In-game artificial intelligence-based auxiliary triggering method and system

Abstract

The present invention discloses an artificial intelligence-based auxiliary triggering method and system in a game, which relates to the technical field of data processing and includes: connecting to a user terminal to obtain player operation data; deeply mining and constructing a player operation profile; using the behavior pattern nodes of the player operation profile as the first one-way association nodes, the skill preference nodes of the player operation profile as the second one-way association nodes, and the level conquest nodes of the player operation profile as the third one-way association nodes to perform data mining and establish a player information tree; using a data analysis engine to establish an auxiliary operation network in combination with game scenario data; optimizing through the auxiliary operation network based on the player information tree to generate auxiliary operations and making a triggering decision under a positive incentive cycle. The present invention solves the technical problem that the prior art cannot generate personalized operation suggestions in real time according to player behaviors and the game environment, and achieves the technical effect of improving the player game experience and operation efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to an auxiliary triggering method and system based on artificial intelligence in a game. Background Art

[0002] Currently, many game systems rely on fixed rules to provide operation suggestions. These suggestions are usually based on static data of players, such as character attributes or task progress, and it is difficult to make personalized adjustments in real time according to the dynamic behaviors of players and the game environment. Especially in complex and ever-changing game scenarios, traditional methods cannot flexibly respond to the needs of players and game situations, resulting in insufficient real-time performance and accuracy of operation suggestions, and unable to optimize game strategies in a timely manner according to the performance of players. Therefore, there is a lack of a flexible game assistance system that can dynamically perceive the behaviors of players and the game environment and generate personalized and instant optimized operation suggestions, and the game experience of players is limited. Summary of the Invention

[0003] This application provides an auxiliary triggering method and system based on artificial intelligence in a game, which is used to solve the technical problem that the prior art cannot generate personalized operation suggestions in real time according to the behaviors of players and the game environment.

[0004] In view of the above problems, this application provides an auxiliary triggering method and system based on artificial intelligence in a game.

[0005] In the first aspect of this application, an auxiliary triggering method based on artificial intelligence in a game is provided. The method includes:

[0006] Connect to the user terminal to obtain player operation data; based on the player operation data, deeply mine and construct a player operation portrait, where the player operation portrait includes behavior pattern nodes, skill preference nodes, and level conquest nodes; use the behavior pattern nodes of the player operation portrait as the first unidirectional association nodes, use the skill preference nodes of the player operation portrait as the second unidirectional association nodes, and use the level conquest nodes of the player operation portrait as the third unidirectional association nodes to perform data mining and establish a player information tree; use a data analysis engine to establish an auxiliary operation network in combination with game scene data; based on the player information tree, optimize through the auxiliary operation network to generate auxiliary operations, and make a triggering decision under a positive incentive cycle.

[0007] In the second aspect of this application, an auxiliary triggering system based on artificial intelligence in a game is provided. The system includes:

[0008] A data acquisition module for connecting to a user terminal to acquire player operation data; an operation portrait construction module for deeply mining and constructing a player operation portrait based on the player operation data, where the player operation portrait includes a behavior pattern node, a skill preference node, and a level conquest node; a player information tree establishment module for using the behavior pattern node of the player operation portrait as a first unidirectional association node, the skill preference node of the player operation portrait as a second unidirectional association node, and the level conquest node of the player operation portrait as a third unidirectional association node to perform data mining and establish a player information tree; an auxiliary operation network establishment module for using a data analysis engine to establish an auxiliary operation network in combination with game scenario data; an auxiliary operation module for optimizing through the auxiliary operation network based on the player information tree to generate auxiliary operations and make trigger decisions under a positive incentive loop.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] This application connects to a user terminal to acquire player operation data; based on the player operation data, deeply mines and constructs a player operation portrait, where the player operation portrait includes a behavior pattern node, a skill preference node, and a level conquest node; uses the behavior pattern node of the player operation portrait as a first unidirectional association node, the skill preference node of the player operation portrait as a second unidirectional association node, and the level conquest node of the player operation portrait as a third unidirectional association node to perform data mining and establish a player information tree; uses a data analysis engine to establish an auxiliary operation network in combination with game scenario data; based on the player information tree, optimizes through the auxiliary operation network to generate auxiliary operations and make trigger decisions under a positive incentive loop. This invention solves the technical problem that the prior art cannot generate personalized operation suggestions in real time according to player behavior and game environment. By connecting to the user terminal to acquire player operation data, deeply mining and constructing a player operation portrait, extracting behavior patterns, skill preferences, and level conquest nodes, using data mining to establish a player information tree, and combining game scenario data, establishing an auxiliary operation network through a data analysis engine, finally generating personalized auxiliary operations based on the player information tree and the auxiliary operation network, and triggering an optimization decision through a positive incentive loop, achieving the technical effect of improving the player's game experience and operation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0012] Figure 1 Schematic flowchart of the in-game AI-based auxiliary trigger method provided by the embodiments of the present application;

[0013] Figure 2 Schematic structural diagram of the in-game AI-based auxiliary trigger system provided by the embodiments of the present application.

[0014] Explanation of reference numerals: data acquisition module 11, operation portrait construction module 12, player information tree establishment module 13, auxiliary operation network establishment module 14, auxiliary operation module 15. Detailed implementation manners

[0015] By providing an in-game AI-based auxiliary trigger method and system, the present application aims to solve the technical problem that the prior art cannot generate personalized operation suggestions in real time according to player behaviors and game environments. By connecting to the user terminal to obtain player operation data, deeply mining and constructing the player operation portrait, extracting behavior patterns, skill preferences, and level conquest nodes, using data mining to establish the player information tree, and combining with game scene data, an auxiliary operation network is established through a data analysis engine. Finally, based on the player information tree and the auxiliary operation network, personalized auxiliary operations are generated, and the optimization decision is triggered through a positive incentive loop, achieving the technical effects of improving the player's game experience and operation efficiency.

