Disclosure of Invention
The invention mainly solves the technical problem of how to accurately recommend the game software in the application store.
According to a first aspect, the present invention provides a software recommendation method based on big data analysis and artificial intelligence, comprising: acquiring running video of an installed game, setting information of the installed game and mobile phone hardware information; determining a game touch frequency of a user by using a touch frequency determination model based on the running video of the installed game; determining the operation preference degree of the user based on the touch frequency of the user game; determining a user's image quality requirement level based on the setting information of the installed game; determining communication interaction data in the user game by using a communication interaction model based on the running video of the installed game; determining the game social degree of the user based on the communication interaction data in the user game; and determining a plurality of game software to be recommended by using a game determination model based on the operation preference degree of the user, the image quality requirement degree of the user, the game social degree of the user and the mobile phone hardware information.
Further, the interactive data includes text interactive data, picture interactive data and voice interactive data.
Further, the touch frequency determining model is a long-short period neural network model, the input of the touch frequency determining model is a running video of the installed game, and the output of the touch frequency determining model is game touch frequency of the user.
Furthermore, the communication interaction model is a long-short-period neural network model, the input of the communication interaction model is the running video of the installed game, and the output of the communication interaction model is communication interaction data in the user game.
Still further, the method further comprises: if the game social degree of the user exceeds a social threshold, processing and outputting based on communication interaction data in the user game corresponding to the game social degree of the user by using a social software recommendation model to obtain recommended social software, wherein the social software recommendation model is a deep neural network model, the input of the social software recommendation model is communication interaction data in the user game corresponding to the game social degree of the user, and the output of the social software recommendation model is recommended social software.
According to a second aspect, the present invention provides a software recommendation system based on big data analysis and artificial intelligence, comprising: the acquisition module is used for acquiring running video of the installed game, setting information of the installed game and mobile phone hardware information; the touch frequency module is used for determining game touch frequency of a user by using a touch frequency determination model based on the running video of the installed game; the preference degree determining module is used for determining the operation preference degree of the user based on the touch frequency of the user game; an image quality requirement determining module for determining an image quality requirement degree of a user based on the setting information of the installed game; the communication interaction module is used for determining communication interaction data in the user game by using a communication interaction model based on the running video of the installed game; the game social module is used for determining the game social degree of the user based on the communication interaction data in the user game; and the recommendation module is used for determining a plurality of game software to be recommended by using a game determination model based on the operation preference degree of the user, the image quality requirement degree of the user, the game social degree of the user and the mobile phone hardware information.
Further, the interactive data includes text interactive data, picture interactive data and voice interactive data.
Further, the touch frequency determining model is a long-short period neural network model, the input of the touch frequency determining model is a running video of the installed game, and the output of the touch frequency determining model is game touch frequency of the user.
Furthermore, the communication interaction model is a long-short-period neural network model, the input of the communication interaction model is the running video of the installed game, and the output of the communication interaction model is communication interaction data in the user game.
Still further, the system is further configured to: if the game social degree of the user exceeds a social threshold, processing and outputting based on communication interaction data in the user game corresponding to the game social degree of the user by using a social software recommendation model to obtain recommended social software, wherein the social software recommendation model is a deep neural network model, the input of the social software recommendation model is communication interaction data in the user game corresponding to the game social degree of the user, and the output of the social software recommendation model is recommended social software.
The invention provides a software recommending method and a system based on big data analysis and artificial intelligence, wherein the method comprises the steps of obtaining running video of an installed game, setting information of the installed game and mobile phone hardware information; determining a game touch frequency of a user by using a touch frequency determination model based on the running video of the installed game; determining the operation preference degree of the user based on the touch frequency of the user game; determining a user's image quality requirement level based on the setting information of the installed game; determining communication interaction data in the user game by using a communication interaction model based on the running video of the installed game; determining the game social degree of the user based on the communication interaction data in the user game; the method can accurately recommend the game software in the application store by determining a plurality of game software to be recommended by using a game determination model based on the operation preference degree of the user, the image quality requirement degree of the user, the game social degree of the user and the mobile phone hardware information.
Detailed Description
In the embodiment of the invention, a software recommendation method based on big data analysis and artificial intelligence is provided as shown in fig. 1, and the software recommendation method based on big data analysis and artificial intelligence comprises the following steps of S1-S7:
step S1, acquiring running video of the installed game, setting information of the installed game and mobile phone hardware information.
The installed game is game software downloaded and installed on the mobile phone by the user. Because the installed games are games which are remained after the installed games are selected by users, the installed games are games which are favored by the users after the users experience, certain information in the installed games can express the favorites of the users, and the favorites of the users can be obtained by analyzing the information of the installed games, so that recommendation of game software can be performed based on the favorites of the users.
