Game recommendation method and device based on big data, electronic equipment and medium

Abstract

The invention discloses a game recommendation method and system based on big data, electronic equipment and a computer readable storage medium, and behavior and preference data of a user are collected through diversified data sources, such as game history, game evaluation, social network behaviors and the like. And then, modeling and analyzing the data by utilizing an advanced machine learning algorithm, such as deep learning and collaborative filtering, so that the preference of the user can be predicted more accurately, and the most suitable game can be recommended for the user. The system has higher accuracy, more real-time feedback and better expansibility, and can better meet the requirements of users. The game recommendation system aims at overcoming the defects of a traditional recommendation system and providing more accurate, personalized and real-time game recommendation services for users. Along with continuous development of the game market, the game market will become an important trend and a development direction in the future game recommendation field.

Description

Technical Field

[0001] The present invention belongs to the technical field of big data and data analysis, and in particular, relates to a game recommendation method, system, electronic device and computer-readable storage medium for big data. Background Art

[0002] With the rapid and vigorous development of the game market, game products continue to emerge, and the game market shows a trend of diversification and rapid development. However, users often have no idea where to start when choosing games, because it is not easy to quickly find the right game for them among the massive amount of games. This phenomenon is vividly called information overload.

[0003] At present, traditional game recommendation systems often use simple rule matching or popular game lists. Although this method can provide users with some choices, it is difficult to meet the personalized needs of users. For example, a player who likes action-adventure games may not be interested in simulation business games, and traditional recommendation systems may not be able to distinguish these differences. At the same time, to build an efficient game recommendation system, it is also a challenge to collect and analyze user behavior and preference data. Users' game preferences may be diverse, including preferences in game types, themes, difficulty, etc., and these preferences may change over time and in different situations. Therefore, it becomes crucial to accurately capture and understand user preferences.

[0004] Due to the bottlenecks of traditional recommendation systems in terms of accuracy, real-time and scalability, coupled with the rapid development of the game market and the diversified needs of users, traditional methods often cannot keep up with the pace of change. Therefore, there is an urgent need for a game recommendation method and system based on big data to meet the personalized needs of users. Summary of the invention

[0005] The purpose of the present invention is to provide a game recommendation method, system, electronic device and computer-readable storage medium based on big data, which collects user behavior and preference data through diversified data sources, and uses recommendation algorithms for modeling and analysis, so as to solve the problems of insufficient personalization, poor real-time performance and limited scalability in the prior art.

[0006] The first technical solution adopted by the present invention is a game recommendation method based on big data, comprising: Data collection: Obtaining user behavior data related to game browsing, game purchases, game ratings, game reviews, and game time; Data preprocessing: clean the behavioral data to obtain preliminary screening data; Feature extraction: Based on the initial screening data, extract user preference features through feature engineering; User portrait construction: Use clustering algorithm (K-means) to construct user portraits, identify users' unique preferences, and obtain user portrait data; Model training: Use collaborative filtering, content-based recommendation or hybrid recommendation algorithms, combined with user portrait data, to train the model; Recommendation generation: By combining the model with user portrait data, we can filter out game recommendation results that match the user from the game database, and dynamically adjust the sorting to obtain the recommendation results; Result push: Push and display the recommended results to users.

[0007] The characteristics of this technical solution are also: Furthermore, behavioral data is obtained from social media, game forums, and in-game activity records through API interfaces, log records, and web crawlers.

[0008] Going a step further, web crawlers include Selenium and Appnium.

[0009] Furthermore, the cleaning process includes removing invalid data, filling missing values, and data standardization.

[0010] Furthermore, regular expressions and outlier detection methods are used to eliminate invalid or erroneous records, and interpolation is used to fill missing values; the numerical data are z-score standardized so that the mean of all features is 0 and the standard deviation is 1.

[0011] Furthermore, the preference characteristics include game type preference and game difficulty preference.

[0012] Furthermore, the model consists of an embedding layer, a multi-layer perceptron layer, and an output layer. The embedding layer is used to convert discrete features into low-dimensional dense vectors; the multi-layer perceptron layer is responsible for learning high-level feature interactions; and the output layer predicts the user's interest score for games that have not been played.

