Interaction method and system for virtual exhibition hall

Abstract

The invention relates to the technical field of virtual exhibition hall interaction, and discloses a virtual exhibition hall interaction method and system. The method comprises the steps of firstly obtaining user interaction data collected by various terminals such as touch, voice and somatosensory terminals, extracting behavior characteristics and matching to obtain interaction pairs, and specifically, extracting characteristic descriptors, distributing time sequence labels and searching matched nodes to obtain the interaction pairs. And then constructing an interaction model by using the user interaction data and the interaction pairs, and performing scene encapsulation to generate a standard interaction packet, including obtaining the interaction model by taking the interaction pairs as anchor points, and performing layered scene encapsulation through two levels. In addition, topological data of the exhibition hall is obtained to generate an interaction path, and multiple terminals collect data in sequence and screen interaction pairs. The system comprises an interaction controller which can execute the operation. According to the invention, effective processing of multi-terminal interaction data and construction of the interaction model are realized, and the fluency and accuracy of interaction of the virtual exhibition hall are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of virtual exhibition hall interaction, and in particular to an interaction method and system for a virtual exhibition hall. Background Art

[0002] With the rapid development of information technology, virtual exhibition halls, as a new form of display, are gradually gaining widespread application in various fields. They can break the limitations of time and space and provide users with a convenient and efficient display experience. However, in the interactive process of virtual exhibition halls, many problems still need to be solved.

[0003] When it comes to processing multi-terminal interaction data, traditional virtual exhibition hall interaction methods are often only able to process interaction data from a single terminal, making it difficult to simultaneously integrate user interaction data collected from multiple terminals, such as touch interaction terminals, voice interaction terminals, and somatosensory interaction terminals. This results in an inability to fully and accurately capture user behavior characteristics, which in turn affects the effectiveness of interaction and user experience. For example, when users interact using both voice and gestures simultaneously, traditional methods are unable to effectively integrate the data from these two interaction methods, resulting in inaccurate and untimely interaction responses.

[0004] When it comes to interaction model construction and scene encapsulation, existing methods lack effective mechanisms for constructing reasonable interaction models based on user interaction data and interaction pairs. Furthermore, it's difficult to effectively encapsulate interaction models into standard interaction packages. This results in incomplete and incoherent interaction scenes in virtual exhibition halls, hindering the user's immersive interactive experience. For example, when constructing interaction models, traditional methods may fail to accurately determine the relationships between interaction pairs, leading to loopholes in the interaction model and compromising the smoothness of interaction.

[0005] When it comes to interactive path planning, traditional methods often fail to fully consider factors such as the target scenario, node status, and terminal load data when generating interactive paths based on exhibition hall topology data. This results in suboptimal interactive paths, potentially leading to path choppy paths, insufficient interactive responses, or high operational complexity. For example, in some complex exhibition hall scenarios, interactive paths generated by traditional methods may experience congestion, impacting the user's interactive experience.

[0006] Existing methods lack effective mechanisms for filtering interaction pairs, making it difficult to remove mismatched interaction pairs, which in turn affects the quality of interaction model construction. This can lead to incorrect interaction pairs in the interaction model, further impacting the accuracy and reliability of virtual exhibition hall interactions.

[0007] During the interaction process in the entire virtual exhibition hall, the coordination among various links is poor, lacking unified management and control, resulting in an inefficient and unstable interaction process. For example, there may be problems such as untimely data transmission or inconsistent processing among links such as terminal data collection, interaction model construction, scene encapsulation, and interaction path planning, affecting the overall interaction effect of the virtual exhibition hall. Summary of the Invention

[0008] The purpose of the present invention is to provide an interaction method and system for a virtual exhibition hall to solve the problems raised in the above background technology.

