Player Intelligent Matching System and Method Based on Big Data

Abstract

The present application discloses a player intelligent matching system and method based on big data. By using artificial intelligence technology based on deep learning, semantic information mining and clustering analysis are respectively performed on the game history records of the first game player and the second game player, and the overall game behavior characteristics of the first game player and the second game player are respectively extracted. Then, through feature interaction response analysis of the two, it is intelligently determined whether the first game player should be matched with the second game player. In this way, more personalized player matching can be achieved, which helps to improve the interactivity and user experience of the game, and at the same time is also beneficial to improving the operation efficiency and commercial value of the game.

Description

Technical Field

[0001] The present application relates to the field of intelligent matching, and more specifically, to a player intelligent matching system and method based on big data. Background Art

[0002] With the rapid development of Internet technology, online games have become one of the important ways for people to relax and entertain, and its user base is becoming increasingly large and diverse. In this context, how to improve players' gaming experience and promote effective interaction between players has become the focus of game developers and operators.

[0003] Traditional player matching systems are often based on simple rules or statistical features, such as matching by level, win rate, region, online time and other factors. Although these methods can meet basic needs to a certain extent, they have obvious shortcomings in accuracy, personalization and efficiency.

[0004] In the current gaming environment, players’ behavioral data is massive and diverse, including but not limited to game character selection, social behavior, game time, consumption behavior, etc. These behavioral data contain a lot of information about players’ interests, habits, and preferences. However, traditional matching methods fail to fully tap and utilize these data, resulting in matching results that often fail to meet players’ personalized needs, and may miss potential high-quality matches, which in turn affects players’ retention and activity.

[0005] Therefore, we look forward to an optimized player intelligent matching system and method based on big data. Summary of the invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a player intelligent matching system and method based on big data, which respectively mines semantic information and performs cluster analysis on the game history records of the first game player and the second game player by using artificial intelligence technology based on deep learning, extracts the overall game behavior characteristics of the first game player and the second game player respectively, and then performs feature interaction response analysis on the two, so as to intelligently determine whether the first game player and the second game player should be matched. In this way, more personalized player matching can be achieved, which helps to improve the interactivity and user experience of the game, and is also conducive to improving the operational efficiency and commercial value of the game.

[0007] According to one aspect of the present application, a player intelligent matching system based on big data is provided, which includes:

[0008] A player game history acquisition module is used to acquire the game history of a first game player and the game history of a second game player;

[0009] a game record semantic coding module, configured to semantically code each game record in the game history record of the first game player and each game record in the game history record of the second game player to obtain a set of first game record semantic coding feature vectors and a set of second game record semantic coding feature vectors;

[0010] A feature distribution cluster analysis module, used to perform feature distribution cluster analysis on the set of the first game record semantic coding feature vectors and the set of the second game record semantic coding feature vectors to obtain a first game player game behavior semantic cluster representation vector and a second game player game behavior semantic cluster representation vector;

[0011] An interactive response analysis module, used for performing feature interactive response analysis on the first game player's game behavior semantic cluster representation vector and the second game player's game behavior semantic cluster representation vector to obtain a first game player-second game player game behavior semantic response joint representation vector;

[0012] The matching result generation module is used to determine whether to push the first game player to the second game player based on the first game player-second game player game behavior semantic response joint representation vector.

[0013] According to another aspect of the present application, a player intelligent matching method based on big data is provided, which includes:

[0014] Obtaining the game history records of the first game player and the game history records of the second game player;

[0015] Semantically encoding each game record in the game history record of the first game player and each game record in the game history record of the second game player to obtain a set of first game record semantic encoding feature vectors and a set of second game record semantic encoding feature vectors;

[0016] Performing feature distribution cluster analysis on the set of the first game record semantic coding feature vectors and the set of the second game record semantic coding feature vectors respectively to obtain a first game player game behavior semantic cluster representation vector and a second game player game behavior semantic cluster representation vector;

[0017] Performing feature interaction response analysis on the first game player's game behavior semantic cluster representation vector and the second game player's game behavior semantic cluster representation vector to obtain a first game player-second game player game behavior semantic response joint representation vector;

[0018] Based on the first game player-second game player game behavior semantic response joint representation vector, determine whether to push the first game player to the second game player.