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts belong to the scope of protection of the present application.

[0017] It should be noted that any variations of the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or server that includes a series of steps or units does not necessarily have to be limited to those clearly listed steps or units, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products, or devices.

[0018] Embodiment 1, as Figure 1 shown, the present application provides an in-game AI-based auxiliary trigger method, and the method includes:

[0019] Step S100: Connect to the user terminal and obtain player operation data.

[0020] In the embodiment of the present application, first establish communication with the user terminal through a network connection, and perform data interaction with the player's device (such as a smart phone, PC, game console, etc.). This connection is achieved through a wireless network (such as Wi-Fi, Bluetooth, 4G / 5G, etc.) or a wired protocol (such as USB, Ethernet, etc.).

[0021] After establishing the connection, start collecting the player's operation data in the game. The player's operation data includes the selected path, residence time, number of attempts, and skill usage data. To obtain the selected path, the movement trajectory of the player in the game world is obtained through position tracking and path recording techniques. This can be achieved through the map engine, coordinate system in the game, and the position information of the player each time they move. The residence time records the duration of the player's stay in a certain area, level, or task through timestamps. Whenever the player enters a new area or level, the entry time is automatically recorded, and the departure time is recorded when the player leaves, so as to calculate the total duration of the stay. The number of attempts records the number of attempts of the player when challenging a specific level or task through an event trigger mechanism. Whenever the player fails to pass a level or complete a certain task successfully, a count is triggered, increasing the record of the number of attempts. When the player triggers or uses a certain skill during the game, the time point, skill type, and its trigger conditions of the skill usage are captured in a timely manner through an event listening mechanism. Through this process, the skill usage data is obtained. Among them, the skill usage data includes the skill type and the skill usage frequency.

[0022] Through the above process, the player operation data is obtained.

[0023] Step S200: Based on the player operation data, deeply mine and construct a player operation portrait, where the player operation portrait includes a behavior pattern node, a skill preference node, and a level conquest node.

[0024] In the embodiment of the present application, based on the player operation data, the K-means clustering algorithm is used to group and identify patterns in the player's behavior data. For example, by analyzing the selected path and residence time of the player, the players are divided into different behavior patterns. For example, some players may tend to choose the shortest path to quickly clear the level, while some players may like to explore every detail and stay in certain areas for a long time. In this way, the behavior pattern nodes of the players are identified.

[0025] Next, the association rule learning method is used to mine the skill preferences of players. By analyzing the skill data used by players in the game (such as skill types, usage frequencies), the skill selection patterns of players in battles or tasks are identified. By using the Apriori algorithm, the frequently used skill combinations of players are identified, and the usage scenarios of these skills are understood. For example, a certain player may prefer to use attack skills in offensive levels and defense skills in defensive levels. Through the association analysis of these skill usages, skill preference nodes are generated, reflecting the common skill types and skill combinations used by players in the game.

[0026] After that, the level conquest nodes are constructed through statistical analysis methods. According to the recorded number of attempts and passing times of players in each level, the frequency statistics of these data are carried out to evaluate the performance of players in each level. Specifically, first, according to the number of attempts and passing times of players in each level, the average number of attempts and passing times of each level are calculated. If the number of attempts of a certain level is significantly higher than that of other levels and the passing time is longer, it will be considered that this level is more difficult for players. Through these statistical data, it is identified which levels pose greater challenges to players. In order to convert the performance of players in these levels into level conquest nodes, first, the number of attempts of players in each level is compared with the average number of attempts of other levels. If the number of attempts of a certain level exceeds 1 / 2 of that of other levels, it is marked as a challenging level. Secondly, considering the passing time, if the time spent by players on a certain level is significantly higher than the average level, it indicates that this level has a higher difficulty for players. Through these data, the identification of the challenging level is combined with the actual performance of players (such as the number of attempts, passing time), and converted into level conquest nodes, which reflect the performance of players on specific levels.

[0027] Finally, the obtained behavior pattern nodes, skill preference nodes, and level conquest nodes are integrated to obtain the player operation portrait.

[0028] Step S300: Use the behavior pattern nodes of the player operation portrait as the first one-way association nodes, the skill preference nodes of the player operation portrait as the second one-way association nodes, and the level conquest nodes of the player operation portrait as the third one-way association nodes to perform data mining and establish a player information tree.

[0029] In the embodiment of the present application, first, the behavior pattern nodes are used as the first one-way association nodes for data mining. Through this node, key indicators related to player behavior are extracted, such as the frequency of movement paths, interaction behavior rules, and supply recovery behaviors, to construct the first search space. In this space, the behavior patterns of players are analyzed to identify the common action routes and decision-making rules of players in the game, and then the first feature branch is generated, which reflects the behavior tendencies and decision-making styles of players.

[0030] Next, the skill preference node is used as the second unidirectional association node for similar data mining. By analyzing the data of players' skill usage, indicators such as skill usage frequency and skill combination preference are extracted to establish a second search space. In this space, the skill usage patterns of players are mined to identify the types and combinations of skills that players tend to use, thereby generating a second feature branch, which depicts the battle strategies and skill usage habits of players.

[0031] After that, the level conquest node is used as the third unidirectional association node for data mining. By recording data such as the number of attempts, success rate, and passing time of players in each level, a third search space is established to analyze the performance of players in the levels. By calculating the average number of attempts and passing time of the levels, the challenges faced by players are identified, and a third feature branch is generated.