[0009] To achieve the above purpose, the present invention provides the following technical solution: An interaction method for a virtual exhibition hall, the method comprising: Obtain user interaction data collected by multiple terminals, extract behavior characteristics respectively according to the user interaction data, and obtain interaction pairs through matching according to the extracted behavior characteristics; wherein the multiple terminals include a touch interaction terminal, a voice interaction terminal, and a somatosensory interaction terminal; wherein obtaining interaction pairs through matching according to the extracted behavior characteristics includes: extracting a feature descriptor at the behavior characteristic, assigning a timing label to the behavior characteristic, and finding a matching node according to the feature descriptor, thereby obtaining a matching interaction pair; wherein the feature descriptor is an attribute sequence describing the behavior characteristic; Construct an interaction model according to the user interaction data and the interaction pairs, and perform scene encapsulation on the interaction model according to the user interaction data to generate a standard interaction package, including: Use the mutually matched interaction pairs as anchor points to obtain an interaction model through continuous matching, and adopt a method based on scene segmentation. Based on the sparse scenes obtained by matching the behavior characteristics among multiple terminals, obtain a complete or partial interaction path model covering the target area through extension and screening; Adopt hierarchical scene encapsulation, perform scene encapsulation through two levels. The first level encapsulates from the original interaction space to an intermediate scene that can be expressed by available protocols; the second level encapsulates the data of the intermediate scene into the target space that needs to be interacted to generate a standard interaction package.

[0010] Preferably, the method further includes the following content: Obtain the topological data of the exhibition hall that needs to be interacted, and generate an interaction path according to the topological data of the exhibition hall; multiple terminals collect user interaction data according to the interaction path; The topological data of the exhibition hall includes the target scene, node status, and terminal load data of the area that needs to be interacted. Generating an interaction path according to the topological data of the exhibition hall includes the following content: Perform boundary recognition according to the target scene to construct an interaction area, and plan an interaction path according to the interaction area; And establish an interaction benchmark according to the node status and terminal load data, and collect user interaction data according to the interaction benchmark.

[0011] Preferably, planning the interaction path according to the interaction area includes: Construct a two-dimensional exhibition hall topology map for the target scene to be interacted with, and divide the target scene to be interacted with into several grid scenes with a preset interval as the side length according to the two-dimensional exhibition hall topology map; Classify the two-dimensional grid scenes according to the target scene, node status, and terminal load data. The classification includes three types: core interaction area, ordinary scene area, and blocked interaction area, and mark the priority of the core interaction area; Randomly generate several initial interaction paths to form an initial interaction path set; Construct a fitness evaluation system, and the fitness evaluation system includes a path smoothness evaluation item, an interaction response volume evaluation item, and an operation complexity evaluation item; The path smoothness evaluation item is the accumulated value of the switching delays between scenes in the initial interaction path; The interaction response volume evaluation item is the total interaction traffic that each node can carry per unit time in the initial interaction path; The operation complexity evaluation item is the number of terminal configurations and node parameters that need to be adjusted in the initial interaction path; Use the initial interaction path set as the initial population, and perform iterative optimization according to the fitness evaluation system to obtain the optimal interaction path and complete the planning of the interaction path.

[0012] Preferably, using the initial interaction path set as the initial population and performing iterative optimization according to the fitness evaluation system includes: According to the fitness evaluation system, calculate the evaluation values of each initial interaction path, and sort the initial interaction paths from high to low according to the three evaluation items respectively to obtain three sorting sets of initial interaction paths; According to the evaluation values, randomly select several initial interaction paths from the three sorting sets respectively to form three sub-populations in different directions. The selected probability is proportional to the evaluation value; Within each sub-population, perform gene recombination, mutation adjustment, and individual learning according to the genetic optimization algorithm, and eliminate the unreachable path schemes, and iterate a preset number of times respectively to form three iterated sub-populations; Randomly select individuals from the three sub-populations according to the evaluation values to form a basic gene pool. Cross between the three sub-populations in pairs to form three groups of cross sub-populations. For each group of cross sub-populations, calculate the three evaluation values of each individual, perform multi-objective sorting on the two evaluation values involved in the cross sub-population, and perform gene recombination, mutation adjustment, and individual learning according to the sorting results, and then eliminate the unreachable path schemes; After the three groups of cross-subpopulations are iterated for a preset number of times, the three subpopulations are combined together, and the weighted sum of the three evaluation items is obtained to get the comprehensive evaluation value. According to the comprehensive evaluation value, gene recombination, mutation adjustment, and individual learning are carried out, and after iterating for the preset number of times, the individual with the largest evaluation value is obtained as the optimal interaction path.