[0019] Compared with the prior art, the player intelligent matching system and method based on big data provided by the present application respectively mines semantic information and performs cluster analysis on the game history records of the first game player and the second game player by using artificial intelligence technology based on deep learning, extracts the overall game behavior characteristics of the first game player and the second game player respectively, and then intelligently determines whether the first game player and the second game player should be matched by analyzing the characteristic interaction response of the two. In this way, more personalized player matching can be achieved, which helps to improve the interactivity and user experience of the game, and is also conducive to improving the operational efficiency and commercial value of the game. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0021] Figure 1 A block diagram of a player intelligent matching system based on big data according to an embodiment of the present application;

[0022] Figure 2 A data flow diagram of a player intelligent matching system based on big data according to an embodiment of the present application;

[0023] Figure 3 The present invention is a flowchart of a method for intelligently matching players based on big data according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0025] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0026] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0027] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0028] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0029] Traditional player matching systems are often based on simple rules or statistical features, such as matching by level, win rate, region, online time and other factors. Although these methods can meet basic needs to a certain extent, they have obvious deficiencies in accuracy, personalization and efficiency. In the current gaming environment, players' behavioral data is massive and diverse, including but not limited to game character selection, social behavior, game time, consumption behavior, etc. These behavioral data contain a large amount of information about players' interests, habits and preferences. However, traditional matching methods fail to fully explore and utilize these data, resulting in matching results that often fail to meet the personalized needs of players, and may miss potential high-quality matches, thereby affecting player retention and activity. Therefore, an optimized player intelligent matching system and method based on big data is expected.

[0030] In the technical solution of the present application, a player intelligent matching system based on big data is proposed. Figure 1 It is a block diagram of a player intelligent matching system based on big data according to an embodiment of the present application. Figure 2 : is a data flow diagram of a player intelligent matching system based on big data according to an embodiment of the present application. Figure 1 and Figure 2As shown, according to an embodiment of the present application, a player intelligent matching system 300 based on big data includes: a player game history acquisition module 310, which is used to acquire the game history of a first game player and the game history of a second game player; a game record semantic coding module 320, which is used to semantically encode each game record in the game history of the first game player and each game record in the game history of the second game player to obtain a set of first game record semantic coding feature vectors and a set of second game record semantic coding feature vectors; a feature distribution clustering analysis module 330, which is used to semantically encode the set of first game record semantic coding feature vectors and the set of second game record semantic coding feature vectors; A set of second game record semantic coding feature vectors are respectively subjected to feature distribution cluster analysis to obtain a first game player's game behavior semantic cluster representation vector and a second game player's game behavior semantic cluster representation vector; an interactive response analysis module 340 is used to perform feature interactive response analysis on the first game player's game behavior semantic cluster representation vector and the second game player's game behavior semantic cluster representation vector to obtain a first game player-second game player game behavior semantic response joint representation vector; a matching result generation module 350 is used to determine whether to push the first game player to the second game player based on the first game player-second game player game behavior semantic response joint representation vector.

[0031] In particular, the player game history acquisition module 310 is used to obtain the game history of the first game player and the game history of the second game player. It should be understood that the game history record is an important data source that reflects the player's game behavior and preferences, including but not limited to the player's game scores, the game types he is good at, the characters or equipment he uses, the game time, consumption behavior, social behavior and other multi-dimensional information. Intelligent matching of players based on the player's game history records can ensure that the matched players have high compatibility in terms of skill level, game interests, etc., thereby improving the player's game experience and satisfaction. Based on this, in the technical solution of the present application, first, the game history record of the first game player and the game history record of the second game player are obtained.

[0032] In particular, the game record semantic coding module 320 is used to semantically encode each game record in the game history record of the first game player and each game record in the game history record of the second game player to obtain a set of first game record semantic coding feature vectors and a set of second game record semantic coding feature vectors. That is, in order to fully understand the deep semantic information in the game history record, each game record in the game history record of the first game player and each game record in the game history record of the second game player are further semantically encoded to mine the player's behavior pattern information. In an embodiment of the present application, a pre-trained deep learning model, such as BERT or Transformer, can be used to perform embedding coding and contextual semantic association understanding on each game record to capture the complex game behavior semantic features in each game record, and convert the textual game records into feature vector form, thereby generating a set of first game record semantic coding feature vectors and a set of second game record semantic coding feature vectors.

