Virtual space social interaction engine system based on AI enhancement

Abstract

The invention discloses a virtual space social interaction engine system based on AI enhancement, and relates to the technical field of AI enhancement, the virtual space social interaction engine system constructs a multi-level and three-dimensional user portrait network structure, greatly improves the precision and dimension of user feature expression, and improves the user experience through multi-source fusion analysis of path data, physiological feature data and voice chat data. According to the method, the system can comprehensively capture dominant interests and recessive preferences of users, richer and more accurate user portraits are formed, subsequent user matching is facilitated, positive attraction factors and negative rejection factors can be considered at the same time, the matching accuracy and the satisfaction degree of user matching experience are greatly improved, and the user experience is improved. Meanwhile, by constructing a behavior feature layer, the system can dynamically adjust matching and interaction strategies according to different social scenes, and continuous optimization of system performance is achieved based on a self-adaptive optimization mechanism fed back by a user.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of AI enhancement, in particular to a virtual space social interaction engine system based on AI enhancement. BACKGROUND

[0002] With the rapid development and popularization of virtual reality technology and the concept of metaverse, virtual space has gradually become an important field for people's social interaction. As a core technology platform connecting users and promoting interaction, the performance and experience of the virtual space social system directly affect the social quality and satisfaction of users in the digital world. Therefore, it is necessary to build an intelligent social interaction engine system based on AI enhancement for multi-dimensional user feature capture and accurate matching.

[0003] In the prior art, most virtual space social systems rely on single-dimensional user data collection methods, such as basic personal information or simple behavior tags, which cannot cover multi-level feature recognition of users in virtual environments. Although some existing technologies introduce interest matching algorithms and interaction data analysis, due to the limited data collection dimensions, it is difficult to achieve intelligent judgment of user social preferences and behavior patterns. In complex and variable virtual social scenarios, using only text or voice data for matching is time-consuming, and when facing a large number of users, simple tag matching may result in inefficient matching. At the same time, the existing technology lacks joint collection, analysis and matching linkage capability of multi-modal data such as path behavior, physiological response characteristics and voice emotional characteristics of users in virtual space, lacks recognition mechanism of user aversion factors, and cannot effectively avoid matching that may lead to negative social experience. The existing system often needs to rely on user active feedback for experience optimization, which has the defects of long feedback delay, insufficient data value mining and low optimization efficiency. In terms of perceiving user emotions and social intentions, the existing technology mainly relies on text analysis or simple emoticons, which cannot capture the potential emotional state and social needs of users, resulting in a significant gap between the matching result and the actual expectation of users. At the same time, it lacks adaptive strategies for different social scenarios, and cannot dynamically adjust the matching mechanism according to the behavior characteristics of users in different virtual environments. SUMMARY

[0004] The purpose of the present application is to provide a virtual space social interaction engine system based on AI enhancement, which solves the problems in the background art.

[0005] To solve the above technical problems, the technical scheme adopted by the present application is as follows: A virtual space social interaction engine system based on AI enhancement, comprising: a user portrait construction module, configured to collect path data, physiological feature data and voice chat data of a target user in a virtual space, and construct a portrait network structure of the target user based on the collected data.

[0006] The intelligent interest matching module is used for intelligently matching different users according to the portrait network structure of the target user, and obtaining a matching user group of the target user.

[0007] The user feedback analysis optimization module is used for randomly matching the target user with users in the corresponding matching user group and generating an interest area, and optimizing the portrait network structure of the target user according to the feedback data after matching.

[0008] The present application has the beneficial effect that the present application constructs a multi-level and three-dimensional user portrait network structure, including an interest layer, a behavior aversion layer and a behavior characteristic layer, which greatly improves the precision and dimension of user feature expression compared with the single label portrait in the prior art. Through multi-source fusion analysis of path data, physiological characteristic data and voice chat data, the system can comprehensively capture the explicit interest and implicit preference of the user, form a more rich and accurate user portrait, facilitate subsequent user matching, and realize a two-way consideration matching mechanism by analyzing the interest correlation coefficient and aversion correlation coefficient between users. Compared with the traditional matching method based on common interest, the present application can consider both positive attraction factors and negative repulsion factors, greatly improving the matching accuracy and user matching experience satisfaction. Meanwhile, the present application constructs a behavior characteristic layer through fine analysis of the behavior characteristics and communication methods of users in different areas, so that the system can dynamically adjust the matching and interaction strategy according to different social scenarios, and based on the adaptive optimization mechanism of user feedback, the system can automatically evaluate the satisfaction degree according to the physiological response data after social interaction, and adjust the weight of the interest layer or the fine behavior aversion layer, realizing continuous optimization of system performance. BRIEF DESCRIPTION OF DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0010] Figure 1 The figure is a schematic diagram of the system module of the present application. DETAILED DESCRIPTION

[0011] The technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments only constitute some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0012] Referring to Figure 1 As shown in the figure, the application provides an AI-enhanced virtual space social interaction engine system, comprising a user portrait construction module, an intelligent interest matching module and a user feedback analysis optimization module.