[0013] Preferably, for the multi-objective sorting of the two evaluation values involved in the cross-subpopulation, and for gene recombination, mutation adjustment, and individual learning according to the sorting results, before the cross-subpopulation crosses, a dynamic adjustment mechanism is adopted to select the cross object, where the selection probability of the cross object is related to the position of the individual in the sorting. The later the position of the individual in the sorting, the higher the probability of being selected.

[0014] Preferably, the method further includes: assigning a selection probability according to the sorting order of the individuals to be crossed. The later the individual is sorted, the larger the selection probability value.

[0015] Preferably, the method further includes the following content: Screen the matched interaction pairs, screen out the mis-matched interaction pairs, and when constructing the interaction model, construct it according to the screened interaction pairs.

[0016] Preferably, screening the matched interaction pairs is: using two-way verification and uniqueness verification to eliminate the wrongly matched interaction pairs.

[0017] Preferably, the present invention further includes an interaction system for a virtual exhibition hall, which is applied to the above-mentioned interaction method for a virtual exhibition hall. The system includes an interaction controller: The interaction controller is used to obtain user interaction data collected by multiple terminals, extract behavior characteristics respectively according to the user interaction data, and obtain interaction pairs according to the extracted behavior characteristics; The interaction controller is also used to construct an interaction model according to the user interaction data and the interaction pairs, and perform scenario encapsulation on the interaction model according to the user interaction data to generate a standard interaction package.

[0018] Preferably, the interaction controller is also used to generate an interaction path according to the exhibition hall topology data, and multiple terminals collect user interaction data according to the interaction path.

[0019] Compared with the prior art, the beneficial effects of the present invention are: The virtual exhibition hall interaction method and system provided by the present invention can comprehensively and accurately capture the user's interaction intention by obtaining user interaction data collected by various terminals and performing behavior feature extraction and matching. The various terminals include touch, voice, and somatosensory, etc., which can obtain user behavior information from multiple dimensions, extract feature descriptors, assign temporal tags, and then find matching nodes to form interaction pairs. This enables the system to more accurately understand the user's operation behavior in the virtual exhibition hall and provide a more reliable basis for subsequent interactions.

[0020] When constructing the interaction model and scene encapsulation, the interaction pair is used as the anchor point to obtain the interaction model through continuous matching, and then the hierarchical scene encapsulation method is adopted to encapsulate from the original interaction space to the intermediate scene, and then to the target space to generate the standard interaction package. This method can construct a complete and coherent interaction model, covering the interaction paths in the target area, making the interaction scenarios in the virtual exhibition hall more rich and real, and bringing an immersive interaction experience to users. Based on the scene segmentation method, a complete or partial interaction path model is expanded and screened from the sparse scene to ensure the rationality and coverage of the interaction path.

[0021] In the process of obtaining the exhibition hall topology data to generate the interaction path, a two-dimensional exhibition hall topology map is constructed and the grid scene is divided, the types such as the core interaction area are classified and the priorities are marked, and the initial interaction path set is iteratively optimized through the fitness evaluation system. The fitness evaluation system includes evaluation items such as path smoothness, interaction response volume, and operation complexity, which can comprehensively consider various factors and finally obtain the optimal interaction path. This makes the interaction path smoother and more efficient, reduces the switching delay, increases the interaction response volume, reduces the operation complexity, and improves the user's interaction experience in the virtual exhibition hall.

[0022] The matching interaction pairs are screened by two-way verification and uniqueness verification to eliminate the wrongly matched interaction pairs and ensure the accuracy of the interaction pairs used when constructing the interaction model. This can improve the quality of the interaction model, and further make the interaction in the virtual exhibition hall more accurate and reliable, reducing the occurrence of misoperations.

[0023] The interaction controller in the interaction system can execute the above series of operations to realize functions such as the processing of multi-terminal interaction data, the construction of the interaction model, the scene encapsulation, and the generation of the interaction path, making the entire interaction process in the virtual exhibition hall more efficient, stable, and collaborative. From data collection to interaction model generation, to interaction path planning and interaction pair screening, each link cooperates closely, improving the overall performance of virtual exhibition hall interaction and user satisfaction. Brief Description of the Drawings

[0024] Figure 1 It is the working principle diagram of the interaction method of the virtual exhibition hall described in the present invention; Figure 2 It is the design diagram of the exhibition hall topology data processing; Figure 3 Design drawings for interactive path planning; Figure 4 Design drawings for interactive path optimization; Figure 5 Design drawings for dynamic adjustment mechanism. Specific implementation manners

