Goods price abnormity detection method of game and terminal
By training a game price prediction model and using cluster analysis to detect abnormal game prices, the problem of price fluctuations caused by malicious transactions was solved, thus achieving fairness and security in game transactions.
Patent Information
- Application Number
- CN202410776664.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-17
- Publication Date
- 2025-12-19
AI Technical Summary
Malicious trading in the game causes price fluctuations, affecting the fairness and security of game transactions. Existing technology is insufficient to effectively detect and adjust abnormal prices.
By collecting attribute information of game items, supply and demand data, player transaction records, and game behavior data, a game price prediction model is trained to identify price anomalies and make automatic adjustments. Cluster analysis is then used to detect abnormal transactions.
This improves the comprehensiveness and accuracy of abnormal price detection, ensuring fairness and security in the game transaction process and reducing the impact of malicious trading behavior.
Smart Images
Figure CN121155136A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of game data processing, and particularly relates to a game price abnormality detection method and a terminal. BACKGROUND
[0002] At present, there are malicious transaction behaviors in game auction houses. For example, game studios usually control a large number of characters to perform in-game activities such as monster killing and collecting by means of multi-account opening, using automated scripts (robots), etc., so as to efficiently produce game currency. Such behaviors can cause a substantial increase in the supply of in-game currency, which may trigger inflation. Because these game studios produce a large amount of currency and put it into the market, it may cause a sharp fluctuation in the price of in-game items. Some originally rare items may depreciate due to the increase in the supply of currency, while the prices of some commonly used items may skyrocket due to the flooding of currency. Therefore, the operation of these game studios can cause abnormality in the economic system of the game.
[0003] Therefore, the above-mentioned malicious transaction behaviors can adversely affect the fairness and normal operation of the entire game auction house. It is difficult to ensure the fairness and security of the game transaction process when malicious transaction behaviors occur, so it is urgent to reasonably detect the abnormality of game prices. SUMMARY
[0004] The technical problem to be solved by the present application is to provide a game price abnormality detection method and terminal, which can improve the comprehensiveness of price abnormality detection when malicious transaction behaviors occur, so as to ensure the fairness and security of the game transaction process.
[0005] To solve the above technical problems, the technical scheme adopted by the present application is:
[0006] A game price abnormality detection method, comprising the steps of:
[0007] S1, collecting data of game items, and determining the occurrence probability of the game items according to attribute information of the game items;
[0008] S2, training and establishing a game price prediction model according to collected game events, supply and demand data of game items, transaction record data of players and game behavior data;
[0009] S3, predicting the price data of the game items according to the game price prediction model, identifying price abnormality problems based on the price data prediction results, and if corresponding price abnormality problems are identified, determining that the price data of the game items is abnormal, and adjusting the price data and occurrence probability of the game items.
[0010] In order to solve the above technical problems, another technical solution adopted by the present application is:
[0011] A game price anomaly detection terminal, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements each step of the game price anomaly detection method described above when executing the computer program.
[0012] The present application has the beneficial effect that after collecting the data of the game items, the occurrence probability of the game items is determined according to the attribute information of the game items; the game events, the supply and demand data volume of the game items, the transaction record data and the game behavior data of the players in the game are collected, and the game price prediction model is trained and established based on this, so that a more comprehensive and accurate prediction model can be constructed through multi-dimensional data analysis; then, the price data of the game items is predicted according to the game price prediction model, and the price anomaly problem is identified based on the price data prediction result, if the anomaly is identified, it is determined that the price data of the game items is abnormal, and the price and occurrence probability of the game items are adjusted. In this way, it can be identified whether the item price is abnormal after the malicious transaction behavior occurs, the comprehensiveness of the price anomaly detection is improved, and the price data and occurrence probability of the game items can be automatically adjusted to ensure the fairness and safety of the game transaction process. BRIEF DESCRIPTION OF DRAWINGS
[0013] Figure 1 A flowchart of a game price anomaly detection method according to an embodiment of the present application;
[0014] Figure 2 A schematic diagram of a game price anomaly detection terminal according to an embodiment of the present application;
[0015] Figure 3 A specific step flowchart of a game price anomaly detection method according to an embodiment of the present application.
[0016] REFERENCE NUMERALS:
[0017] 1. A game price anomaly detection terminal; 2. a memory; 3. a processor. DETAILED DESCRIPTION
[0018] In order to explain the technical content, the purposes and effects of the present application in detail, the following will be described in conjunction with the embodiments and the accompanying drawings.