[0033] In particular, the feature distribution cluster analysis module 330 is used to perform feature distribution cluster analysis on the set of the first game record semantic encoding feature vectors and the set of the second game record semantic encoding feature vectors to obtain the first game player game behavior semantic cluster representation vector and the second game player game behavior semantic cluster representation vector. In a specific example of the present application, the set of the first game record semantic encoding feature vectors and the set of the second game record semantic encoding feature vectors are respectively input into the feature distribution cluster analysis network based on reverse mask enhancement to obtain the first game player game behavior semantic cluster representation vector and the second game player game behavior semantic cluster representation vector. Considering that there may be individual abnormal or accidental behaviors in the player's game history records, they cannot represent their real interests. Therefore, in order to extract representative core features of player game behavior from multiple game records, the set of the first game record semantic encoding feature vectors and the set of the second game record semantic encoding feature vectors are further clustered to reveal the main game behavior patterns of the first game player and the second game player. In the technical solution of the present application, a feature distribution cluster analysis network based on reverse mask enhancement is introduced to implement this clustering process. Specifically, the feature distribution cluster analysis network dynamically allocates weights based on the semantic similarity of each game record semantic coding feature vector relative to the cluster center of the set, and uses a reverse masking mechanism to strengthen the feature representation that deviates from the cluster center in the set of game record semantic coding feature vectors, that is, highlighting the abnormal or unique behavior of the player, and then removes accidental behavior features from the set of game record semantic coding feature vectors through positional difference operations to achieve feature purification. Finally, the set of purified and optimized game record semantic coding feature vectors is calculated by positional mean to obtain a refined representation of the player's long-term stable game behavior pattern, thereby obtaining the first game player game behavior semantic cluster representation vector and the second game player game behavior semantic cluster representation vector.

[0034] In an embodiment of the present application, a feature distribution cluster analysis is performed on the set of the first game record semantic coding feature vectors and the set of the second game record semantic coding feature vectors to obtain a first game player game behavior semantic cluster representation vector and a second game player game behavior semantic cluster representation vector, including: clustering the set of the first game record semantic coding feature vectors to obtain a first game record semantic feature cluster center vector; calculating the semantic correlation coefficient between each first game record semantic coding feature vector in the set of the first game record semantic coding feature vectors and the first game record semantic feature cluster center vector to obtain a sequence of semantic correlation coefficients; taking the reciprocal of each semantic correlation coefficient in the sequence of semantic correlation coefficients to obtain a sequence of semantic anti-correlation coefficients; using Sigma The id function normalizes the sequence of semantic anti-correlation coefficients to obtain a sequence of semantic anti-correlation weight coefficients; using each semantic anti-correlation weight coefficient in the sequence of semantic anti-correlation weight coefficients as a weight, weighting each first game record semantic coding feature vector in the set of the first game record semantic coding feature vectors to obtain a set of suppressed first game record semantic coding feature vectors; calculating the positional difference between the set of the first game record semantic coding feature vectors and the set of suppressed first game record semantic coding feature vectors to obtain a set of optimized first game record semantic coding feature vectors; calculating the positional mean vector of the set of optimized first game record semantic coding feature vectors to obtain the first game player game behavior semantic clustering representation vector.

[0035] Among them, the process of clustering the set of the first game record semantic coding feature vectors to obtain the first game record semantic feature clustering center vector includes: calculating the positional mean vector of the set of the first game record semantic coding feature vectors to obtain the first game record semantic feature clustering center vector; and the process of calculating the semantic correlation coefficient between each first game record semantic coding feature vector in the set of the first game record semantic coding feature vectors and the first game record semantic feature clustering center vector to obtain a sequence of semantic correlation coefficients includes: calculating the cosine similarity between each first game record semantic coding feature vector in the set of the first game record semantic coding feature vectors and the first game record semantic feature clustering center vector as the semantic correlation coefficient to obtain a sequence of semantic correlation coefficients.