[0032] Finally, the first unidirectional association node and the first feature branch generated therefrom are fitted to capture the detailed features of players' behavior patterns. Similarly, fitting is performed on the second unidirectional association node and the second feature branch, and the third unidirectional association node and the third feature branch, respectively revealing the skill preferences and level conquest characteristics of players. Through the fitting of these feature branches, the deep data of the behavior pattern node, skill preference node, and level conquest node are associated, thereby accurately depicting players' behaviors and preferences in three dimensions. Finally, by integrating these nodes and their feature branches, a complete player information tree is established.

[0033] Furthermore, in the method provided by the application embodiment, using the behavior pattern node of the player operation portrait as the first unidirectional association node for data mining further includes:

[0034] Taking the behavior pattern node as the root node, obtaining the first unidirectional association indicators, the first unidirectional association indicators including the frequency of movement paths, the rules of interaction behaviors, and the supply and recovery behaviors; establishing a first search space through the frequency of movement paths, the rules of interaction behaviors, and the supply and recovery behaviors in the first unidirectional association indicators; in the first search space, using the behavior pattern node of the player operation portrait as the first unidirectional association node for data mining to generate a first feature branch.

[0035] In the embodiments of the present application, first, taking the behavior pattern node as the root node, the first unidirectional association index is obtained. Specifically, first, by analyzing the operation data of players in detail, the first unidirectional association index is obtained. To extract the frequency of movement paths, using the position tracking technology and the built-in map engine, the position information of players is collected in real time. When a player moves in the game, their current position and timestamp are recorded to generate continuous trajectory data. Subsequently, by statistically analyzing the trajectory data, the number of visits of the player on a specific path is calculated, thus forming the index of "frequency of movement paths", which reflects the exploration habits and path preferences of players. Next, to obtain the interaction behavior pattern, the event listening mechanism is used to monitor the interaction operations of players with the game environment, such as talking to NPCs, picking up items, or completing tasks. The time, type, and target object (such as NPC ID or item ID) of each interaction event are recorded, and by statistically analyzing the time distribution and frequency of these events, an interaction behavior pattern index is generated to reveal the tendencies of players in game tasks or resource utilization. In addition, to extract the supply recovery behavior, the supply operation behavior of players is recorded through the resource usage monitoring method, such as using potions or triggering recovery points. The trigger time, type, and quantity of each supply behavior are recorded, and through statistical analysis, the supply usage frequency and distribution of players are calculated to show their decision-making preferences in resource management.

[0036] After obtaining the above first unidirectional association indexes such as the frequency of movement paths, interaction behavior pattern, and supply recovery behavior, the first search space is then constructed. The search space, as a multi-dimensional analysis framework for integrating player behavior data, is constructed using multi-dimensional data modeling technology. First, the frequency of movement paths, interaction behavior pattern, and supply recovery behavior are defined as three independent dimensions of the search space. To ensure the proportional consistency between data, the data of each dimension is normalized. For example, the path frequency is normalized to the range of 0 to 100, the interaction pattern is statistically distributed according to time periods, and the supply recovery behavior is classified according to frequency or trigger time distribution. Then, each player operation record is mapped to a point in the three-dimensional space, and each point reflects the behavior characteristics of the player within a certain time period. For example, a point may represent the behavior characteristics of a player with a high path frequency, concentrated interaction behavior, and a low supply usage frequency. By observing the distribution of points, the aggregation areas and difference characteristics of player behavior can be intuitively understood, such as the frequently selected path areas or high supply usage hotspots.

[0037] After constructing the first search space, the Apriori algorithm is used to perform data mining and analyze the potential correlation between player behaviors. The Apriori algorithm is a classic association rule mining algorithm used to discover frequent item sets and association rules in a data set. At this stage, the behavior pattern node is used as the first unidirectional association node, and the potential correlation between the node and other behavior indicators (such as movement path frequency, interaction behavior rules, and supply recovery behavior) is analyzed. By using the Apriori algorithm, the relationship between the player's path frequency and supply use in a specific mission area is analyzed. For example, it may be found that the player has a high path frequency in a specific area and frequently uses supplies, which indicates that the player may conduct more intensive exploration or combat in this area. By mining the associations between these behaviors, the frequent patterns in the player's behavior pattern are revealed, thereby refining the player's behavior characteristics and further refining the player's behavior portrait.

[0038] Next, we count the frequency of occurrence of each behavioral feature through frequency analysis, and combine it with the mined association rules to further refine the player's behavior pattern. For example, a player may show a high path frequency and frequent use of supplies in multiple areas. Based on these data, we can find the player's behavior pattern and identify him as an "exploration player" or a "combat player who relies on supplies." These analysis results provide a basis for the subsequent generation of feature branches. Through further analysis of frequency statistics and association rules, we gradually extract the typical behavior patterns of players and map them into feature branches.

[0039] Finally, based on these analysis and mining results, the first feature branch is generated. The feature branch represents a behavior pattern or strategy of the player in the game. For example, if the player shows a high path frequency and high supply recovery behavior in the exploration area, a feature branch of "exploration, high supply dependence" will be generated. If the player tends to interact frequently and use less supplies, a feature branch of "high interaction, low supply use" may be generated.