[0019] Please refer to Figure 1 The embodiment of the present application provides a game price anomaly detection method, comprising the steps of:
[0020] S1, data collection is performed on game items, and occurrence probability of the game items is determined according to attribute information of the game items;
[0021] S2, a game price prediction model is trained and established according to collected game events in the game, supply-demand data volume of the game items, transaction record data and game behavior data of the players;
[0022] S3, price data of the game items are predicted according to the game price prediction model, and price anomaly problems are identified based on the price data prediction results, if corresponding price anomaly problems are identified, it is determined that the price data of the game items are abnormal, and the price data and occurrence probability of the game items are adjusted.
[0023] From the above description, it can be known that the beneficial effects of the present application are that after the data of the game items are collected, the occurrence probability of the game items is determined according to the attribute information of the game items, game events in the game, supply-demand data volume of the game items, transaction record data and game behavior data of the players are collected, and a game price prediction model is trained and established based on this, so that a more comprehensive and accurate prediction model can be constructed through multi-dimensional data analysis, then the price data of the game items are predicted according to the game price prediction model, and price anomaly problems are identified based on the price data prediction results, if abnormalities are identified, it is determined that the price data of the game items are abnormal, and the price and occurrence probability of the game items are adjusted. In this way, whether the item price is abnormal after the malicious transaction behavior occurs can be identified, the comprehensiveness of the price anomaly detection is improved, and the price data and occurrence probability of the game items can be automatically adjusted to ensure the fairness and safety of the game transaction process.
[0024] Further, step S2 comprises:
[0025] Collecting game events in the game, supply-demand data volume of the game items, transaction record data and game behavior data of the players;
[0026] Preprocessing the collected data, and creating a game price prediction model using the preprocessed data;
[0027] Cross-validation and hyperparameter adjustment are performed on the game price prediction model.
[0028] From the above description, it can be known that in the process of establishing the game price prediction model, the game events, supply-demand data volume of the game items, transaction record data and game behavior data of the players are comprehensively considered, a more comprehensive prediction model can be constructed through multi-dimensional data analysis, and the accuracy and generalization ability of the model are improved through cross-validation and hyperparameter adjustment.
[0029] Further, the identification of the price anomaly problem based on the price data prediction result in step S3 includes:
[0030] If the price data prediction result of the game item indicates that the price data continues to grow, and the money supply in the game increases, it is identified that there is an inflation price anomaly problem;
[0031] If the price data prediction result of the game item indicates that the price data continues to decline, and the money supply in the game decreases, it is identified that there is a deflation price anomaly problem.
[0032] From the above description, the economic problems that may exist are identified through the price data prediction result, so as to adjust the prices and the appearance probability of the game items for the possible economic problems.
[0033] Further, the adjustment of the price data and the appearance probability of the game item in step S3 includes:
[0034] Based on the current price P of the game item and the current moving average price P a Adjustment:
[0035] P u = P + c x P a ;
[0036] In the formula, P u represents the adjusted price, and c represents the real-time adjustment coefficient;
[0037] The real-time adjustment coefficient is an adjustment coefficient generated according to the difference between the predicted price and the actual price of the game price prediction model;
[0038] When the real-time adjustment coefficient is positive, the appearance probability of the corresponding game item is reduced, and when the real-time adjustment coefficient is negative, the appearance probability of the corresponding game item is increased.
[0039] From the above description, the AI real-time adjustment coefficient is generated according to the difference between the predicted price and the actual price of the game item, and the current price of the game item is adjusted in combination with the moving average price. The appearance probability of the game item is adjusted correspondingly while the price is adjusted, which can adjust the price to ensure the fairness and safety of the game transaction process and reduce the interference of malicious transaction behavior on the overall game transaction.
[0040] Further, it further includes:
[0041] According to the clustering analysis of the transaction record data of the player, the abnormal transaction type is obtained through the clustering analysis result, and the transaction record data includes the transaction frequency and the transaction amount;
[0042] According to the abnormal transaction type, transaction detection is performed, and if an abnormal transaction is detected, transaction restriction is performed on the account corresponding to the abnormal transaction.
[0043] As can be seen from the above description, by analyzing and classifying historical transaction data, abnormal transaction behavior can be detected, and the account of the abnormal transaction can be restricted, further reducing the case of malicious transaction behavior and improving the security of game transactions.
[0044] Please refer to Figure 2 Another embodiment of the present application provides a game price anomaly detection terminal, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to realize each step of the game price anomaly detection method described above.