[0036] In summary, in the above embodiment, the set of the first game record semantic coding feature vectors and the set of the second game record semantic coding feature vectors are respectively subjected to feature distribution cluster analysis to obtain the first game player game behavior semantic cluster representation vector and the second game player game behavior semantic cluster representation vector, including: processing the set of the first game record semantic coding feature vectors with the following reverse mask aggregation formula to obtain the first game player game behavior semantic cluster representation vector, wherein the reverse mask aggregation formula is:

[0037] X={x (1) ,x (2) ,…,x (N)}

[0038]

[0039] Y={y (1) ,y (2) ,…,y (N)},y (i) =w (i) ·x (i)

[0040] Z={z (1) ,z (2) ,…,z (N)}, z (i) =x (i) -y (i)

[0041]

[0042] Wherein, X is the set of semantic encoding feature vectors of the first game records, x (1) 、x (2) 、x (i) and x (N) are the first, second, i-th and N-th first game record semantic encoding feature vectors in the set of the first game record semantic encoding feature vectors, respectively, and the value of N is the number of feature vectors in the set of the first game record semantic encoding feature vectors, (·) T represents the transpose of the feature vector, c is the center vector of the semantic feature cluster of the first game record, ||·|| represents the modulus of the vector, and r (i) represents the semantic correlation coefficient between the i-th first game record semantic encoding feature vector and the first game record semantic feature clustering center vector, ρ (i) is the i-th semantic anti-correlation coefficient, w (i) is the i-th semantic anti-correlation weight coefficient, e (·) represents an exponential function with a natural constant as the base, Y is the set of semantic encoding feature vectors that suppress the first game record, y(1) ,y (2) ,y (i) and (N) are respectively the first, second, i-th and N-th first game record semantic coding feature vectors in the set of the first game record semantic coding feature vectors, Z is the set of the first game record semantic coding feature vectors, z (1) 、z (2) 、z (i) and z (N) are respectively the first, second, i-th and N-th optimized first game record semantic encoding feature vectors in the set of optimized first game record semantic encoding feature vectors, v a A semantic clustering representation vector for the game behavior of the first game player.

[0043] In particular, the interactive response analysis module 340 is used to perform feature interactive response analysis on the first game player's game behavior semantic cluster representation vector and the second game player's game behavior semantic cluster representation vector to obtain a first game player-second game player game behavior semantic response joint representation vector. In a specific example of the present application, in order to explore the behavioral interactivity between the first game player and the second game player, the present application introduces a feature interactive response module based on an adaptive distinguishable mechanism to perform feature interactive matching analysis on the first game player's game behavior semantic cluster representation vector and the second game player's game behavior semantic cluster representation vector. Specifically, the first game player's game behavior semantic cluster representation vector and the second game player's game behavior semantic cluster representation vector are input into the feature interactive response module based on an adaptive distinguishable mechanism to obtain the first game player-second game player game behavior semantic response joint representation vector.

[0044] In an embodiment of the present application, a feature interaction response analysis is performed on the first game player's game behavior semantic cluster representation vector and the second game player's game behavior semantic cluster representation vector to obtain a first game player-second game player game behavior semantic response joint representation vector, including: first calculating the position-by-position response between the first game player's game behavior semantic cluster representation vector and the second game player's game behavior semantic cluster representation vector to obtain a player game behavior semantic feature position-by-position response representation vector; that is, first calculating the position-by-position response feature between the game behavior semantic cluster representation vectors of the first game player and the second game player through the feature interaction response module to establish behavioral correlation between the players. Next, the Softmax function is used to normalize the position-by-position response representation vector of the player's game behavior semantic feature to obtain a normalized position-by-position response weight vector of the player's game behavior semantic feature; and the normalized position-by-position response weight vector of the player's game behavior semantic feature is input into a learnable gating function to obtain a player's game behavior semantic feature response weight screening mask vector; here, the response weight is generated by the normalization operation, and the gating mechanism is further introduced to adaptively output the weight mask, and the weight distribution is further refined by performing a mask operation on the generated response weight to improve the distinguishability of feature selection. Then, the position-by-position point multiplication between the player's game behavior semantic feature response weight screening mask vector and the normalized position-by-position response weight vector of the player's game behavior semantic feature is calculated to obtain a distinguishable weight mask vector of the player's game behavior semantic feature position-by-position response; then the position-by-position point multiplication between the player's game behavior semantic feature position-by-position response distinguishable weight mask vector and the player's game behavior semantic feature position-by-position response representation vector is calculated to obtain the first game player-second game player game behavior semantic response joint representation vector. That is, the original position-by-position response features are weighted by position using the response weights after mask filtering, so as to achieve information screening in the feature association response analysis process, highlight the similarities and differences in the game behaviors of the first game player and the second game player, and thus generate a joint representation vector of the semantic responses of the game behaviors of the first game player and the second game player.

