Game strategy intelligent layout generation and dynamic optimization method, device and storage medium

Abstract

The present application provides a method, device and storage medium for intelligent layout generation and dynamic optimization of game strategies. The method includes: when a layout trigger event is detected, generating a layout generation request; in response to the layout generation request, calling a knowledge graph engine to calculate the correlation of relational data, and screening a number of initial layout candidate templates from a preset layout template library based on the correlation result; inputting the initial layout candidate template into a low-code layout editing module, and generating a corresponding layout description file according to the mapping rules between page components and component parameters; inputting the layout description file as the current state into a reinforcement learning decision model, and outputting the target layout configuration in combination with user portrait data; sending the target layout configuration to the front-end rendering engine, and rendering the game strategy page according to the target layout configuration. The present application can realize the intelligent generation and real-time dynamic optimization of the game strategy page layout, and improve the layout personalization and user interaction experience.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, device and storage medium for intelligent layout generation and dynamic optimization of game strategies. Background Art

[0002] With the continuous expansion of the online and mobile gaming ecosystem, players' reliance on strategy guides has increased significantly. This is especially true when new games are released or versions are updated, as users are eager to learn about level-clearing strategies, character development paths, and the latest balance changes. However, existing strategy guide platforms mostly rely on keyword searches combined with manual layout: when new games or major versions are released, operators must manually create entries, assign columns, and edit page layouts. Subsequent patches or events require them to manually review and update relevant content. With the surge in the number of strategy guides and the frequency of updates, this highly manual model has become difficult to ensure responsiveness and content consistency, resulting in poor information timeliness, high error rates, and difficulty meeting players' demands for real-time and accuracy.

[0003] To improve the search experience, some platforms have begun introducing message queues for asynchronous backend task distribution or using low-code tools to assist with page construction. However, these solutions typically only address a single pain point: message queues are often only responsible for simple task triggering, lacking the ability to identify and respond to complex event sequences such as "version popularity, player behavior peaks," and so on. While low-code platforms have lowered the threshold for page editing, they still rely on manual judgment of component placement and are unable to automatically generate the optimal layout based on player profiles or entity semantic relationships. Knowledge graph technology is used to organize game entities and their relationships, but its application remains largely at the search or recommendation level, without deep integration with page layout or a dynamic optimization loop for strategy content.

[0004] Therefore, existing technologies generally have four shortcomings: 1. The creation efficiency of the encyclopedia structure is low when new games or major versions are launched, and entities and pages cannot be automatically initialized on demand; 2. Strategy content updates are delayed and error-prone, and there is a lack of a unified versioning and consistency verification mechanism; 3. The page layout is static, lacking intelligent recommendation and adaptive capabilities, and cannot dynamically adjust information weights based on players' real-time behavior; 4. The event-driven granularity is coarse, and it is impossible to achieve linked updates of content generation and layout for complex multi-dimensional events. Summary of the Invention

[0005] In view of this, the embodiments of the present application provide a method, device and storage medium for intelligent layout generation and dynamic optimization of game guides to solve the problem that the existing game guide page layout generation method is static and lacks personalization, and cannot automatically generate and dynamically optimize the page layout based on game entity relationships, player behavior and real-time events.

[0006] A first aspect of an embodiment of the present application provides a method for intelligent layout generation and dynamic optimization of a game guide, including: collecting contextual information related to a target page, the contextual information including relational data related to a game guide entity, and writing the relational data into a knowledge graph database; using a message queue or a complex event processing engine to listen to layout trigger events corresponding to a game identifier or a level identifier, and generating a layout generation request when a layout trigger event is detected; in response to the layout generation request, calling a knowledge graph engine to perform correlation calculation on the relational data, and screening a number of initial layout candidate templates from a preset layout template library based on the correlation result; inputting the initial layout candidate template into a low-code layout editing module, and generating a corresponding layout description file according to the mapping rules between page components and component parameters; inputting the layout description file as the current state into a reinforcement learning decision model, and outputting a target layout configuration in combination with user portrait data; sending the target layout configuration to a front-end rendering engine, and rendering the game guide page according to the target layout configuration.

[0007] The second aspect of the embodiment of the present application provides a game strategy intelligent layout generation and dynamic optimization device, including: an acquisition module, used to collect context information related to the target page, the context information includes relationship data related to the game strategy entity, and write the relationship data into a knowledge graph database; a monitoring module, used to use a message queue or a complex event processing engine to monitor layout trigger events corresponding to the game identifier or level identifier, and generate a layout generation request when a layout trigger event is detected; a screening module, used to call the knowledge graph engine to calculate the correlation of the relationship data in response to the layout generation request, and filter out several initial layout candidate templates from a preset layout template library based on the correlation result; a generation module, used to input the initial layout candidate template into a low-code layout editing module, and generate a corresponding layout description file according to the mapping rules between page components and component parameters; a decision module, used to input the layout description file as the current state into the reinforcement learning decision model, and output the target layout configuration in combination with user portrait data; a rendering module, used to send the target layout configuration to the front-end rendering engine, and render the game strategy page according to the target layout configuration.

[0008] According to a third aspect of an embodiment of the present application, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the steps of the above method are implemented when the processor executes the computer program.