[0045] The game price anomaly detection method and terminal described above are suitable for improving the comprehensiveness of price anomaly detection when malicious transaction behavior occurs, to ensure the fairness and security of the game transaction process, which will be described in detail below through specific embodiments:
[0046] Embodiment one
[0047] Please refer to Figure 1 A game price anomaly detection method comprises the following steps:
[0048] S1, collecting data of game items, and determining the occurrence probability of the game items according to attribute information of the game items.
[0049] Specifically, the attribute information of the game items can include item rarity, and the corresponding rarity ratio is set according to the item rarity, that is, the higher the rarity of the item, the lower the occurrence probability of the item in the game. According to this ratio, the player can judge the value of the item according to the rarity of the item.
[0050] In some embodiments, the price calculation can also be combined with the popularity coefficient of the game item, the player's purchase willingness coefficient, and the seasonal demand coefficient, for example:
[0051] Price = base price + (rarity coefficient × item level) + (emotional analysis coefficient × popularity) + (purchase willingness coefficient × player's purchase willingness) + (seasonal demand coefficient × seasonal factor) + AI real-time adjustment coefficient × current moving average price.
[0052] S2, training and establishing a game price prediction model according to the collected game events, supply and demand data of game items, transaction record data and game behavior data of players in the game.
[0053] Specifically, step S2 comprises the following steps:
[0054] Collecting game events, supply and demand data of game items, transaction records of players, and game behavior data in the game;
[0055] Preprocessing the collected data and creating a game price prediction model using the preprocessed data;
[0056] Cross-validation and hyperparameter adjustment of the game price prediction model.
[0057] In this embodiment, the collection and arrangement of game data includes: transaction data, collecting players' transaction records, including the items, quantity, price and time of transactions; player behavior data, recording players' game behavior, such as login frequency, game duration, purchase and sale habits; game events, collecting special events in the game, such as promotional activities, update releases, etc., which may affect economic data; market supply and demand data, analyzing the supply and demand of items in the game, as well as their price changes.
[0058] The collected data is preprocessed, specifically including: data cleaning, removing erroneous, duplicate or incomplete records; feature extraction, extracting useful features from raw data, such as price trends, transaction volume, player activity, etc.; data conversion, converting non-numeric data into numeric data, such as processing categorical variables through One-Hot Encoding.
[0059] Establishing a prediction model specifically includes: selecting appropriate algorithms, selecting appropriate machine learning algorithms according to data characteristics and prediction targets, such as linear regression, decision tree, random forest, neural network, etc.; model training, training the selected algorithm using historical data to create a prediction model.
[0060] Model evaluation and optimization are then performed, specifically including: cross-validation, using cross-validation and regularization techniques to prevent model overfitting. At the same time, constantly learning from new data to improve the generalization ability of the model; hyperparameter adjustment, adjusting the hyperparameters of the model through methods such as Grid Search or Random Search to optimize the performance of the model.
[0061] In some embodiments, to improve the prediction accuracy of the prediction model, multiple machine learning algorithms can be used for prediction, and expert systems and rule engines can be combined to improve the accuracy of the prediction; at the same time, the model is updated regularly to adapt to changes in player behavior.
[0062] S3, predicting the price data of the game item according to the game item price prediction model, and identifying the price anomaly problem based on the price data prediction result, if the corresponding price anomaly problem is identified, it is determined that the price data of the game item is abnormal, and the price data and the occurrence probability of the game item are adjusted.
[0063] Specifically, the price of the game item is predicted according to the trained game item price prediction model, such as item price, transaction volume, and explosion rate. According to the analysis and prediction result, possible economic problems such as inflation, deflation or market bubble are identified.
[0064] If the price data prediction result of the game item indicates that the price data continues to grow, and the money supply in the game increases, it is identified that there is an inflation price anomaly problem.
[0065] If the price data prediction result of the game item indicates that the price data continues to decline, and the money supply in the game decreases, it is identified that there is a deflation price anomaly problem.
[0066] In this embodiment, dynamic adjustment rules can also be developed according to the prediction result and the game design target, for example, if the prediction shows that the price of a certain item will continue to decline, the explosion rate of the item may need to be reduced. The effect of the adjustment strategy is tested in the simulation environment to ensure that the adjustment does not cause game balance problems.