[0013] It should be noted that the user portrait construction module is connected to the intelligent interest matching module, and the intelligent interest matching module is connected to the user feedback analysis optimization module.

[0014] The user portrait construction module is configured to collect path data, physiological feature data and voice chat data of a target user in a virtual space, and construct a portrait network structure of the target user based on the collected data.

[0015] In one specific embodiment, the path data, physiological feature data and voice chat data of the target user in the virtual space are collected by an API interface of a VR virtual space device.

[0016] It should be noted that the path data includes spatial coordinates of each region, time points at which each spatial coordinate appears, and stay duration of each region.

[0017] It should be noted that the physiological feature data includes physiological response data and body feature data.

[0018] It should also be noted that the physiological response data includes blood pressure value, skin electricity response, heart rate variability and other physiological response data.

[0019] It should also be noted that the body feature data includes body posture action data points and other body feature data.

[0020] It should be noted that the voice chat data includes text data, text data appearance time points and voice tone.

[0021] In a specific embodiment of the application, the portrait network structure of the target user is constructed by constructing an interest layer of the target user based on the path data and voice chat data of the target user in the virtual space.

[0022] The behavior aversion layer of the target user is constructed based on the physiological feature data and voice chat data of the target user in the virtual space.

[0023] The behavior feature layer of the target user is constructed based on the path data, physiological feature data and voice chat data of the target user in the virtual space.

[0024] The interest layer, behavior aversion layer and behavior feature layer of the target user are used as the portrait network structure.

[0025] In the specific embodiment of the present application, the interest layer of the target user is constructed, and the specific method is as follows: according to the path data and voice chat data of the target user in the virtual space, and using the TF-IDF algorithm to extract each interest keyword of the target user and the region where each interest keyword appears.

[0026] It should be noted that the each interest keyword includes interest keywords such as liking playing cards, liking drinking, and being keen on running.

[0027] In one specific embodiment, the extraction of each interest keyword of the target user and the region where each interest keyword appears is performed by the following specific extraction method: according to the existing TF-IDF algorithm and the text data in the voice chat data of the target user in the virtual space, each interest keyword of the target user is extracted, the time point of the appearance of each interest keyword of the target user is obtained according to the time point of the appearance of the text data in the voice chat data of the target user in the virtual space, the space coordinates of the appearance of each interest keyword of the target user are mapped, the coordinate interval of each region in the virtual space is obtained from the local database, if the space coordinates of the appearance of a certain interest keyword of the target user are in the coordinate interval of a certain region, the region is taken as the region where the interest keyword of the target user appears, and thus the region where each interest keyword of the target user appears is obtained.

[0028] It should be noted that the local database is used to store the coordinate interval of each region in the virtual space, the characteristic keywords of each interest, the center coordinates of each region in the virtual space, each interest involved in each region in the virtual space, the conventional value of each physiological reaction data of the target user, the physiological feature mutation coefficient threshold, the comprehensive matching coefficient threshold, the appearance proportion threshold, the satisfaction index threshold, and the correlation index threshold.

[0029] According to each interest keyword of the target user, the first love coefficient of the target user for each interest is analyzed.

[0030] In one specific embodiment, the first love coefficient of the target user for each interest is analyzed by the following specific analysis method: the characteristic keywords of each interest are obtained from the local database, the semantic association of each interest keyword of the target user and the characteristic keywords of each interest is compared, the semantic association of each interest keyword of the target user and the characteristic keywords of each interest is obtained, if the semantic association of a certain interest keyword of the target user and the characteristic keyword of a certain interest is greater than a preset association threshold, the interest keyword is taken as the target keyword of the interest, and thus each target keyword of each interest of the target user is screened, and the total number of target keywords of each interest of the target user is counted. wherein n represents the number of each interest, , m is a positive integer greater than 2, and the first love coefficient of the target user for each interest is calculated .