[0009] According to a fourth aspect of the embodiments of the present application, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the steps of the above method are implemented.

[0010] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects:

[0011] By collecting context information related to the target page, the context information includes relational data related to the game strategy entity, and writing the relational data into the knowledge graph database; using a message queue or complex event processing engine to listen to layout trigger events corresponding to the game identifier or level identifier, when a layout trigger event is detected, a layout generation request is generated; in response to the layout generation request, the knowledge graph engine is called to calculate the correlation of the relational data, and based on the correlation result, several initial layout candidate templates are screened from the preset layout template library; the initial layout candidate template is input into the low-code layout editing module, and the corresponding layout description file is generated according to the mapping rules between page components and component parameters; the layout description file is input into the reinforcement learning decision model as the current state, and the target layout configuration is output in combination with the user portrait data; the target layout configuration is sent to the front-end rendering engine, and the game strategy page is rendered according to the target layout configuration. This application can realize the intelligent generation and real-time dynamic optimization of the game strategy page layout, and improve the layout personalization and user interaction experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0013] Figure 1 This is a flow chart of a method for intelligent layout generation and dynamic optimization of game strategies provided in an embodiment of the present application;

[0014] Figure 2 This is a schematic diagram of the structure of the device for intelligent layout generation and dynamic optimization of game strategies provided in an embodiment of the present application;

[0015] Figure 3 It is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0016] In the following description, specific details such as specific system structures and techniques are provided for purposes of illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obscuring the description of the present application with unnecessary detail.

[0017] To solve the problem of users being unable to quickly and accurately find the content they want in a vast amount of game guides, this application provides a low-code intelligent layout platform driven by knowledge graphs, allowing users to quickly and accurately find specific games, specific levels, and specific themes.

[0018] Specifically, the encyclopedia page layout is automatically recommended based on the relationship between strategy entities in the knowledge graph; when the user drags the layout on the low-code platform, the platform dynamically analyzes the user's intention and combines the entity association information in the knowledge graph in real time to intelligently recommend layout templates or components; the platform backend automatically drives the front-end rendering by generating a description configuration of the layout, providing a more intelligent and personalized layout generation experience.

[0019] In addition, this application also introduces a technology that combines message queues with event-driven dynamic content automation. This application goes beyond the simple "add game, create encyclopedia" function and further explores a more refined event-driven messaging mechanism to achieve real-time automated content replenishment and optimization driven by complex strategy events (such as version updates and hot event triggers).

[0020] Real-time content generation and synchronous update mechanism based on complex event processing (CEP) technology; message queue and AI-driven content automation expansion, revision and version management technology; intelligent error correction and strategy knowledge verification mechanism based on message queue.

[0021] In addition, this application also introduces a low-code platform and AI-driven layout automatic optimization technology. By introducing AI algorithms to dynamically analyze user preferences and strategy content characteristics, it can automatically recommend and optimize page layouts to significantly improve user experience.

[0022] Automatic layout generation and optimization algorithm based on reinforcement learning; dynamic layout intelligent adaptation technology driven by user personalized data; AI algorithm that automatically adjusts page layout and information weight based on content type and user usage habits.

[0023] The contents of the technical solution of this application are described in detail below with reference to the accompanying drawings and specific embodiments.

[0024] Figure 1 This is a flow chart of the method for intelligent layout generation and dynamic optimization of game strategies provided by the embodiment of the present application. Figure 1 As shown, the method for intelligent layout generation and dynamic optimization of game strategies may specifically include:

[0025] S101, collecting context information related to the target page, the context information including relationship data related to the game strategy entity, and writing the relationship data into the knowledge graph database;

[0026] S102, using a message queue or complex event processing engine to monitor a layout trigger event corresponding to a game identifier or a level identifier, and generating a layout generation request when a layout trigger event is detected;

[0027] S103, in response to the layout generation request, calling the knowledge graph engine to calculate the relevance of the relationship data, and screening a number of initial layout candidate templates from a preset layout template library based on the relevance results;

[0028] S104: Input the initial layout candidate template into the low-code layout editing module, and generate a corresponding layout description file according to the mapping rules between page components and component parameters;

[0029] S105, inputting the layout description file as the current state into the reinforcement learning decision model, and outputting the target layout configuration in combination with the user portrait data;

[0030] S106: Send the target layout configuration to the front-end rendering engine, and render the game strategy page according to the target layout configuration.

[0031] In some embodiments, collecting context information related to the target page includes:

[0032] Obtain information about game identifiers, level identifiers, and other related entities through multiple data sources related to the game;

[0033] Using graph construction technology, the acquired information is associated with game strategy entities to form entity nodes and relationships in the knowledge graph;

[0034] Based on relational data, contextual information related to target page rendering is extracted through graph query and calculation.

[0035] Specifically, in order to achieve intelligent generation and dynamic optimization of the layout of a specific game strategy page, the system first collects contextual information related to the target page to provide data support for subsequent layout generation and optimization. First, the system obtains necessary game-related entity information through multiple data sources related to the game. These data sources include but are not limited to the official data interface provided by the game developer, third-party content platforms or operational data backgrounds. The entity information obtained includes game identification, level identification, in-game characters, skill information, and player tasks and progress data. For example, the system may obtain the unique identification of game A (such as the game ID) from the official data interface of game A, as well as the identification of level A in the game (such as the level number or level name), basic attribute information of game character A, attribute information of skill A, etc.