[0040] Furthermore, in the method provided in the embodiment of the application, in the first search space, the behavior pattern node of the player's operation portrait is used as a first unidirectional association node for data mining to generate a first feature branch, and further includes:

[0041] In the first search space, a sliding window analysis is performed on the activity data pointed to by mining through the first unidirectional association node, and a first activity feature set is set; in the first search space, a sliding window analysis is performed on the resource usage ratio data pointed to by mining through the first unidirectional association node, and a first resource usage ratio feature set is set; with the behavior pattern node as the root node, a bidirectional traversal pointer is configured according to the first activity feature set and the first resource usage ratio feature set.

[0042] In the embodiment of the present application, first, the activity data is mined. This is the activity level of players at various time periods in the game, involving the frequency of task participation, the number of interactions, etc. To analyze the changes in activity more precisely, data processing is performed through the sliding window analysis method. Specifically, based on the distribution density and periodic characteristics of the behavior pattern nodes, the sliding window parameters are configured, including the time window length (e.g., 30 minutes or 1 hour) and the sliding step size (e.g., 5 minutes or 10 minutes). These parameters help to slide on the time series and analyze the activity data of players. By this method, the activity of players is calculated within each time window, and the first associated area is marked to reveal the volatility and periodicity of the activity. Through the sliding window analysis, the first activity feature set is generated. These feature sets reflect the active behavior patterns of players at different time periods and help to reveal the behavior changes and participation of players.

[0043] Next, the first unidirectional association node is used to mine the resource usage ratio data of players. The resource usage ratio data reflects the allocation and usage strategies of players for various resources (such as supplies, props, etc.) in the game. These data reveal the resource management characteristics of players in scenarios such as battles and explorations. To analyze the time-varying trend of resource usage, the sliding window analysis is also adopted. Based on the distribution characteristics of the resource usage data, the sliding window parameters (time window length and sliding step size) are configured for each player to ensure the analysis of the resource usage ratio within each time window. Through the sliding window, the resource usage frequency and ratio of players in that time period are statistically analyzed segment by segment on the time series. For example, calculate the proportion of potion usage or the frequency of prop consumption of players within a certain period of time. The concentrated areas of resource usage are also marked. These areas indicate the dependence of players on certain resources within a specific time period. Combining the time neighborhood, the data within these marked areas is aggregated to generate the first resource usage ratio feature set, comprehensively reflecting the management strategies and tendencies of players in resource usage.

[0044] After setting up the first activity feature set and the first resource usage ratio feature set, a bidirectional traversal pointer is further configured. This pointer has the behavior pattern node as the root node and is used for efficient navigation and query between the two feature sets. Specifically, a bidirectional traversal path is set for the behavior pattern node, enabling the pointer to start from the node and access the first activity feature set and the first resource usage ratio feature set either upward or downward. Through this configuration, it is possible to quickly search for and match the player's behavior characteristics in multi-dimensional data. For example, when the player's behavior pattern shows "high activity, high resource dependence", the bidirectional pointer can locate the corresponding activity and resource usage feature regions in the time series and extract the player's key behavior patterns. Through the efficient navigation of the pointer, an accurate mapping of the behavior pattern and resource management features is achieved, providing a solid data foundation for subsequent personalized recommendations and intelligent assisted decision-making.

[0045] Furthermore, in the method provided by the application embodiment, in the first search space, the first unidirectional association node is used to mine the activity data in the pointing for sliding window analysis to establish the first activity feature set, and it further includes:

[0046] Based on the distribution density and periodic characteristics of the behavior pattern node, sliding window parameters are configured, and the sliding window parameters include the time window length and the sliding step size; through the sliding window parameters, the player operation data is analyzed on the time series to mark the first association region; based on the first association region and multiple time neighborhoods, the first feature branch of the player information tree is set, and the first activity feature set is obtained by traversing upward or downward from any node with the first feature branch.

[0047] In the embodiment of the present application, first, the sliding window parameters are configured based on the distribution density and periodic characteristics of the behavior pattern node. Among them, the behavior pattern node reflects the player's typical behaviors in different time periods, such as task completion, interaction frequency, etc.; the distribution density represents the concentration degree of the player's behavior in the time dimension, and the periodic characteristics reveal the regularity and repeatability of the player's behavior. According to these characteristics, the time window length (such as 30 minutes or 1 hour) and the sliding step size (such as 5 minutes or 10 minutes) are configured, and these parameters help to accurately capture the changes and laws of the player's behavior in the time series.

[0048] Next, based on the configured sliding window parameters, analyze the player operation data on the time series. The time series is the player behavior data arranged in chronological order. Through the sliding window analysis method, capture the player's behavior characteristics period by period, especially the change in activity. Calculate the player's activity as a key indicator to measure the behavior intensity of the player within each time window. Specifically, the calculation of activity is based on the number of task completions, interaction frequency, exploration behavior, and resource usage frequency. The number of task completions is obtained by counting the number of tasks completed by the player during this period. The interaction frequency is obtained by recording the number of interaction behaviors of the player with other players, NPCs, or the game system, such as chatting, trading, or cooperation. The exploration behavior is obtained by counting the movement path or area access frequency of the player in the game map. The resource usage frequency is obtained by counting the number of supplies or items used by the player during this period. By summing up these data and calculating the average value, calculate the activity value within each time window to comprehensively measure the behavior intensity of the player during this period. Subsequently, through the dynamic comparison of the activity values, mark the time periods with activity values significantly higher than the average level, and define these time periods as the first associated area.

[0049] After marking the first associated area, combine multiple time neighborhoods (i.e., the associated data of the time periods before and after the sliding window) to further extract the player's behavior characteristics. Set the first feature branch of the player information tree for the player based on these data. The feature branch comprehensively reflects the player's active behavior pattern by analyzing the behavior patterns within the associated area and its time neighborhoods. For example, if the player frequently participates in tasks and uses a large amount of resources during a certain period, the feature branch will summarize these behavior patterns into the player's behavior characteristic nodes.