[0031] According to the path data of the target user in the virtual space, each spatial coordinate of each region of the target user in the virtual space is obtained, and the like degree of the target user for each region is analyzed according to the spatial coordinates, and the second love coefficient of the target user for each interest is mapped by weighting the region where the keyword of each interest of the target user appears.

[0032] In one embodiment, the like degree of the target user for each region is analyzed by the following method: the center coordinates of each region in the virtual space are obtained from a local database , the staying time of the target user in each region of the virtual space is extracted according to the path data of the target user in the virtual space , and the spatial coordinates of each region of the target user in the virtual space are obtained , wherein p represents each spatial coordinate, , q is a positive integer greater than 2, and the like degree of the target user for each region is calculated .

[0033] In one embodiment, the second love coefficient of the target user for each interest is mapped by the following method: each interest involved in each region in the virtual space is obtained from a local database, the target keyword of each interest involved in each region in the virtual space is mapped according to the region where the keyword of each interest of the target user appears, and the target keyword of each interest involved in each region in the virtual space is combined with the target keyword of each interest of the target user, and the total number of target keywords of each interest involved in each region in the virtual space is counted , the second love coefficient of the target user for each interest is calculated .

[0034] Based on the first love coefficient and the second love coefficient of the target user for each interest, an interest layer of the target user is generated.

[0035] In one embodiment, the interest layer of the target user is generated by the following method: the comprehensive love coefficient of the target user for each interest is calculated according to the first love coefficient and the second love coefficient of the target user for each interest , and the comprehensive love coefficient of the target user for each interest is taken as the interest layer of the target user.

[0036] In the specific embodiment of the present application, the behavior aversion layer of the target user is constructed, and the specific method is as follows: according to the physiological characteristic data of the target user in the virtual space, the measured values of the physiological response data of the target user when communicating with other users in the virtual space are extracted, the abnormal physiological response time points of the target user are analyzed, and the body posture characteristic data and voice chat data of other users at the abnormal physiological response time points of the target user are extracted.

[0037] In one specific embodiment, the abnormal physiological response time points of the target user are analyzed, and the specific analysis method is as follows: the normal values of the physiological response data of the target user are obtained from the local database , wherein t represents the number of the physiological response data, w is a positive integer greater than 2, the measured values of the physiological response data of the target user at each communication time point with other users are extracted according to the measured values of the physiological response data of the target user when communicating with other users in the virtual space , wherein r represents the communication time point, s is a positive integer greater than 2, the physiological feature mutation coefficients of the target user at each communication time point with other users are calculated The physiological feature mutation coefficient threshold is obtained from the local database, if the physiological feature mutation coefficient of the target user at a certain communication time point with other users is greater than the physiological feature mutation coefficient threshold, the communication time point is marked as an abnormal physiological response time point, and thus the abnormal physiological response time points of the target user are obtained.

[0038] According to the body posture characteristic data and voice chat data of other users at the abnormal physiological response time points of the target user, the behavior data correlation analysis is performed, the aversion actions, aversion tones and aversion words of the target user are obtained, and the behavior aversion layer of the target user is generated.

[0039] In one specific embodiment, the method for obtaining the aversion actions, the aversion tones and the aversion words of the target user and generating the behavior aversion layer of the target user comprises: extracting the body posture action data points of the other users at the physiological abnormal reaction time points of the target user according to the body posture feature data of the other users at the physiological abnormal reaction time points of the target user, and obtaining the actions of the other users at the physiological abnormal reaction time points of the target user by using the existing action analysis method; classifying the actions, and classifying similar actions into one category and counting the number of the actions of the other users at each type of action of the target user; calculating the occurrence proportion of the actions of the other users at each type of action of the target user; obtaining the occurrence proportion threshold from the local database; and regarding the action of the target user as an aversion action if the occurrence proportion of the action of the target user is greater than the occurrence proportion threshold, so as to obtain the aversion actions of the target user.

[0040] According to the speech chat data of the other users at the physiological abnormal reaction time points of the target user, the text data and the tone data of the other users at the physiological abnormal reaction time points of the target user are extracted, and the aversion actions, the aversion tones and the aversion words of the target user are obtained by using the method for obtaining the aversion actions of the target user, so as to obtain the behavior aversion layer of the target user.