[0036] Next, the system uses knowledge graph construction technology to perform semantic analysis and entity extraction on the collected entity information, and construct the entity nodes of the knowledge graph and the relationship between entities. Specifically, the system uses preset natural language processing technology and entity extraction methods to define the game identifiers, level identifiers, characters, skills, equipment, tasks and other information obtained above as different types of entity nodes in the graph, and establish semantic association relationships between entities. For example, character A is defined as a character entity node, skill A is defined as a skill entity node, and a relationship is established in which character A "possesses" skill A; further, an association relationship of the type "applicable to" or "recommended for use" is established between level A and a specific task or character node, thereby forming a complete game strategy knowledge graph.

[0037] Subsequently, in order to extract contextual information related to the rendering of the target strategy page, the system performs graph query and graph calculation operations based on the entity and relationship data constructed above to determine the weight and layout priority of each element in the target page. Specifically, the system uses predefined graph path query and graph algorithm analysis technology to calculate the association strength between each entity node in the graph and the target page, such as calculating the correlation between level A and skill A, the recommended association strength between character A and specific equipment, etc., and further extracts key information that is highly relevant to layout optimization, such as the skills, equipment, tasks and player focus that should be presented on the current strategy page. The extracted contextual information is eventually output to the next layout generation module to ensure that the generated strategy page layout can more accurately match the player's needs and realize the intelligent and personalized presentation of the page layout.

[0038] In some embodiments, a message queue or complex event processing engine is used to monitor layout trigger events corresponding to a game identifier or a level identifier. When a layout trigger event is detected, a layout generation request is generated, including:

[0039] Configure a message queue or complex event processing engine to monitor event streams related to game or level identifiers;

[0040] Define the rules for layout trigger events. When a layout trigger event that meets the rules is detected, parse the content of the layout trigger event to extract relevant game status, player behavior, and level information.

[0041] Based on the parsing results, a corresponding layout generation request is generated, and the layout generation request is sent to the layout generation module through a message queue or event-driven mechanism.

[0042] Specifically, in order to realize the dynamic layout generation of the game strategy page, the system uses a message queue or complex event processing engine to continuously monitor the game-related event stream, and when the preset conditions are met, it automatically generates and pushes a layout generation request to drive the subsequent page layout generation process.

[0043] By deploying a messaging middleware (such as Kafka or RabbitMQ) or a Complex Event Processing (CEP) engine that supports event-driven architectures, the system is configured to monitor various events associated with specific game or level identifiers. These events include, but are not limited to, a player entering level A in game A, completing task A, and triggering specific game behaviors (such as consecutive failures or an increase in hotspot queries). Event monitoring is implemented through an event topic and subscription mechanism, enabling the system to detect key behaviors occurring during game play in real time.

[0044] Furthermore, after event monitoring is complete, the system pre-sets layout triggering rules to determine which events should trigger page layout generation. These rules can be based on a single event (such as a player entering level A for the first time) or a complex event pattern (such as a player visiting character A's guide page multiple times and engaging in search behavior within a short period of time), combined with the context of the event. For example, if the CEP engine detects the event combination of "player visits exceeding a threshold + the current game level matches level A," it will be considered to have met the triggering conditions.

[0045] Furthermore, once a layout trigger event that meets the conditions is detected, the system will parse the event and extract key information carried in the event, including the game identifier (such as Game A), level identifier (such as Level A), the player's recent behavior trajectory (such as click path and search terms), and current game status data (such as character usage frequency and equipment selection). This information is structured and encapsulated into a layout generation request object.

[0046] Finally, the system sends the layout generation request to the backend layout generation module via a message queue or event-driven mechanism. Upon receiving the request, the module initiates subsequent processes such as layout template screening, configuration generation, and page rendering based on the extracted context information and subsequent graphs and AI components, thus achieving dynamic page layout generation.

[0047] Through the event monitoring and request generation mechanism in the above embodiment, the system can achieve real-time response based on the game running status and user behavior, ensuring the automation, intelligence and high timeliness of the page layout generation process.

[0048] In some embodiments, a knowledge graph engine is called to calculate the relevance of the relational data, and based on the relevance results, several initial layout candidate templates are screened from a preset layout template library, including:

[0049] Use the knowledge graph engine to perform graph query and analysis on the relevant relationship data of the target page and calculate the correlation between each entity node;

[0050] Based on the semantic relationship between entity nodes, the strength of association, and the characteristics of player behavior, a comprehensive association scoring model is generated to quantify the correlation between various elements.

[0051] Based on the calculated relevance score, a number of initial layout candidate templates whose matching degree with the context information of the target page is greater than a threshold are selected from the preset layout template library;

[0052] The initial layout candidate templates obtained by screening are sorted, and multiple initial layout candidate templates with higher sorting are selected as templates for the final layout.

[0053] Specifically, in order to achieve personalized recommendation and intelligent initialization of page layout, the system calls the knowledge graph engine to process the collected relational data to generate an initial layout candidate template that matches the target page.