[0045] In summary, in the above embodiment, the feature interaction response analysis is performed on the first game player game behavior semantic cluster representation vector and the second game player game behavior semantic cluster representation vector to obtain the first game player-second game player game behavior semantic response joint representation vector, including: processing the first game player game behavior semantic cluster representation vector and the second game player game behavior semantic cluster representation vector using the following feature interaction response formula to obtain the first game player-second game player game behavior semantic response joint representation vector, wherein the feature interaction response formula is:

[0046] vr =v b / v a

[0047] v n =softmax(v r )

[0048]

[0049] Among them, v a represents the semantic clustering representation vector of the first game player’s game behavior, v b represents the semantic clustering representation vector of the second game player’s game behavior, v r represents the position-by-position response representation vector of the semantic features of the player's game behavior, softmax is a normalized exponential function, v n represents the normalized player game behavior semantic feature position-by-position response weight vector, v e represents the weight mask vector of the response screening of the semantic feature of the player's game behavior, exp(·) represents the exponential function operation with e as the base, ⊙ represents the point multiplication by position, A joint representation vector representing the semantic responses of the game behaviors of the first game player and the second game player.

[0050] In particular, the matching result generation module 350 is used to determine whether to push the first game player to the second game player based on the first game player-second game player game behavior semantic response joint representation vector. In a specific example of the present application, the first game player-second game player game behavior semantic response joint representation vector is input into a classifier-based intelligent matcher to obtain a matching result, and the matching result is used to indicate whether to push the first game player to the second game player. That is, the classifier model is used to learn the feature pattern of the first game player-second game player game behavior semantic response joint representation vector, analyze the compatibility of the first game player and the second game player in terms of game behavior features, and classify and map them to determine whether to match the first game player with the second game player.

[0051] Preferably, the applicant of the present application takes into account that the first game player's game behavior semantic cluster representation vector and the second game player's game behavior semantic cluster representation vector respectively express the semantic encoding reverse mask enhanced feature distribution cluster analysis features of the game history records of the first game player and the second game player. In this way, when the first game player's game behavior semantic cluster representation vector and the second game player's game behavior semantic cluster representation vector are input into the feature interaction response module based on the adaptive distinguishable mechanism, the obtained first game player-second game player game behavior semantic response joint representation vector will also lead to outlier fusion class reasoning mapping deviation due to insufficient coverage of adaptive distinguishable interaction responses caused by differences in semantic feature distribution of different players' game history records, thereby affecting the accuracy of the matching result obtained by the first game player-second game player game behavior semantic response joint representation vector through a classifier-based intelligent matcher.

[0052] Therefore, in a preferred example, inputting the first game player-second game player game behavior semantic response joint representation vector into a classifier-based intelligent matcher to obtain a matching result includes:

[0053] Calculate the feature mean of the first game player-second game player game behavior semantic response joint representation vector, and divide the feature mean by the difference between the maximum eigenvalue and the minimum eigenvalue of the first game player-second game player game behavior semantic response joint representation vector to obtain the first game player-second game player game behavior semantic response joint representation distribution representation value;

[0054] Subtract the first game player-second game player game behavior semantic response joint representation distribution representation value from one and divide it by the first game player-second game player game behavior semantic response joint representation distribution representation value to obtain the first game player-second game player game behavior semantic response joint representation distribution modulation value;

[0055] Activating the first game player-second game player game behavior semantic response joint representation vector through a probabilistic function to obtain a probabilistic first game player-second game player game behavior semantic response joint representation vector;

[0056] Subtract the probabilistic first game player-second game player game behavior semantic response joint representation vector from the first game player-second game player game behavior semantic response joint representation distribution modulation value, take the absolute value and calculate the negative of the logarithmic value with base 2 to obtain the probabilistic first game player-second game player game behavior semantic response joint representation distribution modulation information vector;