[0050] Finally, traverse up or down through the first feature branch to further analyze the player's behavior pattern. Upward traversal allows the system to trace back from a certain feature branch node to the player's historical behavior characteristics to understand the trend of the player's behavior changes; downward traversal allows the system to start from the feature branch node and predict the possible future behavior patterns or active periods of the player. This traversal method not only enhances the understanding of the player's behavior but also provides support for personalized recommendations and dynamic game adjustments.

[0051] Through the above steps, finally generate the first activity feature set. This feature set comprehensively reveals the player's active behavior patterns at different time periods through sliding window analysis and dynamic calculation of activity.

[0052] Furthermore, in the method provided by the application embodiment, setting the first feature branch of the player information tree based on the first associated area and multiple time neighborhoods further includes:

[0053] In the first associated region and multiple time neighborhoods, establish an activity change sequence; select the points that are among the top 3 in the activity change sequence and not on the same straight line as the leading solutions, and perform iterative search; use the behavior pattern node as the root node, and fit the first feature branch with reference to the real-time updated leading solutions.

[0054] In the embodiment of the present application, first, the operation data of the player is segmented in the time series through the sliding window analysis method to generate an activity change sequence. Specifically, using the sliding window technology, the data is segmented according to a fixed time window length (such as 30 minutes or 1 hour) and a sliding step length (such as 5 minutes or 10 minutes). In each time window, key behavior indicators of the player are extracted, including the number of task completions, the frequency of interaction behaviors, the number of resource usages, etc. Then, the activity value is obtained by weighted calculation of these indicators, where the weights of each indicator are the same. Subsequently, the activity values of each time window are arranged in chronological order to form an activity change sequence.

[0055] Next, select the top 3 points with the highest activity values from the activity change sequence as the initial leading solutions. These time periods represent the high-activity moments of the player's behavior. To ensure that the selected points can reflect the diversity of the behavior pattern, geometric constraints are used to detect whether these three points are collinear. If the three points are collinear, it means that their change trends are too single, and continue to select the fourth point or more points until a non-collinear point set is formed as the initial leading solution.

[0056] After the leading solutions are selected, the leading solutions are optimized through the iterative search method to further improve the fitting accuracy of the player's behavior pattern. In each iteration, the selection of points in the activity change sequence is re-evaluated, and a better set of points is screened by calculating the fitness value of the points. The fitness value is determined by calculating the distance and behavior characteristic differences between the point and other points in the sequence, and the goal is to find a set of points that can most reflect the player's active pattern. The iterative process stops after meeting the preset optimization conditions (such as the number of iterations reaching the upper limit), and finally the optimized leading solutions are determined.

[0057] After determining the leading solutions, the first feature branch of the player is obtained through curve fitting. Specifically, the points in the leading solutions are used as feature nodes, and a smooth curve is generated according to the distribution trend of these points. The fitted curve not only comprehensively reflects the change trend of the player's activity over time, but also highlights the key behavior characteristics in specific time periods. The weight of each node is determined by the activity value, and the node with a higher weight indicates that the player's behavior in this time period is more important. The generated feature branch intuitively depicts the player's behavior pattern and is stored as part of the player information tree for further analysis.

[0058] Finally, the first feature branch is obtained by fitting through the above method. This branch is an accurate expression of the player behavior pattern, fully reflecting the changing trend of player activity and key behavior characteristics, providing strong data support for subsequent personalized recommendation, intelligent auxiliary decision-making, and dynamic game optimization.

[0059] Furthermore, the method provided by the application embodiment further includes:

[0060] If the points in the top 3 positions in the activity change sequence are on the same straight line, it is defined as the first fitness straight line, and the point in the 4th position of the activity change sequence is determined; it is judged whether the point in the 4th position of the activity change sequence is on the first fitness straight line. If it is not, the points in the top 4 positions in the activity change sequence that are not on the same straight line are selected as the leader solution for iterative search.

[0061] In the embodiment of the present application, first, the top 3 points with the highest activity values are selected from the activity change sequence. These three points represent the high-activity periods in the player's behavior pattern and are a preliminary capture of the player's behavior fluctuations. These points serve as the preliminary leader solution, laying the foundation for subsequent optimization.

[0062] Then, geometric constraints are imposed on the selected top 3 points to check whether they are on the same straight line. If these points are collinear, it indicates that their behavior pattern changes are relatively single and cannot effectively reflect the diverse behavior patterns of players. To ensure that the point set can fully display the player's behavior changes, the vector cross product method is used to calculate whether these three points are collinear. If the three points are collinear, it is judged that these points lack sufficient variability, so more points need to be introduced for supplementation to enrich the diversity of the behavior pattern.

[0063] If it is found that the first three points are collinear, these three points are defined as the first fitness straight line. This straight line represents a relatively simple behavior pattern, lacking behavior fluctuations and complexity. To solve this problem, the point in the 4th position is introduced for supplementation, and the geometric method is used to check whether the 4th point is on the first fitness straight line. If the 4th point is not on the straight line, it indicates that this point has a large change in the behavior pattern, which can help optimize the leader solution and increase the diversity of the point set.

[0064] If the fourth point successfully breaks the collinearity of the first three points, these four points are selected as the new leading solution. The distribution of these points is more balanced, which can more comprehensively reflect the diversity and complexity of player behavior, providing broader support for subsequent behavioral pattern analysis. At this time, it is ready to enter the iterative search stage, and these points are optimized through fitness value calculation. In each iteration, the Euclidean distance between each point and other points is calculated to measure the distribution of points in the active state change sequence, ensuring that the selected points can cover different regions of the behavioral pattern. Points with larger fitness values represent more behavioral differences. By selecting these points for iterative search, the leading solution is gradually updated.