[0041] In the specific embodiment of the application, the method for constructing the behavior feature layer of the target user comprises: mapping the physiological feature data and the speech chat data of the target user in each region of the virtual space according to the path data, the physiological feature data and the speech chat data of the target user in the virtual space.

[0042] The physiological feature data of the target user in each region of the virtual space is analyzed to obtain the behavior feature map of the target user in each region.

[0043] In one specific embodiment, the method for obtaining the behavior feature map of the target user in each region comprises: extracting the body posture action data points of the target user at each monitoring time point in each region of the virtual space according to the physiological feature data of the target user in each region of the virtual space, and obtaining the habitual actions of the target user in each region by using the method for obtaining the aversion actions of the target user according to the data, so as to obtain the behavior feature map of the target user in each region.

[0044] The speech chat data of the target user in each region of the virtual space is analyzed to obtain the communication mode feature set of the target user in each region.

[0045] In one specific embodiment, the target user's communication mode feature set in each area is obtained by extracting the text data and intonation data of the target user at each monitoring time point in each area of the virtual space according to the voice chat data of the target user in each area of the virtual space, and obtaining the target user's each habitual intonation and each habitual vocabulary according to the above data and the method of obtaining each aversion action of the target user.

[0046] The behavior feature map and the communication mode feature set of the target user in each area are unified as data of a behavior feature layer, so as to obtain the behavior feature layer of the target user.

[0047] The intelligent interest matching module is configured to perform intelligent interest matching on different users according to the portrait network structure of the target user, so as to obtain a matching user group of the target user.

[0048] In the specific embodiment of the present application, the intelligent interest matching is performed to obtain a matching user group of the target user by extracting the interest layer of the target user according to the portrait network structure of the target user, obtaining the interest layer of other users, and analyzing the interest correlation between the target user and the other users.

[0049] According to the portrait network structure of the target user, the behavior aversion layer and the behavior feature layer of the target user are extracted, and the behavior aversion layer and the behavior feature layer of other users are obtained, and the aversion correlation between the target user and the other users is analyzed.

[0050] Based on the interest correlation and the aversion correlation between the target user and the other users, a comprehensive matching coefficient of the target user and the other users is evaluated.

[0051] In one specific embodiment, the comprehensive matching coefficient of the target user and the other users is evaluated by calculating the interest correlation between the target user and the other users and the aversion correlation between the target user and the other users to obtain the comprehensive matching coefficient of the target user and the other users .

[0052] The comprehensive matching coefficient threshold is obtained from the local database, and based on the comprehensive matching coefficient of the target user and the other users, if the comprehensive matching coefficient of the target user and a certain other user is greater than the comprehensive matching coefficient threshold, the other user is divided into a user in the matching user group of the target user, and so on, so as to obtain the matching user group of the target user.

[0053] In the specific embodiment of the present application, the interest correlation between the target user and the other users is analyzed by analyzing the interest of the target user according to the interest layer of the target user.

[0054] In one embodiment, the analysis of the interest weight of the target user is performed by calculating the interest weight of the target user according to the comprehensive interest coefficient of the target user. In one embodiment, the analysis of the interest weight of the other user is performed by calculating the interest weight of the other user according to the interest weight of the target user.

[0055] In one embodiment, the analysis of the interest weight of the other user is performed by calculating the interest weight of the other user according to the interest weight of the target user.

[0056] In one embodiment, the analysis of the interest weight of the other user is performed by calculating the interest weight of the other user according to the interest weight of the target user.

[0057] In one embodiment, the calculation of the correlation index of the target user and the other user is performed by calculating the correlation index of the target user and the other user according to the interest weight of the target user and the interest weight of the other user.

[0058] In one embodiment, the calculation of the correlation index of the target user and the other user is performed by calculating the correlation index of the target user and the other user according to the interest weight of the target user and the interest weight of the other user. wherein e represents a natural constant.

[0059] In one embodiment, the calculation of the correlation index of the target user and the other user is performed by calculating the correlation index of the target user and the other user according to the interest weight of the target user and the interest weight of the other user.

[0060] In one embodiment, the calculation of the correlation index of the target user and the other user is performed by calculating the correlation index of the target user and the other user according to the interest weight of the target user and the interest weight of the other user.

[0061] In one embodiment, the analysis of the aversion correlation coefficient of the target user and the other user is performed by performing an aversion action correlation analysis according to the behavior aversion layer and the interest layer of the target user, and combining the behavior characteristic layer and the interest layer of the other user, so as to obtain a first aversion index of the target user and the other user.