[0054] First, based on the previously constructed game knowledge graph, the system invokes the graph query module to traverse the graph and calculate the relationship paths for the entity nodes associated with the current target page. Entity nodes include, but are not limited to, game identifiers (e.g., Game A), level identifiers (e.g., Level A), character entities (e.g., Character A), skill entities (e.g., Skill A), and equipment entities. Based on the graph structure between these entities, the system extracts direct or indirect relationship paths between them, such as "Game A → Level A → Recommended Character A → Skill A → Recommended Equipment."

[0055] Furthermore, based on the graph query, the system further combines the semantic relationship type between entities (such as "recommended," "restraint," and "adaptable") with the length of the associated path, using a graph computation algorithm (e.g., PageRank-based or path-weighted) to score the strength of the association between each pair of entities. This score is then integrated with the behavioral characteristics contained in the user profile (such as recently clicked characters, skills, and equipment) to construct a comprehensive relevance scoring model. This model allows the system to quantify the relevance between each guide entity and the core theme of the page, and serves as a basis for template selection.

[0056] During the screening phase, the system uses the aforementioned relevance score results as screening criteria to filter the preset layout template library. The layout template library stores several different types of page structure templates, which differ in component arrangement, information display focus, and interaction methods. For example, some templates are suitable for pages centered around character skills, while others focus on displaying task progress. The system calculates the degree of match between the contextual information of the target page and the content weight vector corresponding to each template, and selects several templates with a match degree above a set threshold as initial candidates.

[0057] Finally, the system ranks the initial candidate layout templates based on their comprehensive relevance scores and selects the top candidate templates as the input for subsequent layout optimization and reinforcement learning phases. These templates are further personalized to generate the final page layout presented to the user.

[0058] Through the above method, the system can accurately screen and initialize the layout of the initial structure of the strategy page based on the semantic relationship between entities and user behavior data, providing a high-quality template foundation for subsequent intelligent optimization.

[0059] In some embodiments, the initial layout candidate template is input into the low-code layout editing module, and a corresponding layout description file is generated according to the mapping rules between page components and component parameters, including:

[0060] Import the selected initial layout candidate template into the low-code layout editing module. The initial layout candidate template contains the basic framework of the page structure and preset component configuration;

[0061] According to the mapping rules between page components and component parameters, each component element in the initial layout candidate template is automatically identified and mapped, and corresponding parameters are configured for each component element;

[0062] Based on component parameters and mapping rules, a layout description file containing detailed information of each component is generated. The layout description file is used to guide the subsequent page rendering process and provide flexible configuration for page optimization.

[0063] Specifically, to enable rapid construction and flexible configuration of game guide page layouts, the system inputs initial layout candidate templates into a low-code layout editing module and automatically generates a rendering-ready layout description file based on component-parameter mapping rules. This process helps transform structured layout intent into specific page rendering configurations, thereby enhancing the automation and intelligence of page construction.

[0064] The system first imports one or more initial layout candidate templates, selected through knowledge graph analysis and matching, into the low-code layout editing module. Each initial layout candidate template is stored in a JSON or XML structure and defines the basic framework of the page, including the top navigation area, main content display area, side auxiliary area, interactive control area, etc. The template also includes several preset components, such as: text display component, image and text card component, video playback component, progress display bar, accordion panel, user comment box, etc.

[0065] Furthermore, after importing the template, the low-code layout editing module will automatically parse the structural definition in the template, and identify the various component types and their configuration requirements based on the preset mapping rules between page components and component parameters. For example, after the system recognizes a "skill display" component in the template, it will automatically extract skill-related information from the associated knowledge graph entity node and fill it into the parameter configuration of the component. Parameter configuration includes but is not limited to the following: component ID, data binding path (such as the graph node path for binding skill A), display fields (such as skill name, skill description, skill cooldown time), style attributes (such as font size, color, layout width), component interactive behavior (such as click jump, floating prompt), etc.

[0066] After identifying and binding all components, the system integrates the structural information and parameter configurations of each component into a standardized layout description file based on a unified layout rule template. This file, using a structured configuration format (such as JSON Schema), guides the front-end page rendering engine in dynamically constructing pages. It can also be used by subsequent reinforcement learning models or AI optimization modules to further adjust component arrangement and content display strategies.

[0067] In addition, the low-code editing module also provides a visual interactive interface. Based on the automatic generation of the initial layout description file, it supports developers or operators to perform fine-grained editing of the layout by dragging components, adjusting parameter values, modifying the structural order, etc. The system will update the corresponding layout description file in real time to ensure that the editing operation is synchronized with the underlying configuration.

[0068] Through the above implementation method, the system achieves efficient conversion from the initial layout template to the renderable page configuration, and provides a flexible and structured foundation for subsequent intelligent layout optimization and user personalized adaptation.

[0069] In some embodiments, the layout description file is input as the current state into the reinforcement learning decision model, and the target layout configuration is output in combination with the user profile data, including:

[0070] The layout description file and user profile data are input into the reinforcement learning decision model as input features to participate in the decision-making process;

[0071] Based on the layout description file and user portrait data, the reinforcement learning decision model outputs an optimized target layout configuration through training on historical interaction data. The target layout configuration is a page layout with higher user interaction efficiency and personalized features.