[0067] Specifically, adjusting the price data and the occurrence probability of the game item can include the following steps:
[0068] Based on the current price P of the game item and the current moving average price P a Adjustment:
[0069] P u = P + c x P a ;
[0070] In the formula, P u represents the adjusted price, and c represents the real-time adjustment coefficient;
[0071] The real-time adjustment coefficient is an adjustment coefficient generated according to the difference between the price predicted by the game item price prediction model and the actual price;
[0072] When the real-time adjustment coefficient is positive, the occurrence probability of the corresponding game item is reduced, and when the real-time adjustment coefficient is negative, the occurrence probability of the corresponding game item is increased.
[0073] For example, the original pricing formula for rare collectibles is: current price = base price + (rarity coefficient x collectible level), where the base price is 1000 gold coins, the rarity coefficient is 30, and the probability of obtaining a rare collectible is 0.05%. The adjusted price formula is: adjusted price = current price + AI real-time adjustment coefficient x current moving average price.
[0074] Wherein, according to the difference between the predicted market price and the actual market price, the AI real-time adjustment coefficient can be calculated. This coefficient can be positive or negative, depending on whether the predicted price is higher or lower than the actual price. Apply the calculated adjustment coefficient to the adjusted price formula of the item to get the final adjusted price. The result of price adjustment and the actual reaction of the market are input again as data for continuous learning and optimization of the model. In this embodiment, the AI real-time adjustment coefficient c can be defined as the ratio of the difference between the predicted price and the actual price, formula as follows c=(P1-P) / Pa, in the formula, P1 represents the predicted price obtained according to the prediction model. If c is positive, it means that the predicted price is higher than the actual price, and the probability of the appearance of the item should be reduced; if c is negative, it means that the predicted price is lower than the actual price, and the probability of the appearance of the item should be increased.
[0075] In some embodiments, in order to formulate dynamic adjustment rules that can maintain economic stability and not destroy the fun of the game, a set of preliminary adjustment rules can be formulated by combining economic theory and game design principles. Then, these rules are evaluated and optimized through simulation and A / B testing.
[0076] At the same time, in order to avoid too much system intervention that may affect the game experience of players and make players feel that the game has lost control, a transparent and explainable adjustment mechanism can be designed to let players understand the reasons and goals of the adjustment, while limiting the degree of system intervention and giving players a certain degree of freedom.
[0077] Embodiment two
[0078] Please refer to Figure 3 The difference between this embodiment and embodiment one is that it also includes:
[0079] According to the transaction record data of the players, the transaction record data including the transaction frequency and the transaction amount, the clustering analysis is performed to obtain the abnormal transaction type;
[0080] According to the abnormal transaction type, the transaction detection is performed, and if the abnormal transaction is detected, the transaction restriction is performed on the account corresponding to the abnormal transaction.
[0081] Specifically, the transaction patterns can be analyzed by algorithms to identify irregular transaction behaviors; thresholds can be set to monitor the transaction frequency of individual players or between players. If the number of transactions within a short period of time exceeds the normal range, these transactions will be marked for further review. By monitoring the transaction amount, abnormal large transactions can be identified. The behavior patterns of players can be analyzed to identify abnormal patterns that do not conform to normal game behavior.
[0082] For example, a player suddenly receives or sends a large amount of currency or high-value items that do not match its game progress, which triggers an alarm; or a group of players coordinate to buy or sell a certain item, which may cause temporary distortion of market prices, which can be warned and measures can be taken to alleviate the impact.
[0083] In this embodiment, clustering algorithms can be used to make mixed judgments based on transaction behavior, transaction amount, etc. to identify irregular transaction behaviors. The objects in the data set can be grouped so that the similarity between objects in the same group (i.e. a cluster) is high, while the similarity between objects in different groups is low. In transaction data analysis, this method can help identify abnormal patterns that may indicate irregular transaction behavior.
[0084] Among them, the transaction amount and transaction frequency features are selected for clustering analysis; when determining the number of clusters, for algorithms that require the number of clusters to be specified in advance (such as K-means), methods such as elbow rule, silhouette coefficient, etc. can be used to determine the optimal number of clusters; when selecting a clustering algorithm, select a suitable clustering algorithm according to the characteristics of the data and the analysis target, such as K-means, DBSCAN, hierarchical clustering, etc.