[0057] After dividing the first game player-second game player game behavior semantic response joint representation distribution characterization value by one minus the difference of each eigenvalue of the probabilistic first game player-second game player game behavior semantic response joint representation vector, summing all eigenvalues ​​of the probabilistic first game player-second game player game behavior semantic response joint representation vector and dividing it by the length of the first game player-second game player game behavior semantic response joint representation vector to obtain the probabilistic first game player-second game player game behavior semantic response joint representation distribution modulation bias value;

[0058] Performing point multiplication of the probabilistic first game player-second game player game behavior semantic response joint representation distribution modulation information vector and the probabilistic first game player-second game player game behavior semantic response joint representation distribution modulation bias value and a weight as a hyperparameter to obtain an optimized first game player-second game player game behavior semantic response joint representation vector; and

[0059] The optimized first game player-second game player game behavior semantic response joint representation vector is input into a classifier-based intelligent matcher to obtain a matching result.

[0060] The optimized first game player-second game player game behavior semantic response joint representation vector is expressed as:

[0061]

[0062] And among them:

[0063]

[0064] in, The feature mean of the joint representation vector of the semantic response of the game behavior of the first game player and the second game player, v max and v min represents the maximum eigenvalue and the minimum eigenvalue in the joint representation vector of the semantic response to the game behavior of the first game player and the second game player, respectively; p represents the distribution representation value of the joint representation of the semantic response to the game behavior of the first game player and the second game player; V represents the probabilistic joint representation vector of the semantic response to the game behavior of the first game player and the second game player obtained by activating the joint representation vector of the semantic response to the game behavior of the first game player and the second game player through a probabilistic function; v irepresents the i-th eigenvalue of the probabilistic first game player-second game player game behavior semantic response joint representation vector, log represents the logarithmic function with base 2, ε is the weight as a hyperparameter, the value of L is the length of the probabilistic first game player-second game player game behavior semantic response joint representation vector, and V' represents the optimized first game player-second game player game behavior semantic response joint representation vector.

[0065] That is, the probability information distribution planning based on the eigenvalue of the first game player-second game player game behavior semantic response joint representation vector is performed through the Bernoulli probability modulation distribution of the first game player-second game player game behavior semantic response joint representation vector relative to the eigenvalue distribution, and the probability reverse mapping of the probability characteristics of the first game player-second game player game behavior semantic response joint representation vector as a whole is used as the extended coverage of the set mapping space of the first game player-second game player game behavior semantic response joint representation vector, so as to independently understand the interaction path between the intuitive probability information distribution and the abstract probability space mapping of the first game player-second game player game behavior semantic response joint representation vector, so as to improve the accuracy of the matching result obtained by the optimized first game player-second game player game behavior semantic response joint representation vector input into the classifier-based intelligent matcher by avoiding the counterfactual reasoning mapping of the outlier feature distribution of the first game player-second game player game behavior semantic response joint representation vector to the class probability.

[0066] As described above, the player intelligent matching system 300 based on big data according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with a player intelligent matching algorithm based on big data. In a possible implementation, the player intelligent matching system 300 based on big data according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the player intelligent matching system 300 based on big data can be a software module in the operating system of the wireless terminal, or it can be an application developed for the wireless terminal; of course, the player intelligent matching system 300 based on big data can also be one of the many hardware modules of the wireless terminal.

[0067] Alternatively, in another example, the big data-based player intelligent matching system 300 and the wireless terminal may also be separate devices, and the big data-based player intelligent matching system 300 may be connected to the wireless terminal via a wired and / or wireless network, and transmit interactive information in accordance with an agreed data format.

[0068] Furthermore, a player intelligent matching method based on big data is also provided.