[0062] ​​​In one specific embodiment, the aversion action correlation analysis is performed to obtain the first aversion index of the target user and the other user, and the specific method is as follows: according to the behavior aversion layer of the target user, each aversion action of the target user is extracted, according to the behavior characteristic layer of the other user, each habitual action of the other user in each region is extracted, and according to the interest layer of the target user and the interest layer of the other user, the target interest region between the target user and the other user is evaluated, and each habitual action of the other user in the target interest region is extracted, if a certain aversion action of the target user is consistent with a certain habitual action of the other user in the target interest region, the first aversion index is added by 1, and the first aversion index is initially 0, and the first aversion index of the target user and the other user is obtained in this way.

[0063] In one specific embodiment, the target interest region between the target user and the other user is evaluated, and the specific evaluation method is as follows: a certain other user is randomly selected as a matching user, and a comprehensive love coefficient of the matching user for each interest is extracted, according to each interest involved in each region in the virtual space, a comprehensive love coefficient of the matching user for each interest involved in each region in the virtual space is mapped, a comprehensive love coefficient sum of the matching user for each region is obtained by cumulative statistics , and a comprehensive love coefficient sum of the target user for each region is obtained in the same way , an associated love coefficient of the target user and the matching user for each region is calculated , and the region with the largest associated love coefficient is selected as the target interest region.

[0064] According to the behavior characteristic layer and the interest layer of the target user, and in combination with the behavior aversion layer and the interest layer of the other user, aversion action correlation analysis is performed to obtain the second aversion index of the target user and the other user.

[0065] In one specific embodiment, the aversion action correlation analysis is performed to obtain the second aversion index of the target user and the other user, and the specific method is as follows: according to the method of obtaining the first aversion index of the target user and the other user, the second aversion index of the target user and the other user can be obtained in the same way.

[0066] According to the first aversion index and the second aversion index of the target user and the other user, an aversion association coefficient of the target user and the other user is calculated.

[0067] In one specific embodiment, the aversion association coefficient of the target user and the other user is calculated, and the specific calculation method is as follows: the first aversion index and the second aversion index of the target user and the other user are added, and the aversion association coefficient of the target user and the other user is obtained.

[0068] The user feedback analysis optimization module is configured to randomly match the target user with users in the corresponding matching user group and generate an interest area, and optimize the portrait network structure of the target user according to the matched feedback data.

[0069] In one specific embodiment, the interest area generation operation is specifically performed by obtaining the interest area according to an evaluation method of the target interest area between the target user and other users, and generating the interest area in the virtual space.

[0070] In one specific embodiment of the present application, the optimization of the portrait network structure of the target user according to the matched feedback data is specifically performed by collecting measured values of physiological response data of the target user in the interaction process with the users in the matching user group, and analyzing a satisfaction index of the target user with respect to the matching user group according to the measured values.

[0071] In one specific embodiment, the analysis of the satisfaction index of the target user with respect to the matching user group is specifically performed by calculating the physiological feature mutation coefficients of the target user and the users in the matching user group at each communication time point according to the measured values of the physiological response data of the target user in the interaction process with the users in the matching user group, and calculating the physiological feature mutation coefficients of the target user and the users in the matching user group at each communication time point according to a method of calculating the physiological feature mutation coefficients of the target user and other users at each communication time point, and performing cumulative statistics to obtain a total value of the physiological feature mutation coefficients of the target user and the users in the matching user group in the interaction process, and taking the total value as the satisfaction index of the target user with respect to the matching user group.

[0072] Based on the satisfaction index of the target user with respect to the matching user group, the interest layer and the behavior aversion layer in the portrait network structure are adjusted correspondingly.

[0073] In one specific embodiment of the present application, the adjustment of the interest layer and the behavior aversion layer in the portrait network structure is specifically performed by obtaining a satisfaction index threshold from a local database, and adjusting the interest layer in the portrait network structure positively if the satisfaction index of the target user with respect to the matching user group is greater than the satisfaction index threshold, and adjusting the behavior aversion layer in the portrait network structure in a feature refinement manner if the satisfaction index of the target user with respect to the matching user group is less than the satisfaction index threshold.