[0085] Specifically, first, understand the characteristics of transaction data, including transaction frequency, transaction amount, transaction party level, transaction item type, etc. Different features may affect the selection of clustering algorithms. Then, select K-means, DBSCAN, hierarchical clustering clustering algorithms, DBSCAN: suitable for identifying clusters of arbitrary shape and can identify outliers, which is very useful for detecting abnormal transactions; K-Means: if the transaction data can be reasonably divided into several predefined clusters and these clusters are roughly spherical, K-Means is a simple and effective choice; hierarchical clustering: if a visual clustering tree is needed to show different levels of transaction behavior, hierarchical clustering is a good choice. In practical applications, the above three clustering algorithms are combined and adjusted and optimized to adapt to specific game environments.
[0086] Further, run clustering algorithm, analyze clustering results, including: analyze cluster characteristics, check the characteristics of each cluster, determine whether there are clusters significantly different from normal transaction patterns, which may represent normal transaction patterns or contain abnormal transactions; identify abnormal clusters, if the transaction characteristics of a cluster are significantly different from most clusters or do not conform to the rules of the in-game economy, the cluster may contain abnormal transactions; review edge data points, data points marked as noise by the clustering algorithm, i.e. transactions not belonging to any cluster, need special attention as they may be abnormal transactions; time series analysis, analyze the time series of transaction data, see if there are frequent abnormal transactions in a short period of time, which may indicate automated abnormal transaction behavior; comprehensive analysis, combine clustering results and other information (such as player history, market trends, etc.) for comprehensive analysis to determine whether there are abnormalities.
[0087] After that, manual detection can also be carried out, for potential abnormal transactions detected automatically, manual review may be needed to confirm their abnormal nature and take appropriate measures.
[0088] After analyzing the clustering results, according to the feedback of the anomaly detection results, adjust the parameters of the clustering algorithm, such as the number of clusters, distance threshold, etc., to improve the accuracy of anomaly detection. The threshold can be dynamically set, for example: use historical transaction data and statistical analysis to determine the normal transaction frequency range; or, through machine learning algorithms such as clustering analysis or anomaly detection models, automatically identify normal player transaction behavior patterns and set thresholds accordingly.
[0089] Once abnormal transactions are detected, a series of response measures can be automatically taken, such as limiting account transaction functions, freezing accounts, sending warnings to players or notifying game administrators for manual review.
[0090] For example: a certain item has been traded an unusually large number of times in a short period of time, which may be due to players exploiting game bugs or using automated scripts to quickly trade and gain unfair economic benefits, the transaction function of the item involved in abnormal transactions can be temporarily frozen; players may create multiple game accounts and use unfair means (such as exploiting game bugs, automated scripts, etc.) to obtain in-game currency or other valuable items, then transfer these currencies or items to the main account or other players to obtain real-world currencies or other benefits. Appropriate punishment measures can be taken for accounts confirmed to be involved in abnormal transactions, such as temporarily or permanently banning accounts, limiting transaction functions, etc. Different severity levels can also be set according to the amount of money involved in the transaction, the frequency of transactions, whether it involves theft or fraud, etc. Usually divided into minor, moderate, and severe.
[0091] In some embodiments, a data visualization tool can also be included to help the game administrator quickly identify and understand abnormal patterns in the transaction data. In addition, reports can also be generated detailing the abnormal behavior monitored and the measures taken.
[0092] Embodiment Three
[0093] Please refer to Figure 2 A game price anomaly detection terminal 1, comprising a memory 2, a processor 3, and a computer program stored in the memory 2 and executable on the processor 3, wherein the processor 3 implements each step of the game price anomaly detection method of embodiment one or embodiment two when executing the computer program.
[0094] In summary, the game price anomaly detection method and terminal provided by the present application can collect data of game items, determine the occurrence probability of game items according to the attribute information of game items, collect game events, supply and demand data volume of game items, transaction record data and game behavior data of players in the game, train and establish a game price prediction model based on the above data, so as to construct a more comprehensive and accurate prediction model through multi-dimensional data analysis. Then, the price data of game items is predicted according to the game price prediction model, and the price anomaly problem is identified based on the price data prediction result. If an anomaly is identified, it is determined that the price data of the game item is abnormal, and the price and occurrence probability of the game item are adjusted. In this way, the abnormality of the item price can be identified after the malicious transaction behavior occurs, the comprehensiveness of the price anomaly detection is improved, and the price data and occurrence probability of the game item can be automatically adjusted to ensure the fairness and safety of the game transaction process.
[0095] The above description is only an embodiment of the present application, and does not limit the patent scope of the present application. Any equivalent transformation or direct or indirect application in related technical fields based on the content of the specification and drawings is also included in the patent protection scope of the present application.