The running video of the installed game is the running video of the game obtained by recording the screen of the mobile phone after the user has played the game.
The setting information of the installed game is various options and parameters manually set by the user in the installed game, and the setting information of the installed game comprises game image quality definition, special effect requirements in the game, volume requirements, operation experience, game keys and the like.
The mobile phone hardware information comprises relevant hardware information of a mobile phone of a user, including mobile phone model, screen resolution, storage capacity, running memory, processor model, display card information and the like.
And S2, determining the game touch frequency of the user by using a touch frequency determination model based on the running video of the installed game.
The touch frequency determination model can acquire the game behaviors and operation processes of players in the game by analyzing the running video of the installed game. These videos record the player's touch actions in the game, including key presses, swipes, clicks, etc. By analyzing the videos, information such as time stamp and duration of each operation can be obtained, and therefore game touch frequency of a user is obtained.
The game touch frequency is the touch frequency of a user when the user performs game operation, the higher the game touch frequency is, the more intense the user operates in the installed game, the stronger the complexity degree of the installed game is, the more the user likes the game operation feel, and the higher the interest degree of the user in the operation type game is. The running video of the installed game may show the user's touching operation of the mobile phone during the game.
The touch frequency determining model is a long-short-period neural network model. The long-term neural network model is one implementation of artificial intelligence. The Long-Short Term neural network model includes a Long-Short Term neural network (LSTM). The long-term and short-term neural network model can process sequence data with any length, capture sequence information and output results based on the association relationship of front data and rear data in the sequence. The long-short-term neural network model is used for processing the running video of the installed game in the continuous time period, so that the characteristics of the association relationship among the running videos of the installed game comprehensively considering each time point can be output, and the output characteristics are more accurate and comprehensive.
The touch frequency determination model can be obtained by training a training sample through a gradient descent method.
The touch frequency determining model is a long-short-period neural network model, the input of the touch frequency determining model is the running video of the installed game, and the output of the touch frequency determining model is the game touch frequency of the user.
And step S3, determining the operation preference degree of the user based on the touch frequency of the user game.
The operation preference degree of the user indicates the preference degree of the user for the operation in the game. The operational preference of a game user may be proportional to the frequency of touches of the user game. In some embodiments, the user's operation preference level may be determined by a preset relationship between the touch frequency of the user game and the user's operation preference level. The game operation degree can be a value between 0 and 1, and the larger the value is, the more complicated the operation and the stronger the operation feeling are, the more the user likes the game. The preset relationship may be that the touch frequency of the user game is proportional to the operation preference degree of the user.
And step S4, determining the image quality requirement degree of the user based on the setting information of the installed games.
The setting information of the installed game may express the preference of the user, so the degree of image quality requirement of the user may be determined based on the setting information of the installed game. The image quality requirement level refers to a requirement level of a user for game image quality in a game. The degree of image quality requirement may be a value ranging from 0 to 1, and a larger value indicates a higher requirement of the user for game image quality, and it is necessary to recommend game software having a better image quality and finer screen.
In some embodiments, the setting information of the installed game may be constructed as a vector to be matched, and the image quality requirement level of the user corresponding to the reference vector with the distance smaller than the threshold value is determined as the current image quality requirement level of the user by calculating the distance between the vector to be matched and each reference vector in the database. The database is pre-constructed, and comprises reference vectors and image quality requirement degrees of users corresponding to the reference vectors, wherein the reference vectors are constructed based on setting information of installed games in historical data. The image quality requirement level of the user corresponding to the reference vector is an accurate image quality requirement level determined based on the history data.
And S5, determining communication interaction data in the user game by using a communication interaction model based on the running video of the installed game.
The interactive data in the user game is the data for chat interaction with other users in the user game process. The running video of the installed game may display interactive data for communication in the user's game. The interactive data in the user game comprises text interactive data, picture interactive data and voice interactive data. For example, the running video of the installed game may display a text chat record, a voice chat record, a picture communication record, etc. of the user with other users.
And S6, determining the game social degree of the user based on the communication interaction data in the user game.
In some embodiments, the social level of the user's game may be obtained by analyzing the interactive data of the user's game. For example, the more the user communicates with other users and the higher the communication frequency is, the more the game social degree of the user is indicated in the communication interaction data in the user game.
In some embodiments, the interaction data in the user's game may be analyzed by an interaction model to determine the user's game social level. The input of the communication interaction model is communication interaction data in the game of the user, and the output of the communication interaction model is the game social degree of the user. The communication interaction model is a deep neural network model.