[0072] Specifically, to achieve personalized presentation and dynamic optimization of the game guide page layout, the system uses the generated layout description file and user profile data as input into a reinforcement learning decision model to generate an optimized target layout configuration. This process combines historical user behavior with the page structure state, ensuring that the final page layout is more aligned with user preferences and interaction habits.

[0073] The layout description file, structured configuration data generated by the system in the early stages, contains information about the type, display order, location, bound content fields, and style configuration of each component on the page. For example, a typical layout description file might describe a guide title component at the top of the page, skill graphics, level completion points, and recommended equipment in the middle, and user comments and version change information on the right. Each component has a corresponding priority, display fields, and interaction logic definitions.

[0074] At the same time, the system also collects and maintains profile data for each user. This data includes historical user behavior (such as browsing history, click frequency, dwell time, and favorite component types), user preferences (such as preference for graphic or table components, preference for folded or expanded structures), game progress (such as unlocked levels, frequently used characters, and frequently viewed topics), and device type (such as mobile or PC). This user profile data is organized into structured feature vectors, which serve as part of the reinforcement learning input.

[0075] During the layout optimization phase, the system combines layout description files and user profile features into a reinforcement learning decision model. This model is built based on historical training data, which includes feedback from users interacting with different layout configurations, such as page completion rates, component click-through rates, page dwell time, and page bounce rates. The system utilizes a reinforcement learning decision-making framework, treating each layout configuration as an "action," user behavior metrics as "rewards," and the layout description files and user profiles as the "state." During training, the model continuously adjusts its strategy through iterative learning, ensuring that the layout configuration output in a given state maximizes the user behavior reward.

[0076] For example, for user A, his portrait features show that he prefers character-based graphic and text information display, and often uses mobile devices. When generating a layout targeting this user, the reinforcement learning model may prioritize placing the "Character Strategy Card" component in the center area of ​​the main view of the page, and placing the "Skill Comparison Module" with a high interaction frequency above, and automatically adjust the layout to a vertical scrolling form to adapt to small-screen devices; for user B, his historical behavior is more inclined to view data-type content, so the system may prioritize the layout template that includes "Attribute Table" and "Equipment Value Comparison Column", and fold and hide the comment area to improve the page simplicity and information density.

[0077] Ultimately, the target layout configuration output by the reinforcement learning model expresses the display order, weighting, style hierarchy, and interaction priority of each component in a structured manner. This is then pushed to the front-end page rendering engine for dynamic layout generation. Simultaneously, this output is continuously integrated with real-world user interaction data for the next round of model strategy optimization, forming a dynamic closed loop.

[0078] Through the above implementation method, the system can achieve precise optimization of the game guide page layout at the user level based on user portraits and the initial structure of the page, effectively improving the page interaction efficiency and the degree of personalization of the user experience.

[0079] It should be noted that the reinforcement learning decision model in this embodiment plays a core role in the generation and optimization of game guide page layouts. Its primary task is to generate personalized page layout configurations based on layout description files and user profile data, and to continuously optimize its own strategies through user behavior feedback. The following is the structure of the model and its training process:

[0080] 1. Structure of Reinforcement Learning Decision Model

[0081] This application uses a policy optimization reinforcement learning model (such as the PPO model based on the Actor-Critic structure). Its overall structure can be divided into the following main modules:

[0082] 1) State Encoder

[0083] The input data can include the following features:

[0084] The structural features of the layout description file, such as the type, quantity, hierarchical structure, display area location, and component arrangement order of the page components;

[0085] User profile features, such as historical click preferences, component preferences, page browsing habits, user device type, game behavior tags, etc.

[0086] This module encodes and converts the above information to construct a high-dimensional vector as the state input of the model.

[0087] 2) Policy Network

[0088] Based on the current state, output optional layout action strategies. For example, select "adjust component order", "replace component template", or "add / subtract a certain type of component".

[0089] The output is a distribution vector, which represents the probability of selecting different layout actions in the current state.

[0090] 3) Value Network

[0091] Used to evaluate the expected long-term reward value of the current state (i.e., an estimate of the user interaction effect that the current layout may bring in the future).

[0092] This part is used to calculate the advantage function during the training process and assist in the stable update of the policy network.

[0093] 4) Action Space

[0094] Actions are defined as specific adjustments to the page layout, such as: adjusting the order of components (such as placing "skill cards" at the top); replacing component types (such as replacing "chart-based display" with "graphic and text components"); modifying styles (such as color themes and layout density); and adjusting module weight distribution (such as compressing the sidebar and expanding the main content area).

[0095] 2. Training Process of Reinforcement Learning Model

[0096] 1) Initialization training phase

[0097] The initial model is cold-start trained based on expert experience strategy or preset rule strategy; the initial data comes from the platform's existing user behavior records, and a large number of "state-action-feedback" triplets are constructed for model pre-training.

[0098] 2) Online training and strategy update

[0099] After the page is rendered, the user's actual interaction behavior data (clicks, dwell time, scroll depth, bounce rate, etc.) is collected; the system builds a reward function based on this behavior data, such as:

[0100] +1 point: The user clicked a key component in the layout.

