at time t, Let be the feedback fluctuation at time t. It is the minimum value. For feedback impact coefficient, Let t be the number of clicks at time t. This represents the number of clicks at the initial moment.

[0084] Furthermore, in this embodiment, taking click-through rate as an example, by introducing the Q-learning framework, the system can establish a dynamic mapping relationship between feedback data and weight adjustments, effectively responding to real-time changes in user interests. This adaptive learning mechanism enables the recommendation system to more accurately match users' dynamic intentions, improving the relevance of recommended content and user experience. Simultaneously, because Q-learning strategy optimization considers long-term cumulative rewards, the system not only focuses on immediate feedback when adjusting weights but also predicts and optimizes long-term effects, thereby gradually improving recommendation quality through continuous interaction. This dynamic and forward-looking optimization method significantly improves the recommendation system's response speed and adaptability to changes in user preferences.

[0085] S14: Combine the optimization weight ratio and operation feedback information with user tag data to perform optimization calculations, and display the optimized service information link on the user interface.

[0086] Obtain the real-time access page layout of the user account, and adjust the display position of the service information links according to the access page layout.

[0087] Specifically, taking an e-commerce website as an example, when a user visits a product details page, the document object model (DOM) tree structure of the current page is analyzed in real time to identify the layout distribution of the main content area, product parameter module, and user review area. By calculating the visual density values ​​of each area of ​​the page, a blank block with a width of 280 pixels is detected in the right sidebar area, and this area maintains a fixed position when scrolling. Based on the layout analysis results, service information links are dynamically inserted into the second-level container of this blank sidebar area. The link display format is set to a floating card component that matches the page style, with the card height adapting to the link content and moving synchronously with page scrolling. When the user switches to mobile browsing, a fixed navigation bar with a height of 80 pixels is detected at the bottom of the page. The service information links are adjusted to be displayed 10 pixels above the navigation bar, using a semi-transparent background to avoid obscuring the main content of the page.

[0088] Through the above technical solution, this application can dynamically select the optimal display area based on the actual rendering structure of the page. On desktop devices, it utilizes the fixed positioning feature of the sidebar to achieve continuous exposure for recommended content, while on mobile devices, it adapts to the bottom navigation interaction habits for precise targeting. This solution effectively solves the problem of insufficient adaptability between traditional recommendation positions and dynamic page layouts. It maintains the integrity of page content through an embedding method that minimizes visual interference, while simultaneously improving the visibility and click-through conversion efficiency of recommended content by utilizing the inherent interactive hotspots of the page.

[0089] This invention provides a service information recommendation method and system that integrates user tag data. By dynamically adjusting recommendation weights and combining them with real-time user feedback, it solves the problem of insufficient accuracy caused by static weights in traditional recommendation systems, and has the advantages of dynamically optimizing recommendation weights and improving the accuracy of personalized recommendations.

[0090] Example 2:

[0091] Figure 3 This diagram illustrates a service information recommendation system that integrates user tag data according to an embodiment of the present invention. The recommendation system includes:

[0092] Detection Component 10: Used to detect user account login status, retrieve user historical access data, and generate user tag data; user historical access data includes: user account browsing history, search history, data reading history, etc., and uses historical access data as user tag data to provide data support for personalized service information recommendations for users.

[0093] Service link component 20: It is used to perform preliminary analysis of user tag data by introducing an NLP model with XGBoost attention mechanism, generate initial weight ratios, calculate and generate service information links based on the initial weight ratios, and display the service information links at the top of the user interface.

[0094] Weight optimization component 30: used to obtain user operation feedback information on the service information link, analyze and process the operation feedback information based on the Q-learning framework, adjust the initial weight ratio, and generate an optimized weight ratio;

[0095] Link optimization component 40: Used to combine optimization weight ratios and operation feedback information with user tag data to perform optimization calculations and display the optimized service information links on the user interface.

[0096] The recommendation system includes a detection component 10, which retrieves historical access data in real time and generates dynamically updated user tag data upon user login. The service link component 20 uses the XGBoost attention mechanism and a natural language processing model to perform semantic analysis on the tag data, extract keywords, calculate initial weight ratios, and generate service information links that are then displayed prominently at the top of the user interface. The weight optimization component 30 uses the Q-learning framework to collect user click and dwell time feedback, constructing a state-action matrix to dynamically adjust the initial weights. The link optimization component 40, based on real-time page layout characteristics, adaptively adjusts the optimized service information links to blank areas or high-frequency interaction areas on the interface for display.