[0013] The second technical solution adopted by the present invention is a game recommendation system based on big data, comprising the following units: Data collection unit: used to obtain user's game browsing, game purchase, game rating, game review, and game time related behavior data; Data preprocessing unit: used to clean the behavioral data and obtain preliminary screening data; Feature extraction unit: Based on the initial screening data, it is used to extract user preference features through feature engineering; User portrait construction unit: Use clustering algorithm (K-means) to construct user portraits to identify users' unique preferences and obtain user portrait data; Model training unit: uses collaborative filtering, content-based recommendation or hybrid recommendation algorithms, combined with user portrait data, to train the model; Recommendation generation unit: used to combine the model with the user portrait data to filter out the game recommendation results that match the user from the game database, and dynamically adjust the sorting to obtain the recommendation results; Result push unit: used to push and display recommendation results to users.

[0014] The third technical solution adopted by the present invention is a game recommendation electronic device based on big data, including a memory for storing programs in a processor; wherein the program includes instructions, and when the instructions are executed by the processor, the processor executes the above-mentioned game recommendation method based on big data.

[0015] The fourth technical solution adopted by the present invention is a computer storage medium, which stores a computer program. When the program is executed, it executes the above-mentioned game recommendation method based on big data.

[0016] The present invention collects user behavior and preference data through diversified data sources, such as game history, game evaluation, social network behavior, etc. Then, advanced machine learning algorithms, such as deep learning, collaborative filtering, etc., are used to model and analyze these data, so that user preferences can be more accurately predicted and the most suitable games can be recommended for them. This system has higher accuracy, more real-time feedback and better scalability, and can better meet user needs. Innovation aims to solve the shortcomings of traditional recommendation systems and provide users with more accurate, personalized and real-time game recommendation services. With the continuous development of the game market, it will become an important trend and development direction in the field of game recommendation in the future.

[0017] The details are as follows: 1. Personalization improvement: Traditional recommendation systems lack personalization and cannot provide customized recommendations based on users’ unique interests and behavioral characteristics. This invention provides highly personalized game recommendations through user interest models and machine learning algorithms.

[0018] 2. Good real-time performance: Due to the limitations of data processing and recommendation strategies, the traditional system has poor real-time performance and cannot respond quickly to user needs. The present invention uses a big data processing framework and real-time analysis technology to ensure the real-time performance of recommendation results.

[0019] 3. Improved scalability: With the growth of the number of users and the amount of data, traditional systems may face performance bottlenecks and be unable to effectively process large-scale data. The present invention provides highly scalable recommendation services through distributed computing and storage technology.

[0020] 4. Integration of diverse data sources: Game recommendations require the integration of diverse data sources, including game behavior, social network interactions, purchase records, etc. Traditional systems are difficult to process these data effectively. The present invention uses big data technology to collect and process diverse data sources.

[0021] 5. User retention and stickiness: Due to inaccurate recommendations, traditional systems may lead to low user retention and poor stickiness. The present invention improves user experience and promotes user retention through accurate game recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a flowchart of a game recommendation method based on big data of the present invention; Figure 2 It is a schematic diagram of the offline real-time data processing flow based on Hadoop of the present invention; Figure 3 It is a schematic diagram of the user interaction process of the present invention; Figure 4 It is a schematic diagram of the collaborative process of the present invention. DETAILED DESCRIPTION

[0023] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0024] The present invention aims to solve the problems of insufficient personalization, poor real-time performance and limited scalability in the prior art. In the gaming industry, traditional recommendation systems often rely on simple rule matching, popular lists or purchase history, which cannot accurately reflect the interests and preferences of users, resulting in a lack of pertinence in recommendation results and poor user experience. In addition, with the continuous increase in the number of users and the amount of data, the existing system performs poorly in terms of real-time performance and scalability.

[0025] In order to overcome these defects, the present invention collects user behavior and preference data through diversified data sources, and uses recommendation algorithms to model and analyze them, so as to provide users with efficient, accurate and real-time game recommendation services. Figure 1 As shown, the specific implementation is as follows: Step 1: Data collection: Obtain user behavior data related to game browsing, game purchasing, game ratings, game reviews, and game time.

[0026] Through API interfaces and logging as well as Selenium and Appnium web crawler technologies, user browsing, purchase, rating, comment, game time and other behavioral data are collected from different platforms (such as social media, game forums, in-game activity records, etc.).