The communication interaction model is a deep neural network model comprising a deep neural network (Deep Neural Networks, DNN). The deep neural network model is one implementation of artificial intelligence. The deep neural network may include a recurrent neural network (Recurrent Neural Network, RNN), a convolutional neural network (Convolutional Neural Networks, CNN), a generating countermeasure network (Generative Adversarial Networks, GAN), and so on.
The game social degree of a user is the degree to which the user interacts with other players in the game. The game social degree of the user can be a threshold value between 0 and 1, and the higher the game social degree of the user is, the more social preference of the user in the game is indicated, and game software with strong social property, such as a multi-person team game, can be recommended. The lower the game social degree of the user is, the less the user likes the social contact in the game, and the game of the stand-alone class can be recommended.
And S7, determining a plurality of game software to be recommended by using a game determination model based on the operation preference degree of the user, the image quality requirement degree of the user, the game social degree of the user and the mobile phone hardware information.
In some embodiments, the plurality of game software to be recommended may be output through a game determination model.
The game determination model is a deep neural network model, which includes a deep neural network (Deep Neural Networks, DNN). The input of the game determination model is the operation preference degree of the user, the image quality requirement degree of the user and the game social degree of the user, and the output of the game determination model is a plurality of game software to be recommended.
The game determination model can integrate the operation preference degree of the user, the image quality requirement degree of the user, the game social degree of the user and the mobile phone hardware information to determine a plurality of proper game software to be recommended.
In some embodiments, if the game social degree of the user exceeds a social threshold, processing and outputting social software recommended based on communication interaction data in a user game corresponding to the game social degree of the user by using a social software recommendation model, wherein the social software recommendation model is a deep neural network model, input of the social software recommendation model is communication interaction data in the user game corresponding to the game social degree of the user, and output of the social software recommendation model is recommended social software.
If the game social degree of the user exceeds the social threshold, the social desire of the user is indicated to be strong, and the social software can be directly recommended.
In some embodiments, the social software recommendation model may include a word processing sub-model, a picture processing sub-model, a speech processing sub-model, a synthetic output model. The word processing sub-model, the picture processing sub-model, the voice processing sub-model and the comprehensive output model are all deep neural network models. The input of the word processing sub-model is word interaction data, the output of the word processing sub-model is word interaction degree, the input of the picture processing sub-model is picture interaction data, the output of the picture processing sub-model is picture interaction degree, the input of the voice processing sub-model is voice interaction data, and the output of the voice processing sub-model is voice interaction degree. The text interaction degree can be a numerical value between 0 and 1, and the larger the numerical value is, the higher the interaction frequency of the user with other users through the text is. The voice interaction degree can be a value between 0 and 1, and the larger the value is, the higher the frequency of interaction between the user and other users through voice is. The interaction degree of the pictures can be a numerical value between 0 and 1, and the larger the numerical value is, the higher the interaction frequency of the user with other users through the pictures is. Because the social software basically has the functions of text chat, voice chat and picture chat during chat and social contact, but the emphasis of each social software is different, for example, some social software highlights the voice chat function, some social software highlights the text chat function and some social software highlights the picture chat function. The corresponding social software can be processed and selected based on the text interaction degree, the picture interaction degree and the voice interaction degree. For example, the greater the degree of voice interaction, the more prominent the voice chat function is recommended to the social software. The comprehensive output model can carry out comprehensive processing to determine corresponding social software based on the text interaction degree, the picture interaction degree and the voice interaction degree. The input of the comprehensive output model is text interaction degree, picture interaction degree and voice interaction degree, and the output of the comprehensive output model is recommended social software.
Based on the same inventive concept, fig. 2 is a schematic diagram of a software recommendation system based on big data analysis and artificial intelligence according to an embodiment of the present invention, where the software recommendation system based on big data analysis and artificial intelligence includes:
an acquisition module 21, configured to acquire a running video of an installed game, setting information of the installed game, and mobile phone hardware information;
a touch frequency module 22 for determining a game touch frequency of a user using a touch frequency determination model based on a running video of the installed game;
a preference degree determining module 23, configured to determine an operation preference degree of a user based on a touch frequency of the user game;
an image quality requirement determining module 24 for determining an image quality requirement level of a user based on the setting information of the installed game;
an exchange interaction module 25, configured to determine exchange interaction data in the user game using an exchange interaction model based on the running video of the installed game;
a game social module 26 for determining a game social level of a user based on the communication interaction data in the user game;
a recommendation module 27, configured to determine a plurality of game software to be recommended using a game determination model based on the operation preference degree of the user, the image quality requirement degree of the user, the game social degree of the user, and the mobile phone hardware information.