[0065] Through multiple rounds of iteration, the leading solution is optimized to more accurately represent the behavioral characteristics of players. When the change in fitness value tends to be stable or reaches the preset maximum number of iterations, the iteration stops, and the final leading solution is determined. The optimized leading solution can most reflect the diversity and volatility of player behavior and becomes the basis for generating the first feature branch.

[0066] Furthermore, the method provided by the application embodiment further includes:

[0067] Fitting the first feature branch, and at the same time, fitting the second feature branch in the second association area corresponding to the second one-way association node and fitting the third feature branch in the third association area corresponding to the third one-way association node; establishing a player information tree through the first one-way association node and the first feature branch, the second one-way association node and the second feature branch, and the third one-way association node and the third feature branch.

[0068] In the embodiment of the present application, first, based on obtaining the first one-way association node, an active state change sequence is generated through a sliding window analysis method, and the first three points with the highest activity values are selected as the preliminary leading solution. Among these points, a geometric constraint method is applied to check whether they are collinear. If they are collinear, more points are continuously selected until a non-collinear point set is obtained as the optimized initial leading solution. Then, the leading solution is optimized through an iterative search method. In each iteration, a better point set is screened by calculating the fitness value of the points. The calculation of the fitness value combines the Euclidean distance between the points and other points and the difference in behavioral characteristics, with the goal of selecting a point set that can most reflect the player's active pattern. After multiple rounds of iterative optimization, an optimized leading solution is finally determined. Then, through a curve fitting method, the first feature branch is generated based on the optimized leading solution, which accurately describes the behavioral fluctuations and active state changes of players during this time period.

[0069] After fitting the first feature branch, a similar fitting is performed within the second association region corresponding to the second unidirectional association node to generate the second feature branch. At this time, the curve fitting technique is continued to fit the data according to the behavior data of the second unidirectional association node by the weighted least squares method or the spline interpolation method to ensure that the fitting result can accurately depict the player's behavior characteristics in this dimension. Next, the third unidirectional association node and its corresponding third association region are fitted to generate the third feature branch, and this process also uses the curve fitting technique to ensure that the final feature branch can accurately reflect the change in the player's activity level in the third dimension.

[0070] After completing the fitting of the first feature branch, the second feature branch, and the third feature branch, the player information tree begins to be constructed. These feature branches are combined together through a tree structure. Specifically, the first unidirectional association node and its first feature branch serve as the root node of the tree, and the second unidirectional association node and its second feature branch, the third unidirectional association node and its third feature branch serve as the child nodes of the tree respectively, and finally a complete player information tree is constructed.

[0071] Step S400: Use the data analysis engine to establish an auxiliary operation network in combination with the game scenario data.

[0072] In the embodiment of the present application, first, the player's behavior data and game scenario data are collected from the game through the data acquisition interface. The player's behavior data includes task completion status, skill usage, resource consumption, etc., and the game scenario data includes enemy positions, task objectives, resource distributions, etc.

[0073] Next, statistical analysis methods are used to process the collected behavior data and scenario data, extract key features such as the player's activity level in the game, task completion speed, skill usage frequency, etc., and analyze them in combination with the features in the game scenario (such as the number of enemies, resource distribution). These extracted features provide data support for establishing the auxiliary operation network.

[0074] On this basis, a decision-making model is established using the rule engine. The rule engine generates operation suggestions for the player according to the preset rules and the player's current behavior characteristics. For example, if the player is performing a combat task, the rule engine may suggest using a certain skill or recommend that the player avoid specific enemies. The rule engine ensures that intelligent decisions can be made based on real-time behavior and scenario data and provide the most appropriate operation suggestions to the player.

[0075] Once the decision-making model is established, a real-time decision engine is utilized to generate operation suggestions in real time. Through this engine, the optimal operation path is dynamically calculated based on the player's current behavior and the game environment, and the player is recommended to select appropriate skills, task routes, or resource management strategies. These suggestions help the player make the best decisions based on real-time data, thereby improving the game efficiency and operation experience.

[0076] Finally, through a feedback mechanism, the feedback from the player on the operation suggestions is tracked, and it is ensured that each generated suggestion can better match the player's needs.

[0077] Through these steps, an auxiliary operation network is successfully established. This network is based on player behavior data and game scenario data, generates personalized operation suggestions through a rule engine and a real-time decision engine, helps players make the best decisions, and enhances their game experience.

[0078] Step S500: Based on the player information tree, optimize through the auxiliary operation network to generate auxiliary operations and make a trigger decision under a positive incentive loop.

[0079] It should be noted that triggering prompt information too early will reduce the fun of the player's exploration. In addition, novice players may be lost due to the initial difficulty, and the level of assistance provided should be adjusted accordingly, gradually transitioning from relatively basic prompts. Preferably, in the embodiment of the present application, based on the player information tree, the player's behavior patterns and game scenario data are optimized through the auxiliary operation network to generate personalized auxiliary operation suggestions, and the trigger decision is completed under a positive incentive loop. Specifically, first, the characteristic branches in the player information tree (such as behavior pattern nodes, skill preference nodes, and level conquest nodes) are used to analyze the player's behavior characteristics in different dimensions. These characteristics provide comprehensive data support for the auxiliary operation network, enabling it to more accurately adapt to the player's operation needs.