[0101] +2 points: The page dwell time is greater than the threshold;

[0102] −1 point: The user quickly exits the page.

[0103] −2 points: The user did not interact with any component;

[0104] Furthermore, these reward signals are used to perform gradient updates through PPO or other policy optimization algorithms: the advantage function is calculated, the policy network (Actor) is updated to increase the probability of high-reward actions, and the value network (Critic) is simultaneously updated to correct the state value estimate.

[0105] 3) Incremental Iteration and Strategy Convergence

[0106] Each new user behavior enters the training cache as incremental data; the system adopts a periodic batch training mechanism to avoid overfitting the model to a single user; a policy stability threshold is set to avoid excessive model updates that affect user experience.

[0107] 4) Exploration and Exploitation Balance

[0108] The model uses an ε-greedy strategy or entropy regulation mechanism to maintain a certain probability of strategy exploration to avoid falling into local optimality; some high-active user pages use strategy experiments to test the effectiveness of new layout actions.

[0109] In summary, the reinforcement learning decision model in this application takes layout description files and user portraits as input, and user behavior feedback as reward signals. Through continuous learning, it continuously optimizes the page layout strategy, realizes accurate layout recommendation, dynamic optimization and personalized presentation, and significantly improves the user experience and information matching efficiency of the game strategy system.

[0110] In some embodiments, the method further comprises:

[0111] After the game guide page is rendered, the user's page interaction behavior data is collected and fed back to the reinforcement learning decision model so that the parameters of the reinforcement learning decision model can be iteratively updated using the page interaction behavior data.

[0112] Specifically, in order to achieve continuous optimization and adaptive evolution of the reinforcement learning decision model, after the game guide page is rendered and displayed to the user, the system will also collect the user's actual interactive behavior data on the page, and feed this behavior data back to the reinforcement learning model for subsequent iterative updates of parameters.

[0113] When a user visits a game guide page, the system embeds a behavior collection component on the front-end page to record multi-dimensional user interaction data on the page. This collected interaction data includes, but is not limited to, page dwell time, page scroll depth, number and order of clicks on each component, interaction frequency (e.g., number of comments, likes, and favorites), page exit behavior, and on-page navigation behavior (e.g., switching to another character page within the same game).

[0114] For example, the system might record a user's behavior on a strategy page, clicking on the "Skill Comparison Card" within five seconds of entering the page, then spending a considerable amount of time on the "Equipment Recommendation Bar," and ultimately jumping to the comments section to view other user feedback. The system then structures this raw behavior data, packages it with the current page layout configuration and user profile data, and uploads it to the behavioral data analysis module for processing.

[0115] The processed behavioral data is used to calculate rewards and update policies in the reinforcement learning model. The system evaluates the performance of the current layout configuration in response to a user's interaction based on a pre-set behavioral metric weighting model. For example, components with high click-through rates and long dwell times are given positive rewards, while components with high bounce rates or no interaction are given negative rewards. The system combines this positive and negative feedback to train and update the current layout decision-making strategy. Specifically, it uses reinforcement learning algorithmic frameworks such as policy gradient methods, proximal policy optimization (PPO), or deep Q-networks to iteratively adjust model parameters.

[0116] For example, in a certain round of interaction, if multiple users show a high page bounce rate and low click depth when using layout template A, the system will feed this behavior back to the model as a negative signal and reduce the probability of recommending this template in the next same context state; on the contrary, if another template B shows a high click rate and long browsing time among multiple similar users, the model will increase the reward value of this template, increase its priority, and prompt the model to be more inclined to output this configuration under similar conditions.

[0117] In addition, to improve the stability and controllability of training, the system will also batch clean and normalize the collected user behavior data, eliminate abnormal data (such as refreshes, invalid stays, etc.), and regularly perform incremental training on the model to ensure that the model has real-time adaptability to the evolution of user preferences, page content and interaction trends.

[0118] Through the above implementation method, the system can not only optimize the layout based on static user portraits, but also dynamically collect actual user behavior as feedback, and continuously iterate and update the layout strategy model, thereby forming a data-driven reinforcement learning closed loop and realizing the continuous intelligent evolution and optimization of the game strategy page layout.

[0119] The following are device embodiments of the present application, which can be used to implement the method embodiments of the present application. For details not disclosed in the device embodiments of the present application, please refer to the method embodiments of the present application.

[0120] Figure 2 This is a schematic diagram of the structure of the game strategy intelligent layout generation and dynamic optimization device provided by the embodiment of the present application. Figure 2 As shown, the game strategy intelligent layout generation and dynamic optimization device includes:

[0121] The acquisition module 201 is used to collect context information related to the target page, including relationship data related to the game strategy entity, and write the relationship data into the knowledge graph database;

[0122] A monitoring module 202 is configured to monitor layout triggering events corresponding to game identifiers or level identifiers using a message queue or a complex event processing engine, and generate a layout generation request when a layout triggering event is detected;

[0123] The screening module 203 is used to call the knowledge graph engine to calculate the relevance of the relationship data in response to the layout generation request, and to screen a number of initial layout candidate templates from the preset layout template library based on the relevance results;

[0124] The generation module 204 is used to input the initial layout candidate template into the low-code layout editing module and generate a corresponding layout description file according to the mapping rules between page components and component parameters;

[0125] Decision module 205, used to input the layout description file as the current state into the reinforcement learning decision model, and output the target layout configuration in combination with the user profile data;

[0126] The rendering module 206 is used to send the target layout configuration to the front-end rendering engine and render the game strategy page according to the target layout configuration.