[0069] Figure 3 Flowchart of a method for intelligent player matching based on big data according to an embodiment of the present application. Figure 3 As shown, according to the embodiment of the present application, the player intelligent matching method based on big data includes the following steps: S1, obtaining the game history record of the first game player and the game history record of the second game player; S2, semantically encoding each game record in the game history record of the first game player and each game record in the game history record of the second game player to obtain a set of first game record semantic encoding feature vectors and a set of second game record semantic encoding feature vectors; S3, performing feature distribution clustering analysis on the set of the first game record semantic encoding feature vectors and the set of the second game record semantic encoding feature vectors to obtain a first game player game behavior semantic clustering representation vector and a second game player game behavior semantic clustering representation vector; S4, performing feature interaction response analysis on the first game player game behavior semantic clustering representation vector and the second game player game behavior semantic clustering representation vector to obtain a first game player-second game player game behavior semantic response joint representation vector; S5, determining whether to push the first game player to the second game player based on the first game player-second game player game behavior semantic response joint representation vector.

[0070] In summary, the player intelligent matching method based on big data according to the embodiment of the present application is explained, which uses artificial intelligence technology based on deep learning to perform semantic information mining and cluster analysis on the game history records of the first game player and the second game player, respectively, to extract the overall game behavior characteristics of the first game player and the second game player, and then through the feature interaction response analysis of the two, to intelligently determine whether the first game player and the second game player should be matched. In this way, more personalized player matching can be achieved, which helps to improve the interactivity and user experience of the game, and is also conducive to improving the operational efficiency and commercial value of the game.

[0071] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A player intelligent matching system based on big data, characterized in that: include: A player game history acquisition module is used to acquire the game history of a first game player and the game history of a second game player; a game record semantic coding module, configured to semantically code each game record in the game history record of the first game player and each game record in the game history record of the second game player to obtain a set of first game record semantic coding feature vectors and a set of second game record semantic coding feature vectors; A feature distribution cluster analysis module, used to perform feature distribution cluster analysis on the set of the first game record semantic coding feature vectors and the set of the second game record semantic coding feature vectors to obtain a first game player game behavior semantic cluster representation vector and a second game player game behavior semantic cluster representation vector; An interactive response analysis module, used for performing feature interactive response analysis on the first game player's game behavior semantic cluster representation vector and the second game player's game behavior semantic cluster representation vector to obtain a first game player-second game player game behavior semantic response joint representation vector; A matching result generation module, configured to determine whether to push the first game player to the second game player based on a joint representation vector of semantic responses to game behaviors of the first game player and the second game player; Wherein, the interactive response analysis module includes: A position-by-position response unit, used for calculating the position-by-position response between the first game player game behavior semantic cluster representation vector and the second game player game behavior semantic cluster representation vector to obtain a player game behavior semantic feature position-by-position response representation vector; A normalization unit, used for normalizing the player game behavior semantic feature position-by-position response representation vector using a Softmax function to obtain a normalized player game behavior semantic feature position-by-position response weight vector; A mask generating unit, used for inputting the normalized player game behavior semantic feature position-by-position response weight vector into a learnable gating function to obtain a player game behavior semantic feature response weight screening mask vector; A response weight masking unit, used for calculating the position-by-position point multiplication between the player game behavior semantic feature response weight screening mask vector and the normalized player game behavior semantic feature position-by-position response weight vector to obtain a player game behavior semantic feature position-by-position response distinguishable weight mask vector; A response weight applying unit is used to calculate the position point multiplication between the distinguishable weight mask vector of the player game behavior semantic feature position response and the player game behavior semantic feature position response representation vector to obtain the first game player-second game player game behavior semantic response joint representation vector.

2. The player intelligent matching system based on big data according to claim 1 is characterized in that: The feature distribution cluster analysis module is used to: The set of the first game record semantic encoding feature vectors and the set of the second game record semantic encoding feature vectors are respectively input into a feature distribution clustering analysis network based on reverse mask enhancement to obtain the first game player game behavior semantic clustering representation vector and the second game player game behavior semantic clustering representation vector.