[0080] Through the auxiliary operation network, the current behavior state of the player and the game scenario data are combined in real time to dynamically generate the optimal auxiliary operation suggestions. For example, when the player is currently short of resources, it is recommended that they prioritize obtaining resources; when the player approaches a difficult game area, it may be recommended that they adopt a defensive or evasive strategy. These auxiliary operations are intelligent support for the player's behavior and can significantly improve the player's game performance.

[0081] While generating the auxiliary operations, a positive incentive loop mechanism is introduced, that is, by means of rewards or positive feedback, the player's acceptance of the auxiliary operations is strengthened. For example, when the player executes the operation according to the suggestion and successfully completes the task, positive feedback is given by rewarding points, unlocking additional content, or providing visual and audio cues. Through this incentive mechanism, the player's enthusiasm for executing the auxiliary operations is increased, and the player is further guided to gradually optimize their behavior patterns.

[0082] Through this closed-loop design, the optimized operation generation based on the player information tree is realized, and the triggering decision is completed with the support of the positive incentive loop. This process makes the auxiliary operation highly targeted and dynamically adaptable, effectively improving the player's gaming experience and operation efficiency.

[0083] In the embodiments of the present application, in summary, the embodiments of the present application have at least the following technical effects:

[0084] This application connects to the user terminal to obtain player operation data; based on the player operation data, deeply mines and constructs a player operation profile, where the player operation profile includes behavior pattern nodes, skill preference nodes, and level conquest nodes; uses the behavior pattern nodes of the player operation profile as the first one-way association nodes, the skill preference nodes of the player operation profile as the second one-way association nodes, and the level conquest nodes of the player operation profile as the third one-way association nodes to perform data mining and establish a player information tree; uses a data analysis engine to establish an auxiliary operation network in combination with game scenario data; based on the player information tree, optimizes through the auxiliary operation network to generate auxiliary operations, and makes a triggering decision under the positive incentive loop. The present invention solves the technical problem that the prior art cannot generate personalized operation suggestions in real time according to player behavior and game environment. By connecting to the user terminal to obtain player operation data, deeply mining and constructing a player operation profile, extracting behavior patterns, skill preferences, and level conquest nodes, using data mining to establish a player information tree, and combining game scenario data, establishing an auxiliary operation network through a data analysis engine, finally, based on the player information tree and the auxiliary operation network, generating personalized auxiliary operations, and triggering an optimization decision through a positive incentive loop, achieving the technical effect of improving the player's gaming experience and operation efficiency.

[0085] Embodiment 2, based on the same inventive concept as the in-game artificial intelligence-based auxiliary triggering method in the foregoing embodiment, as Figure 2 shown, the present application provides an in-game artificial intelligence-based auxiliary triggering system. The system in the embodiments of the present application and the method embodiments are based on the same inventive concept. Among them, the system includes:

[0086] A data acquisition module 11, which is used to connect to a user terminal and acquire player operation data; an operation portrait construction module 12, which is used to deeply mine and construct a player operation portrait based on the player operation data, and the player operation portrait includes behavior pattern nodes, skill preference nodes, and level conquest nodes; a player information tree establishment module 13, which is used to use the behavior pattern nodes of the player operation portrait as the first unidirectional association nodes, the skill preference nodes of the player operation portrait as the second unidirectional association nodes, and the level conquest nodes of the player operation portrait as the third unidirectional association nodes to perform data mining and establish a player information tree; an auxiliary operation network establishment module 14, which is used to use a data analysis engine and combine game scene data to establish an auxiliary operation network; an auxiliary operation module 15, which is used to optimize through the auxiliary operation network based on the player information tree, generate auxiliary operations, and make trigger decisions under a positive incentive cycle.

[0087] Further, the system is also used to implement the following functions:

[0088] Taking the behavior pattern node as the root node, obtaining the first unidirectional association indicators, where the first unidirectional association indicators include moving path frequencies, interaction behavior rules, and supply recovery behaviors; establishing a first search space through the moving path frequencies, interaction behavior rules, and supply recovery behaviors in the first unidirectional association indicators; in the first search space, using the behavior pattern nodes of the player operation portrait as the first unidirectional association nodes to perform data mining and generate a first feature branch.

[0089] Further, the system is also used to implement the following functions:

[0090] In the first search space, performing a sliding window analysis on the activity data mined by the first unidirectional association node to set a first activity feature set; in the first search space, performing a sliding window analysis on the resource usage ratio data mined by the first unidirectional association node to set a first resource usage ratio feature set; taking the behavior pattern node as the root node, and configuring a bidirectional traversal pointer according to the first activity feature set and the first resource usage ratio feature set.

[0091] Further, the system is also used to implement the following functions:

[0092] Configure the sliding window parameters based on the distribution density and periodic characteristics of the behavior pattern nodes, where the sliding window parameters include the time window length and the sliding step; analyze the player operation data on the time series through the sliding window parameters, and mark the first associated region; based on the first associated region and multiple time neighborhoods, set the first feature branch of the player information tree, and traverse upward or downward from any node with the first feature branch to obtain the first activity feature set.

[0093] Further, the system is also used to implement the following functions:

[0094] In the first associated region and multiple time neighborhoods, establish an activity change sequence; select the points in the top 3 positions and not on the same straight line in the activity change sequence as the leading solutions for iterative search; use the behavior pattern node as the root node, and fit the first feature branch in contrast to the real-time updated leading solutions.

[0095] Further, the system is also used to implement the following functions:

[0096] If the points in the top 3 positions in the activity change sequence are on the same straight line, define it as the first fitness straight line, and determine the point in the 4th position of the activity change sequence; judge whether the point in the 4th position of the activity change sequence is on the first fitness straight line. If not, select the points in the top 4 positions and not on the same straight line in the activity change sequence as the leading solutions for iterative search.