[0097] Through the above technical solution, this application solves the problems of data update lag and recommendation bias caused by insufficient collaboration between modules in traditional recommendation systems. The collaborative operation of the detection component and the service linking component ensures the timeliness of user tag data and the accuracy of initial recommendations; the weight optimization component dynamically adjusts weight parameters through a reinforcement learning mechanism, effectively capturing user interest shifts; the link optimization component optimizes display positions based on interface layout features, improving user operation efficiency. All components form a data closed loop, enabling adaptive iteration throughout the entire process of recommendation result generation and optimization.

[0098] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc.

[0099] Furthermore, the above provides a detailed description of a service information recommendation method and system that integrates user tag data provided by the embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A service information recommendation method that integrates user tag data, characterized in that, The recommendation method includes: S11: Detect user account login status, retrieve user historical access data, and generate user tag data; S12: The NLP model with XGBoost attention mechanism is used to perform preliminary analysis on user tag data, generate initial weight ratios, calculate and generate service information links based on the initial weight ratios, and display the service information links on the top of the user interface. S13: Obtain user operation feedback information on the service information link, analyze and process the operation feedback information based on the Q-learning framework, adjust the initial weight ratio, and generate an optimized weight ratio; S14: Combine the optimization weight ratio and operation feedback information with user tag data to perform optimization calculations, and display the optimized service information link on the user interface; Step S12 includes: S121: Extract several keywords from user tag data based on an NLP semantic recognition model; S122: Perform word frequency analysis on several keywords using the XGBoost attention mechanism algorithm, and set the initial weight ratio of several keywords based on the word frequency analysis data; S123: Extract one or more keywords from a number of keywords based on the initial weight ratio, and generate service information links based on the extracted keywords; Step S121 includes: Semantic analysis of user tag data is performed using a semantic model, and the data content of the user tag data is divided into several semantic categories. The data content within the user tag data is categorized into several semantic categories; Step S123 includes: Based on the semantic categories of the extracted keywords, the keywords are arranged and combined according to their semantic categories to generate several service information links. The priority of each service information link is calculated based on the weight ratio of the keywords within each link, and the links are sorted according to the priority.

2. The service information recommendation method integrating user tag data according to claim 1, characterized in that, Step S11 includes: Get the login request signal of the user account, retrieve the historical access data of the corresponding user account from the database based on the user account information, and set the historical access data as user tag data; During the use of a user account, the access information of the user account is monitored in real time, and the user tag data is updated in real time based on the access information.

3. The service information recommendation method integrating user tag data according to claim 1, characterized in that, The initial weight calculation formula for each keyword is as follows: ； in, The weight of the i-th keyword. The original score for the i-th keyword. The original score for the j-th keyword. This is for positive operations.

4. The service information recommendation method integrating user tag data according to claim 1, characterized in that, Step S13 includes: Detect user actions related to service information links and generate feedback data. The Q-learning framework is used to perform weight optimization calculations on the feedback information data and output the optimized weight ratios.

5. The service information recommendation method integrating user tag data according to claim 4, characterized in that, The formula for weight optimization calculation is as follows: ； in, The weight value for the (t+1)th iteration. The weight value at the t-th iteration. For learning rate, The feedback data at time t, Let be the feedback fluctuation at time t. It is the minimum value. For feedback impact coefficient, Let t be the number of clicks at time t. This represents the number of clicks at the initial moment.

6. The service information recommendation method integrating user tag data according to claim 1, characterized in that, Step S14 includes: Obtain the real-time access page layout of the user account, and adjust the display position of the service information links according to the access page layout.

7. A service information recommendation system that integrates user tag data, characterized in that, The recommendation system includes: Detection component: Used to detect user account login status, retrieve user historical access data, and generate user tag data; Service Link Component: This component is used to perform preliminary analysis of user tag data using an NLP model that incorporates the XGBoost attention mechanism, generate initial weight ratios, calculate and generate service information links based on the initial weight ratios, and display the service information links at the top of the user interface. Based on the NLP semantic recognition model, several keywords are extracted from user tag data; semantic analysis of user tag data is performed through the semantic model, and several semantic categories are divided according to the data content of user tag data; The data content within the user tag data is categorized into several semantic categories; The XGBoost attention mechanism algorithm is used to perform word frequency analysis on several keywords, and the initial weight ratio of several keywords is set according to the word frequency analysis data. Based on the initial weight ratio, extract one or more keywords from several keywords, and generate service information links based on the extracted keywords; Based on the semantic categories of the extracted keywords, the keywords are arranged and combined according to their semantic categories to generate several service information links. The priority of each service information link is calculated based on the weight ratio of the keywords within each link, and several service information links are sorted according to the priority. Weight optimization component: used to obtain user operation feedback information on the service information link, analyze and process the operation feedback information based on the Q-learning framework, adjust the initial weight ratio, and generate an optimized weight ratio; Link optimization component: Used to combine optimization weight ratios and operation feedback information with user tag data to perform optimization calculations and display the optimized service information links on the user interface.

Images