[0027] The present invention collects user behavior and preference data from a variety of data sources through data collection. Data sources may include in-game behavior data, social network interaction data, purchase records, device information, etc. The purpose of data collection is to obtain a comprehensive user portrait and provide a basis for subsequent analysis and recommendation. In order to solve these technical problems, the game recommendation system and method of the present invention can provide more accurate, real-time, and personalized game recommendation services, improve user experience, and enhance the overall benefits of the game industry.

[0028] Step 2: Data preprocessing: Clean the behavioral data to obtain preliminary screening data.

[0029] After collecting behavioral data, the data needs to be cleaned, converted, and normalized. This step includes data deduplication, data format conversion, data normalization, and other operations. The purpose of data preprocessing is to ensure the consistency and quality of the data to facilitate subsequent analysis and modeling.

[0030] Step 3: Feature extraction: Based on the initial screening data, extract user preference features through feature engineering, such as game type preference, game difficulty preference, etc.

[0031] Step 4: User portrait construction: Use clustering algorithm (K-means) to construct user portraits, identify users' unique preferences, and obtain user portrait data.

[0032] Step 5: Model training: Use collaborative filtering, content-based recommendation or hybrid recommendation algorithms, combined with user portrait data, to train the model.

[0033] Step 6: Recommendation generation: By combining the model with user portrait data, the game recommendation results that meet the user's needs are screened out from the game database, and the ranking is dynamically adjusted to obtain the recommendation results.

[0034] Step 7: Push results: Push and display the recommended results to users.

[0035] Based on the model output, a list of games that best suits user preferences is selected from the game database and displayed in a personalized way through the user interface. The recommendation order is dynamically adjusted by taking into account factors such as freshness and popularity. The results are displayed and pushed to operations through Zhiwanyou's self-developed WEB platform and DingTalk's API interface.

[0036] The present invention also provides a game recommendation system based on big data, which includes a data collection unit, a data preprocessing unit, a feature extraction unit, a user portrait construction unit, a model training unit, a recommendation generation unit, and a result push unit.

[0037] The data collection unit can collect behavioral data from multiple dimensions, including but not limited to game platform login logs, in-game event tracking (such as completing tasks, purchasing props), user community interaction records (forum posts, comments), social media sharing and likes, etc. The behavioral data is synchronized to the data center once an hour through crawler scripts and API interfaces.

[0038] The data preprocessing unit cleans the behavior data to obtain preliminary screening data. The specific processing methods include the following: Data denoising: First, use regular expressions and outlier detection methods to remove invalid or erroneous records, such as identifying and removing data entries with abnormal timestamps.

[0039] Missing value processing: For fields with missing values, interpolation methods (such as mean interpolation and nearest neighbor interpolation) are used to fill them, especially for key indicators such as user activity.

[0040] Data standardization: In order to unify the comparison of data of different magnitudes, the numerical data is z-score standardized so that the mean of all features is 0 and the standard deviation is 1.

[0041] The data collection unit and the data preprocessing unit can ensure the efficiency and scalability of data processing based on big data technology through Hadoop. Figure 2 As shown in the figure, offline data uses DataX to import offline data sources into HDFS, and uses MapReduce tools in the Hadoop ecosystem to clean, transform, and aggregate data. Real-time data is captured by deploying Maxwell to monitor the MySQL database binlog logs, and Kafka is used as a message queue to ensure that real-time data changes can be reliably transmitted and stored. By combining DataX and Hadoop, batch processing and analysis of offline data can be performed efficiently. Maxwell and Kafka can be used to capture and process database changes in real time and build a real-time data processing pipeline. These two processing methods complement each other and provide a comprehensive data processing solution.

[0042] Feature extraction unit: Extract meaningful features from the original behavior data. For example, the user's game type preference is determined by counting the frequency of users playing specific types of games. Recursive feature elimination (RFE) and correlation-based feature selection are used to select the most predictive subset from a large number of user behavior features, such as game time in the past week, number of historical purchases, average ratings, etc.

[0043] User portrait construction unit: Use K-means++ as the initialization strategy to group users, and determine the optimal k value through silhouette coefficient evaluation. The clustering results show clear preference differences between user groups, such as "heavy RPG players" and "casual puzzle enthusiasts", and finally construct user portraits.