[0074] In one specific embodiment, the positive reinforcement of the interest layer in the portrait network structure is specifically performed by extracting interests related to the interest area in the virtual space according to the interests related to each area in the virtual space, mapping the comprehensive love coefficient of the target user for the interests related to the interest area in the virtual space according to the comprehensive love coefficient of the target user for each interest, and calculating the comprehensive love coefficient improvement ratio The comprehensive love coefficient of each target user of interest related to the interest area in the virtual space is multiplied by the comprehensive love coefficient promotion ratio, so as to be positively reinforced.

[0075] In one specific embodiment, the behavior aversion layer in the portrait network structure is adjusted in detail, and the specific method is as follows: the behavior aversion layer of the target user is analyzed again, and the behavior characteristics of the users in the matching user group are added together for analysis.

[0076] The present application constructs a multi-level and three-dimensional user portrait network structure, including an interest layer, a behavior aversion layer and a behavior characteristic layer, which greatly improves the accuracy and dimension of user feature expression compared with the single label portrait in the prior art. Through multi-source fusion analysis of path data, physiological characteristic data and voice chat data, the system can comprehensively capture the explicit interest and implicit preference of the user, form a more rich and accurate user portrait, facilitate subsequent user matching, and realize a two-way matching mechanism by analyzing the interest correlation coefficient and aversion correlation coefficient between users. Compared with the traditional matching method based on common interest, the present application can consider both positive attraction factors and negative repulsion factors, greatly improving the accuracy of matching and the satisfaction of user matching experience. Meanwhile, the present application constructs a behavior characteristic layer by fine analysis of the behavior characteristics and communication methods of users in different areas, so that the system can dynamically adjust the matching and interaction strategy according to different social scenarios, and based on the adaptive optimization mechanism of user feedback, the system can automatically evaluate the satisfaction degree according to the physiological response data after social interaction, and adjust the weight of the interest layer or the fine behavior aversion layer, so as to realize continuous optimization of system performance.

[0077] The formulas in the specification are dimensionless numerical calculations, and the formulas are obtained by software simulation of a large amount of data to obtain a formula closest to the real situation. The preset parameters and threshold values in the formula are set by a person skilled in the art according to the actual situation.

[0078] The above content is only an example and description of the concept of the present application. Those skilled in the art can make various modifications or supplements to the described specific embodiments or use similar ways to replace them, as long as they do not deviate from the concept of the present application or exceed the scope defined by the present application.

Claims

1. A virtual space social interaction engine system based on AI enhancement, characterized by: include: The user portrait construction module is used to collect the target user's path data, physiological characteristics data, and voice chat data in the virtual space, and build a portrait network structure based on this data; The intelligent interest matching module is used to intelligently match different users' interests based on the target user's profile network structure to obtain the target user's matching user group; The user feedback analysis and optimization module is used to randomly match the target user with users in the corresponding matching user group and generate interest areas, and optimize the target user's portrait network structure based on the feedback data after matching.

2. The AI-enhanced virtual space social interaction engine system according to claim 1, characterized in that: The specific method of constructing the target user's portrait network structure is as follows: Build the target user's interest layer based on the target user's path data and voice chat data in the virtual space; Build a behavioral aversion layer for target users based on their physiological characteristics and voice chat data in the virtual space; Build a behavioral feature layer for the target user based on the target user's path data, physiological feature data, and voice chat data in the virtual space; The target user's interest layer, behavior aversion layer and behavior feature layer are used as the portrait network structure.

3. The AI-enhanced virtual space social interaction engine system according to claim 2, characterized in that: The specific method of constructing the interest layer of the target user is as follows: Based on the target user's path data and voice chat data in the virtual space, the TF-IDF algorithm is used to extract the target user's interest keywords and the areas where each interest keyword appears; Analyze the target user's first preference coefficient for each interest based on the target user's interest keywords; Based on the target user's path data in the virtual space, the spatial coordinates of each area of ​​the target user in the virtual space are obtained. Based on this, the target user's preference for each area is analyzed. The area where each of the target user's interest keywords appears is weighted, and the second preference coefficient of the target user for each interest is mapped. An interest layer of the target user is generated based on the first liking coefficient and the second liking coefficient of each interest of the target user.

4. The AI-enhanced virtual space social interaction engine system according to claim 2, characterized in that: The specific method of constructing the target user's behavior aversion layer is as follows: Based on the target user's physiological characteristic data in the virtual space, extract the measured values ​​of each physiological reaction data of the target user when communicating with other users in the virtual space, analyze the time points of each abnormal physiological reaction of the target user, and extract the body characteristic data and voice chat data of other users at each abnormal physiological reaction time point of the target user; Based on the body feature data and voice chat data of other users at the time points of the target user's abnormal physiological reactions, behavioral data correlation analysis is performed to obtain the target user's disgusting actions, disgusting tones and disgusting words, and generate the target user's behavioral disgust layer.