[0127] In some embodiments, Figure 2 The acquisition module 201 obtains information about game identifiers, level identifiers and other related entities through multiple data sources related to the game; uses graph construction technology to associate the obtained information with game strategy entities to form entity nodes and relationships in the knowledge graph; based on relational data, through graph query and calculation, extracts context information related to the target page rendering.

[0128] In some embodiments, Figure 2The monitoring module 202 configures a message queue or a complex event processing engine to monitor event flows related to game identifiers or level identifiers; defines rules for layout triggering events, and when a layout triggering event that meets the rules is detected, parses the content of the layout triggering event to extract relevant game status, player behavior, and level information; based on the parsing results, generates a corresponding layout generation request, and sends the layout generation request to the layout generation module through a message queue or event-driven mechanism.

[0129] In some embodiments, Figure 2 The screening module 203 uses a knowledge graph engine to perform graph query and analysis on the relevant relationship data of the target page, and calculates the correlation between each entity node; based on the semantic relationship, correlation strength and player behavior characteristics between the entity nodes, a comprehensive correlation scoring model is generated to quantify the correlation between each element; based on the calculated correlation score, a number of initial layout candidate templates with a matching degree greater than a threshold with the context information of the target page are screened from the preset layout template library; the screened initial layout candidate templates are sorted, and the top-ranked initial layout candidate templates are selected as the templates for the final layout.

[0130] In some embodiments, Figure 2 The generation module 204 imports the screened initial layout candidate template into the low-code layout editing module. The initial layout candidate template contains the basic framework of the page structure and the preset component configuration; according to the mapping rules between page components and component parameters, it automatically identifies and maps the various component elements in the initial layout candidate template, and configures corresponding parameters for each component element; based on the component parameters and mapping rules, it generates a layout description file containing detailed information of each component. The layout description file is used to guide the subsequent page rendering process and provide flexible configuration for page optimization.

[0131] In some embodiments, Figure 2 The decision module 205 inputs the layout description file and the user portrait data into the reinforcement learning decision model as input features to participate in the decision-making process; based on the layout description file and the user portrait data, the reinforcement learning decision model outputs an optimized target layout configuration through training on historical interaction data, wherein the target layout configuration is a page layout with higher user interaction efficiency and personalized features.

[0132] In some embodiments, Figure 2 After the game guide page is rendered, the update module 207 collects the user's page interaction behavior data and feeds the page interaction behavior data back to the reinforcement learning decision model, so as to iteratively update the parameters of the reinforcement learning decision model using the page interaction behavior data.

[0133] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0134] Figure 3 Schematic diagram of the structure of the electronic device 3 provided in the embodiment of the present application. Figure 3 As shown, the electronic device 3 of this embodiment includes: a processor 301, a memory 302, and a computer program 303 stored in the memory 302 and executable on the processor 301. When the processor 301 executes the computer program 303, the steps of the above-mentioned method embodiments are implemented. Alternatively, when the processor 301 executes the computer program 303, the functions of the modules / units in the above-mentioned device embodiments are implemented.

[0135] For example, computer program 303 may be divided into one or more modules / units, which are stored in memory 302 and executed by processor 301 to implement the present application. One or more modules / units may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of computer program 303 in electronic device 3.

[0136] The electronic device 3 may be a desktop computer, a notebook, a PDA, a cloud server or other electronic device. The electronic device 3 may include but is not limited to a processor 301 and a memory 302. Those skilled in the art will understand that Figure 3 It is only an example of electronic device 3 and does not constitute a limitation of electronic device 3. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0137] The processor 301 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0138] Memory 302 can be an internal storage unit of electronic device 3, such as a hard drive or memory of electronic device 3. Memory 302 can also be an external storage device of electronic device 3, such as a plug-in hard drive, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash memory card, etc. Furthermore, memory 302 can include both an internal storage unit of electronic device 3 and an external storage device. Memory 302 is used to store computer programs and other programs and data required by the electronic device. Memory 302 can also be used to temporarily store data that has been output or is about to be output.

[0139] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0140] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0141] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0142] In the embodiments provided in this application, it should be understood that the disclosed apparatus / computer equipment and methods can be implemented in other ways. For example, the apparatus / computer equipment embodiments described above are merely schematic. For example, the division of modules or units is merely a logical function division. In actual implementation, there may be other division methods. Multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection of the apparatus or unit, which may be electrical, mechanical or other forms.

[0143] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0144] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0145] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application can implement all or part of the processes in the above-mentioned embodiment method by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, it can implement the steps of each of the above-mentioned method embodiments. The computer program may include computer program code, which may be in source code form, object code form, executable file, or some intermediate form. Computer-readable media may include: any entity or device capable of carrying computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium.