Claims
1. A method for detecting price anomalies in games, characterized in that, Including the following steps: S1. Collect data on game items and determine the probability of their appearance based on their attribute information; S2. Based on the collected data on game events, supply and demand of game items, player transaction records, and game behavior, train and build a game price prediction model. S3. Predict the price data of game items according to the game price prediction model, and identify price anomalies based on the price data prediction results. If a corresponding price anomaly is identified, it is determined that the price data of the game items is abnormal, and the price data and occurrence probability of the game items are adjusted.
2. The method for detecting price anomalies in games according to claim 1, characterized in that, Step S2 includes: Collect data on game events, supply and demand of game items, player transaction records, and game behavior. The collected data is preprocessed, and the preprocessed data is used to create a game price prediction model; Cross-validation and hyperparameter tuning were performed on the game price prediction model.
3. The method for detecting price anomalies in games according to claim 1, characterized in that, Step S3 involves identifying price anomalies based on price data prediction results, including: If the price data prediction results of the game items indicate that the price data continues to grow, and the currency supply in the game increases, then an inflationary price anomaly is identified. If the price data prediction results for the game items indicate a continuous decline in price data, and the currency supply in the game decreases, then a deflationary price anomaly is identified.
4. The method for detecting price anomalies in games according to claim 3, characterized in that, Step S3 involves adjusting the price data and occurrence probability of the game items, including: Based on the current price P of the game item and the current average price P a Adjustments will be made: P u =P+c×P a ; In the formula, P u 'c' represents the adjusted price, and 'c' represents the real-time adjustment coefficient. The real-time adjustment coefficient is an adjustment coefficient generated based on the difference between the price predicted by the game price prediction model and the actual price; When the real-time adjustment coefficient is positive, the probability of the corresponding game item appearing is reduced; when the real-time adjustment coefficient is negative, the probability of the corresponding game item appearing is increased.
5. The method for detecting price anomalies in games according to claim 1, characterized in that, Also includes: Cluster analysis is performed on the players' transaction records to identify abnormal transaction types. The transaction records include transaction frequency and transaction amount. Transaction detection is performed based on the abnormal transaction type. If an abnormal transaction is detected, transaction restrictions are imposed on the account corresponding to the abnormal transaction.
6. A game price anomaly detection terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it performs the following steps: S1. Collect data on game items and determine the probability of their appearance based on their attribute information; S2. Based on the collected data on game events, supply and demand of game items, player transaction records, and game behavior, train and build a game price prediction model. S3. Predict the price data of game items according to the game price prediction model, and identify price anomalies based on the price data prediction results. If a corresponding price anomaly is identified, it is determined that the price data of the game items is abnormal, and the price data and occurrence probability of the game items are adjusted.
7. A game price anomaly detection terminal according to claim 6, characterized in that, Step S2 includes: Collect data on game events, supply and demand of game items, player transaction records, and game behavior. The collected data is preprocessed, and the preprocessed data is used to create a game price prediction model; Cross-validation and hyperparameter tuning were performed on the game price prediction model.
8. A game price anomaly detection terminal according to claim 6, characterized in that, Step S3 involves identifying price anomalies based on price data prediction results, including: If the price data prediction results of the game items indicate that the price data continues to grow, and the currency supply in the game increases, then an inflationary price anomaly is identified. If the price data prediction results for the game items indicate a continuous decline in price data, and the currency supply in the game decreases, then a deflationary price anomaly is identified.
9. A game price anomaly detection terminal according to claim 8, characterized in that, Step S3 involves adjusting the price data and occurrence probability of the game items, including: Based on the current price P of the game item and the current average price P a Adjustments will be made: P u =P+c×P a ; In the formula, P u 'c' represents the adjusted price, and 'c' represents the real-time adjustment coefficient. The real-time adjustment coefficient is an adjustment coefficient generated based on the difference between the price predicted by the game price prediction model and the actual price; When the real-time adjustment coefficient is positive, the probability of the corresponding game item appearing is reduced; when the real-time adjustment coefficient is negative, the probability of the corresponding game item appearing is increased.
10. A game price anomaly detection terminal according to claim 6, characterized in that, Also includes: Cluster analysis is performed on the players' transaction records to identify abnormal transaction types. The transaction records include transaction frequency and transaction amount. Transaction detection is performed based on the abnormal transaction type. If an abnormal transaction is detected, transaction restrictions are imposed on the account corresponding to the abnormal transaction.