3. The player intelligent matching system based on big data according to claim 2 is characterized in that: The feature distribution cluster analysis module comprises: A cluster center calculation unit, configured to cluster the set of the first game record semantic coding feature vectors to obtain a first game record semantic feature cluster center vector; a semantic correlation measurement unit, configured to calculate a semantic correlation coefficient between each first game record semantic coding feature vector in the set of the first game record semantic coding feature vectors and the first game record semantic feature clustering center vector to obtain a sequence of semantic correlation coefficients; A semantic anti-correlation coefficient calculation unit, used for taking the reciprocal of each semantic correlation coefficient in the sequence of semantic correlation coefficients to obtain a sequence of semantic anti-correlation coefficients; A weighting unit, used for normalizing the sequence of semantic anti-correlation coefficients using a Sigmoid function to obtain a sequence of semantic anti-correlation weight coefficients; a reverse enhancement unit, configured to weight each first game record semantic coding feature vector in the set of the first game record semantic coding feature vectors by using each semantic anti-correlation weight coefficient in the sequence of the semantic anti-correlation weight coefficients as a weight to obtain a set of suppressed first game record semantic coding feature vectors; A feature purification optimization unit, used for calculating the position difference between the set of the first game record semantic coding feature vectors and the set of the suppressed first game record semantic coding feature vectors to obtain a set of optimized first game record semantic coding feature vectors; The feature clustering representation unit is used to calculate the positional mean vector of the set of the optimized first game record semantic encoding feature vectors to obtain the first game player game behavior semantic clustering representation vector.

4. The player intelligent matching system based on big data according to claim 3 is characterized in that: The cluster center calculation unit is used for: The position-wise mean vector of the set of the first game record semantic encoding feature vectors is calculated to obtain the first game record semantic feature cluster center vector.

5. The player intelligent matching system based on big data according to claim 4 is characterized in that: The semantic relevance measurement unit is used to: The cosine similarity between each first game record semantic encoding feature vector in the set of the first game record semantic encoding feature vectors and the first game record semantic feature clustering center vector is calculated as the semantic correlation coefficient to obtain a sequence of the semantic correlation coefficients.

6. The player intelligent matching system based on big data according to claim 5 is characterized in that: The interactive response analysis module is used to: The first game player's game behavior semantic clustering representation vector and the second game player's game behavior semantic clustering representation vector are input into a feature interaction response module based on an adaptive distinguishable mechanism to obtain the first game player-second game player game behavior semantic response joint representation vector.

7. The player intelligent matching system based on big data according to claim 6 is characterized in that: The matching result generating module is used for: The first game player-second game player game behavior semantic response joint representation vector is input into a classifier-based intelligent matcher to obtain a matching result, and the matching result is used to indicate whether to push the first game player to the second game player.

8. A player intelligent matching method based on big data, characterized in that: include: Obtaining the game history records of the first game player and the game history records of the second game player; Semantically encoding each game record in the game history record of the first game player and each game record in the game history record of the second game player to obtain a set of first game record semantic encoding feature vectors and a set of second game record semantic encoding feature vectors; Performing feature distribution cluster analysis on the set of the first game record semantic coding feature vectors and the set of the second game record semantic coding feature vectors respectively to obtain a first game player game behavior semantic cluster representation vector and a second game player game behavior semantic cluster representation vector; Performing feature interaction response analysis on the first game player's game behavior semantic cluster representation vector and the second game player's game behavior semantic cluster representation vector to obtain a first game player-second game player game behavior semantic response joint representation vector; Determining whether to push the first game player to the second game player based on the first game player-second game player game behavior semantic response joint representation vector; Among them, performing feature interaction response analysis on the first game player's game behavior semantic cluster representation vector and the second game player's game behavior semantic cluster representation vector to obtain a first game player-second game player game behavior semantic response joint representation vector includes: Calculating the position-by-position response between the first game player game behavior semantic cluster representation vector and the second game player game behavior semantic cluster representation vector to obtain a player game behavior semantic feature position-by-position response representation vector; Using a Softmax function to normalize the position-by-position response representation vector of the semantic feature of the player's game behavior to obtain a normalized position-by-position response weight vector of the semantic feature of the player's game behavior; Inputting the normalized player game behavior semantic feature position-by-position response weight vector into a learnable gating function to obtain a player game behavior semantic feature response weight screening mask vector; Calculate the position-wise multiplication between the player game behavior semantic feature response weight screening mask vector and the normalized player game behavior semantic feature position-by-position response weight vector to obtain a player game behavior semantic feature position-by-position response distinguishable weight mask vector; Calculate the position-by-position point multiplication between the distinguishable weight mask vector of the player's game behavior semantic feature position-by-position response and the player's game behavior semantic feature position-by-position response representation vector to obtain the first game player-second game player game behavior semantic response joint representation vector.

Images