[0146] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A method for intelligent layout generation and dynamic optimization of game strategies, characterized in that: include: Collecting contextual information related to the target page, the contextual information including relationship data related to the game strategy entity, and writing the relationship data into the knowledge graph database; Using a message queue or a complex event processing engine to monitor a layout trigger event corresponding to a game identifier or a level identifier, and generating a layout generation request when the layout trigger event is detected; In response to the layout generation request, calling the knowledge graph engine to calculate the relevance of the relationship data, and screening a number of initial layout candidate templates from a preset layout template library based on the relevance result; Input the initial layout candidate template into the low-code layout editing module, and generate a corresponding layout description file according to the mapping rules between page components and component parameters; Input the layout description file as the current state into the reinforcement learning decision model, and output the target layout configuration in combination with the user portrait data; Sending the target layout configuration to the front-end rendering engine, and rendering the game strategy page according to the target layout configuration; The step of inputting the layout description file as the current state into the reinforcement learning decision model and outputting the target layout configuration in combination with the user portrait data includes: Input the layout description file and user portrait data into the reinforcement learning decision model as input features to participate in the decision process; Based on the layout description file and user portrait data, the reinforcement learning decision model outputs an optimized target layout configuration by training historical interaction data, wherein the target layout configuration is a page layout with higher user interaction efficiency and personalized features.

2. The method according to claim 1, characterized in that The collecting of context information related to the target page includes: Obtain information about game identifiers, level identifiers, and other related entities through multiple data sources related to the game; Using graph construction technology, the acquired information is associated with game strategy entities to form entity nodes and relationships in the knowledge graph; Based on relational data, contextual information related to target page rendering is extracted through graph query and calculation.

3. The method according to claim 1, characterized in that The method of using a message queue or a complex event processing engine to monitor a layout trigger event corresponding to a game identifier or a level identifier, and generating a layout generation request when the layout trigger event is detected, includes: Configure a message queue or complex event processing engine to monitor event streams related to game or level identifiers; Define the rules for layout trigger events. When a layout trigger event that meets the rules is detected, parse the content of the layout trigger event to extract relevant game status, player behavior, and level information. Based on the analysis result, a corresponding layout generation request is generated, and the layout generation request is sent to the layout generation module through a message queue or an event-driven mechanism.

4. The method according to claim 1, wherein The calling of the knowledge graph engine performs correlation calculation on the relationship data, and based on the correlation result, screens a number of initial layout candidate templates from a preset layout template library, including: Use the knowledge graph engine to perform graph query and analysis on the relevant relationship data of the target page and calculate the correlation between each entity node; Based on the semantic relationship between entity nodes, the strength of association, and the characteristics of player behavior, a comprehensive association scoring model is generated to quantify the correlation between various elements. Based on the calculated relevance score, a plurality of initial layout candidate templates having a matching degree with the context information of the target page greater than a threshold are screened out from a preset layout template library; The initial layout candidate templates obtained by screening are sorted, and multiple initial layout candidate templates with higher sorting are selected as templates for the final layout.

5. The method according to claim 1, wherein The initial layout candidate template is input into the low-code layout editing module, and a corresponding layout description file is generated according to the mapping rules between page components and component parameters, including: Import the selected initial layout candidate template into the low-code layout editing module, wherein the initial layout candidate template includes the basic framework of the page structure and preset component configuration; According to the mapping rules between page components and component parameters, each component element in the initial layout candidate template is automatically identified and mapped, and corresponding parameters are configured for each component element; Based on component parameters and mapping rules, a layout description file containing detailed information of each component is generated. The layout description file is used to guide the subsequent page rendering process and provide flexible configuration for page optimization.

6. The method according to claim 1, characterized in that The method further comprises: After the game strategy page is rendered, the user's page interaction behavior data is collected and fed back to the reinforcement learning decision model, so as to iteratively update the parameters of the reinforcement learning decision model using the page interaction behavior data.

7. A device for intelligent layout generation and dynamic optimization of game strategies, characterized in that: include: A collection module is used to collect context information related to the target page, wherein the context information includes relationship data related to the game strategy entity, and write the relationship data into the knowledge graph database; A monitoring module, configured to monitor a layout triggering event corresponding to a game identifier or a level identifier using a message queue or a complex event processing engine, and generate a layout generation request when the layout triggering event is detected; A screening module is configured to, in response to the layout generation request, call a knowledge graph engine to perform correlation calculation on the relationship data, and screen a plurality of initial layout candidate templates from a preset layout template library based on the correlation result; A generation module is used to input the initial layout candidate template into the low-code layout editing module, and generate a corresponding layout description file according to the mapping rules between page components and component parameters; A decision module, configured to input the layout description file as the current state into a reinforcement learning decision model, and output a target layout configuration in combination with user profile data; A rendering module, configured to send the target layout configuration to a front-end rendering engine and render a game strategy page according to the target layout configuration; Among them, the decision module is used to input the layout description file and user portrait data into the reinforcement learning decision model as input features to participate in the decision-making process; based on the layout description file and user portrait data, the reinforcement learning decision model outputs the optimized target layout configuration through training of historical interaction data, wherein the target layout configuration is a page layout with higher user interaction efficiency and personalized features.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images