[0044] Model training unit: Build a deep learning model consisting of an embedding layer, a multi-layer perceptron (MLP), and an output layer. The embedding layer is used to convert discrete features (such as game ID) into low-dimensional dense vectors; the MLP layer is responsible for learning high-level feature interactions; the output layer predicts the user's interest score for games they have not played. Use the stochastic gradient descent (SGD) optimizer combined with early stopping to prevent overfitting. Use grid search to adjust hyperparameters such as learning rate (0.001, 0.01, 0.1), batch size (32, 64, 128), etc. to find the optimal configuration.

[0045] The model training unit combined with the user portrait building unit can use the machine learning framework, such as TensorFlow or PyTorch, to build and train the user interest model.

[0046] Recommendation generation unit: By combining the model with user portrait data, the game recommendation results that meet the user's needs are screened out from the game database, and the ranking is dynamically adjusted to obtain the recommendation results.

[0047] Result push unit: Provide personalized game recommendation services to users based on model results. This step can be achieved through application programming interface (API), game client interface, email, etc. The recommendation service needs to be real-time and highly scalable to ensure that accurate recommendations can still be provided when a large number of users visit. Figure 3 As shown in the figure, when the user plays the game, the game client records various user operations and events (such as login, level completion, purchase behavior, etc.) and generates dot data. These data are transmitted to the data platform through the real-time data pipeline (Kafka) and stored in a centralized data warehouse or distributed file system (HDFS). Through third-party data service providers or public data sources, user behavior data of similar competitive games are obtained. These data are stored in the data platform through API interfaces and batch imports. The text data such as posts and comments in the game forum are regularly captured through crawler technology. Natural language processing (NLP) technology is used to pre-process the captured data (segmentation, noise removal) and store it in the data platform. After the platform cleans the data, Python is used for feature extraction and transformation, K-means is used to group users, and supervised learning algorithms are used to predict user preferences.

[0048] 1. Enhanced diversity and novelty: In the final recommendation list generation stage, diversity constraints are introduced to ensure that the recommended content not only fits user preferences, but also includes different types or newly launched games, thereby enhancing users' fun in exploring new games.

[0049] 2. Feedback loop: Establish a real-time feedback mechanism to collect user behavioral feedback such as clicks and ratings on recommended content, continuously iterate and update user portraits and recommendation models, and form a closed-loop optimization.

[0050] Depend on Figure 4 It can be seen that in the data preprocessing process, Hadoop is used to denoise the historical data stored in HDFS, and Kafka Steams is used to denoise the streaming data for real-time data. Mean filling, median filling, specific value filling and K-nearest neighbor algorithm (KNN) are used for filling, and the data is filled to meet the standard normal distribution with a mean of 0 and a standard deviation of 1. The extracted user behavior features include login frequency, purchase records, browsing game information, etc., and the user attribute features include age, gender, region, etc. After extracting the features, the clustering algorithm K-means is used to group the users, and the feedback data and evaluation indicators (such as accuracy and recall rate) are used to continuously optimize the user portrait. The appropriate loss function (such as mean square error, cross entropy) and optimization algorithm SGD are selected. Through the above steps, offline data and incremental data can be effectively processed, the data can be denoised, missing values ​​processed and standardized, and key features can be extracted. Based on these data, the user portrait is constructed and optimized, the push algorithm model is trained, and the model is optimized using stochastic gradient descent (SGD), so as to continuously improve the accuracy of the recommendation system and user experience.

[0051] Through the above embodiments, the flexibility and adaptability of the game recommendation system based on big data of the present invention are reflected, proving that the accuracy of game recommendations and user experience can be effectively improved under different parameter configurations.

[0052] In the specific implementation process, the present invention can achieve high scalability through a distributed computing framework. The result push unit is combined with the recommendation generation unit, and can interact with the front end through the back-end API to provide real-time game recommendations.

[0053] The game recommendation system using the method of the present invention was tested on a conventional game platform and compared with a control group that did not use personalized recommendations. The experimental data showed that according to the analysis of the dot data, the user click rate increased by 20% and the game retention rate increased by 15%. The results of the user satisfaction questionnaire showed that more than 80% of users believed that the recommended content was closer to their personal interests and the experience was significantly improved.