[0097] Further, the system is also used to implement the following functions:

[0098] Fit the first feature branch. At the same time, fit the second feature branch in the second associated region corresponding to the second one-way associated node and fit the third feature branch in the third associated region corresponding to the third one-way associated node; establish a player information tree through the first one-way associated node and the first feature branch, the second one-way associated node and the second feature branch, and the third one-way associated node and the third feature branch.

[0099] It should be noted that the above sequence of the embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above description of specific embodiments of this specification has been made. The processes depicted in the drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0100] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0101] This specification and the drawings are merely illustrative of the present application and are considered to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications therein.

Claims

1. An auxiliary triggering method based on artificial intelligence in the game, characterized in that: The method comprises: Connect to the user terminal to obtain player operation data; Based on the player operation data, deeply mine and build player operation profiles, which include behavior pattern nodes, skill preference nodes and level conquering nodes; The behavior pattern node of the player operation portrait is used as the first unidirectional association node, the skill preference node of the player operation portrait is used as the second unidirectional association node, and the level conquering node of the player operation portrait is used as the third unidirectional association node, and data mining is performed to establish a player information tree; Use the data analysis engine and combine it with game scene data to establish an auxiliary operation network; Based on the player information tree, the auxiliary operation network is optimized to generate auxiliary operations and make triggering decisions under a positive incentive cycle; The behavior pattern node of the player's operation portrait is used as a first unidirectional association node to perform data mining, and the method includes: Taking the behavior pattern node as a root node, obtaining a first unidirectional association index, wherein the first unidirectional association index includes a movement path frequency, an interaction behavior rule, and a supply recovery behavior; Establishing a first search space through the movement path frequency, interaction behavior rule and supply recovery behavior in the first unidirectional association index; In the first search space, the behavior pattern node of the player operation portrait is used as a first unidirectional association node for data mining to generate a first feature branch; In the first search space, the behavior pattern node of the player operation portrait is used as a first unidirectional association node for data mining to generate a first feature branch. The method further includes: In the first search space, the first unidirectional association node is used to perform a sliding window analysis on the activity data in the mining direction, and a first activity feature set is set; In the first search space, the first unidirectional association node is used to perform a sliding window analysis on the resource usage ratio data in the mining direction to set a first resource usage ratio feature set; Taking the behavior pattern node as a root node, a bidirectional traversal pointer is configured according to the first activity feature set and the first resource usage ratio feature set.

2. The in-game artificial intelligence-based auxiliary triggering method according to claim 1, characterized in that: In the first search space, the first unidirectional association node is used to mine the activity data in the pointing direction for sliding window analysis, and a first activity feature set is set, the method comprising: Based on the distribution density and periodicity characteristics of the behavior pattern nodes, configuring sliding window parameters, the sliding window parameters including time window length and sliding step length; Analyzing the player operation data in time series by using the sliding window parameter, and marking a first associated area; Based on the first associated area and multiple time neighborhoods, a first feature branch of the player information tree is set, and the first activity feature set is obtained by traversing upward or downward from any node using the first feature branch.

3. The in-game artificial intelligence-based auxiliary triggering method according to claim 2, characterized in that: Based on the first associated area and a plurality of time neighborhoods, setting a first feature branch of the player information tree, the method comprising: Establishing an activity change sequence in the first associated area and multiple time neighborhoods; Select the points that are in the top three of the activity change sequence and are not on the same straight line as the leading solution, and perform iterative search; The behavior pattern node is taken as a root node, and the first feature branch is fitted by comparing with the leadership solution updated in real time.

4. The in-game artificial intelligence-based auxiliary triggering method as claimed in claim 3, characterized in that: The method comprises: If the first three points in the activity change sequence are on the same straight line, it is defined as the first fitness line, and the point at the fourth position in the activity change sequence is determined; Determine whether the point in the 4th position of the activity change sequence is on the first fitness line. If not, select the points in the first 4 positions of the activity change sequence and not on the same line as the leading solution for iterative search.

5. The in-game artificial intelligence-based auxiliary triggering method according to claim 4, characterized in that: The method comprises: Fitting the first characteristic branch, and at the same time, fitting the second characteristic branch in the second association region corresponding to the second unidirectional association node, and fitting the third characteristic branch in the third association region corresponding to the third unidirectional association node; A player information tree is established through the first unidirectional association node and the first characteristic branch, the second unidirectional association node and the second characteristic branch, and the third unidirectional association node and the third characteristic branch.

6. The game's AI-based auxiliary trigger system is characterized by: The system is used to execute the method according to any one of claims 1 to 5, and the system comprises: A data acquisition module, the data acquisition module is used to connect to the user terminal and acquire player operation data, the player operation data includes the selected path, the stay time and the number of attempts; An operation profile building module, which is used to deeply mine and build a player operation profile based on the player operation data. The player operation profile includes a behavior pattern node, a skill preference node, and a level conquering node; A player information tree establishment module, the player information tree establishment module is used to use the behavior pattern node of the player operation portrait as a first unidirectional association node, the skill preference node of the player operation portrait as a second unidirectional association node, and the level conquering node of the player operation portrait as a third unidirectional association node, to perform data mining and establish a player information tree; An auxiliary operation network establishment module, the auxiliary operation network establishment module is used to use the data analysis engine and combine the game scene data to establish an auxiliary operation network; An auxiliary operation module, the auxiliary operation module is used to optimize through the auxiliary operation network based on the player information tree, generate auxiliary operations, and make trigger decisions under a positive incentive cycle.

Images