5. The AI-enhanced virtual space social interaction engine system according to claim 3, characterized in that: The specific method of constructing the target user's behavioral feature layer is as follows: Mapping the target user's path data, physiological characteristic data, and voice chat data in each area of ​​the virtual space to obtain the target user's physiological characteristic data and voice chat data; Conducting behavioral analysis on the physiological characteristic data of the target user in each area of ​​the virtual space to obtain a behavioral characteristic map of the target user in each area; Perform voice communication feature analysis on the target user's voice chat data in each area of ​​the virtual space to obtain a set of communication feature sets of the target user in each area; The target user's behavior feature maps and communication mode feature sets in each area are unified as the data of the behavior feature layer, thereby obtaining the target user's behavior feature layer.

6. The AI-enhanced virtual space social interaction engine system according to claim 2, characterized in that: The specific method of performing intelligent interest matching to obtain a matching user group for the target user is as follows: Based on the target user's profile network structure, the target user's interest layer is extracted, and the interest layers of other users are obtained, and the interest correlation coefficients of the target user and other users are analyzed; According to the target user's portrait network structure, the target user's behavior aversion layer and behavior feature layer are extracted, and the behavior aversion layer and behavior feature layer of other users are obtained to analyze the aversion correlation coefficient between the target user and other users; Evaluate the comprehensive matching coefficient between the target user and other users based on the interest correlation coefficient and aversion correlation coefficient between the target user and other users; A comprehensive matching coefficient threshold is obtained from the local database. Based on the comprehensive matching coefficient between the target user and other users, if the comprehensive matching coefficient between the target user and another user is greater than the comprehensive matching coefficient threshold, the other user is classified as a user in the matching user group of the target user, and so on, thereby obtaining the matching user group of the target user.

7. The AI-enhanced virtual space social interaction engine system according to claim 6, characterized in that: The specific analysis method for analyzing the interest correlation coefficient between the target user and other users is as follows: According to the target user's interest level, analyze the target user's preference weight for each interest; Analyze the weight of other users' preferences for each interest based on their interest levels; Based on the target user's preference weight for each interest and the preference weights of other users for each interest, the correlation index between the target user and other users for each interest is calculated; The correlation index threshold is obtained from the local database. If the correlation index between the target user and other users for a certain interest is greater than the correlation index threshold, the interest is marked as a related interest, thereby obtaining the related interests of the target user and other users, extracting the correlation index between the target user and other users for each related interest, and calculating the interest correlation coefficient of the target user and other users based on the correlation index.

8. The AI-enhanced virtual space social interaction engine system according to claim 6, characterized in that: The specific analysis method for analyzing the aversion correlation coefficient between the target user and other users is as follows: Based on the target user's behavioral aversion layer and interest layer, and combined with the behavioral feature layer and interest layer of other users, a correlation analysis of aversion actions is performed to obtain the first aversion index of the target user and other users; Based on the target user's behavioral feature layer and interest layer, and combined with other users' behavioral disgust layer and interest layer, a disgust action correlation analysis is performed to obtain the second disgust index of the target user and other users; The disgust correlation coefficient between the target user and other users is calculated based on the first disgust index and the second disgust index of the target user and other users.

9. The AI-enhanced virtual space social interaction engine system according to claim 2, characterized in that: The specific method for optimizing the target user's portrait network structure based on the matched feedback data is as follows: Collect the measured values ​​of various physiological response data during the interaction between the target user and users in the matched user group, and analyze the target user's satisfaction index with the matched user group based on the data; Based on the target user's satisfaction index with the matching user group, the interest layer and behavior aversion layer in the portrait network structure are adjusted accordingly.

10. The AI-enhanced virtual space social interaction engine system according to claim 9, characterized in that: The specific method for adjusting the interest layer and the behavior aversion layer in the portrait network structure is as follows: The satisfaction index threshold is obtained from the local database. Based on the target user's satisfaction index with the matching user group, if the target user's satisfaction index with the matching user group is greater than the satisfaction index threshold, the interest layer in the portrait network structure is positively reinforced. Otherwise, the behavioral aversion layer in the portrait network structure is fine-tuned.

Images