[0054] The beneficial effect of the present invention is high accuracy. Through diversified data sources and advanced recommendation algorithms, the system can provide highly accurate recommendations based on the user's interests and behavioral characteristics. Using a big data processing framework, the system can collect and analyze user data in real time to ensure the timeliness of recommendation results. Using a distributed database and a big data processing framework, the system has high scalability and can handle large amounts of data and a growing number of users. Personalized recommendation services can improve user satisfaction and stickiness, promote user retention and sales of game products.

[0055] Another embodiment of the present application provides an electronic device, which includes: a processor and a memory, wherein the memory is used to store a program of the method, and when the program is read and executed by the processor, the above-mentioned game recommendation method is executed.

[0056] Another embodiment of the present application further provides a computer storage medium, which stores a computer program, and the program performs the above-mentioned game recommendation method when executed.

[0057] It should be noted that the detailed description of the electronic device and computer storage medium provided in the embodiments of the present application can refer to the relevant description of the above-mentioned method embodiments provided in the present application, and will not be repeated here.

[0058] Although the present application is disclosed as above in the form of a preferred embodiment, it is not intended to limit the present application. Any technical personnel in this field may make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

[0059] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory. The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0060] 1. Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined in this article, computer-readable media does not include non-transitory computer-readable media (transitor7media), such as modulated data signals and carrier waves.

[0061] 2. Those skilled in the art should understand that the embodiments of the present application can be provided as a system or electronic device. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

Claims

1. A game recommendation method based on big data, characterized in that: include: Data collection: Obtaining user behavior data related to game browsing, game purchases, game ratings, game reviews, and game time; Data preprocessing: clean the behavioral data to obtain preliminary screening data; Feature extraction: Based on the initial screening data, extract user preference features through feature engineering; User portrait construction: Use clustering algorithm (K-means) to construct user portraits, identify users' unique preferences, and obtain user portrait data; Model training: Use collaborative filtering, content-based recommendation or hybrid recommendation algorithms, combined with user portrait data, to train the model; Recommendation generation: By combining the model with user portrait data, we can filter out game recommendation results that match the user from the game database, and dynamically adjust the sorting to obtain the recommendation results; Result push: Push and display the recommended results to users.

2. A game recommendation method based on big data according to claim 1, characterized in that: The behavioral data is obtained from social media, game forums, and in-game activity records through API interfaces, log records, and web crawlers.

3. A game recommendation method based on big data according to claim 2, characterized in that: The web crawlers include Selenium and Appnium.

4. The game recommendation method based on big data according to claim 1, characterized in that: The cleaning process includes removing invalid data, filling missing values, and data standardization.

5. The game recommendation method based on big data according to claim 4, characterized in that: Regular expressions and outlier detection methods were used to remove invalid or erroneous records, and interpolation was used to fill missing values. The numerical data were z-score standardized so that the mean of all features was 0 and the standard deviation was 1.

6. The game recommendation method based on big data according to claim 1, characterized in that: The preference characteristics include game type preference and game difficulty preference.

7. The game recommendation method based on big data according to claim 1, characterized in that: The model consists of an embedding layer, a multi-layer perceptron layer and an output layer. The embedding layer is used to convert discrete features into low-dimensional dense vectors; the multi-layer perceptron layer is responsible for learning high-level feature interactions, and the output layer predicts the user's interest score for games that have not been played.

8. A game recommendation system based on big data, characterized in that: The following units are included: Data collection unit: used to obtain user's game browsing, game purchase, game rating, game review, and game time related behavior data; Data preprocessing unit: used to clean the behavioral data and obtain preliminary screening data; Feature extraction unit: Based on the initial screening data, it is used to extract user preference features through feature engineering; User portrait construction unit: Use clustering algorithm (K-means) to construct user portraits to identify users' unique preferences and obtain user portrait data; Model training unit: uses collaborative filtering, content-based recommendation or hybrid recommendation algorithms, combined with user portrait data, to train the model; Recommendation generation unit: used to combine the model with the user portrait data to filter out the game recommendation results that match the user from the game database, and dynamically adjust the sorting to obtain the recommendation results; Result push unit: used to push and display recommendation results to users.

9. An electronic device comprising a processor and a memory storing a program, wherein the program comprises instructions, characterized in that: When the instructions are executed by the processor, the processor executes a game recommendation method based on big data as described in any one of claims 1-7.

10. A computer storage medium storing a computer program, characterized in that: When the program is executed, it performs a game recommendation method based on big data as described in any one of claims 1 to 7.

Images