[0025] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0026] Please refer to Figures 1-5 , an interactive method for a virtual exhibition hall involved in the present invention, and the specific implementation steps are as follows: Obtain user interaction data collected by multiple terminals, where the multiple terminals include a touch interaction terminal, a voice interaction terminal, and a somatosensory interaction terminal. Respectively extract the behavior characteristics of the obtained user interaction data, extract feature descriptors at the behavior characteristics, where the feature descriptors are attribute sequences describing the behavior characteristics, and at the same time assign time sequence tags to the behavior characteristics, and find matching nodes according to the feature descriptors, so as to obtain matching interaction pairs.

[0027] Use the obtained interaction pairs that match each other as anchor points, and obtain an interaction model through continuous matching. Adopt a method based on scene segmentation, based on the sparse scenes obtained by the behavior characteristic matching between multiple terminals, and obtain a complete or partial interactive path model covering the target area through expansion and screening.

[0028] Adopt hierarchical scene encapsulation, and perform scene encapsulation through two levels: the first level encapsulates from the original interactive space to an intermediate scene that can be expressed by available protocols; the second level encapsulates the data of the intermediate scene into the target space that needs to be interacted to generate a standard interactive package. Embodiment 1:

[0029] This embodiment describes in detail the topological data of the exhibition hall required for obtaining the required interaction. The topological data of the exhibition hall includes the target scene, node status, and terminal load data of the required interaction area. The target scene refers to the specific scene in the virtual exhibition hall that needs to be interacted, such as a certain exhibition area or exhibit display area; the node status includes the running status of each interactive node, such as whether it is available, the load situation, etc.; the terminal load data refers to the load situation of each interactive terminal, such as processing capacity, storage capacity, etc.

[0030] Build an interaction area based on boundary recognition for the target scenario. The process of boundary recognition can be achieved through image recognition technology, 3D modeling technology, etc. For example, use 3D modeling software to model the target scenario, and then identify the boundaries of the scenario through algorithms to determine the scope of the interaction area. Build a 2D exhibition hall topology map for the target scenario to be interacted with. The 2D exhibition hall topology map can be obtained by projecting the 3D scene onto a 2D plane or drawn by specialized drawing software. Divide the target scenario to be interacted with into several grid scenarios with a preset interval as the side length according to the 2D exhibition hall topology map. The preset interval can be set according to actual needs. For example, it can be set to 1 meter, 2 meters, etc. Divide the target scenario into multiple grid scenarios, and the size of each grid scenario is a square with a preset interval.

[0031] Classify the 2D grid scenarios according to the target scenario, node status, and terminal load data. The classification includes three types: core interaction area, ordinary scene area, and blocked interaction area, and label the priority of the core interaction area. The core interaction area refers to the area where user interaction is relatively frequent and important, such as the core area for exhibiting exhibits; the ordinary scene area refers to the area where user interaction is relatively less; the blocked interaction area refers to the area where interaction may be blocked due to node status or terminal load, etc. Label the priority of the core interaction area to give priority to the interaction requirements of the core interaction area in subsequent interaction path planning.

[0032] Randomly generate a number of initial interaction paths to form an initial interaction path set. The generation of the initial interaction paths can adopt a random algorithm. For example, start from the starting point of the interaction area and randomly select the next grid scenario until reaching the end point to form an initial interaction path. Repeat this process multiple times to generate multiple initial interaction paths to form an initial interaction path set.

[0033] Build a fitness evaluation system, which includes a path smoothness evaluation item, an interaction response volume evaluation item, and an operation complexity evaluation item. Among them, the path smoothness evaluation item is the cumulative value of the switching delays between scenarios in the initial interaction path. The switching delay between scenarios refers to the time required for the system to respond when the user moves from one grid scenario to another grid scenario. The smaller the cumulative value, the higher the path smoothness. The interaction response volume evaluation item is the total interaction traffic that each node can carry per unit time in the initial interaction path. The interaction traffic that each node can carry per unit time is determined according to the node status and terminal load data. The larger the total value, the higher the interaction response volume. The operation complexity evaluation item is the number of terminal configurations and node parameters that need to be adjusted in the initial interaction path. The fewer the number of terminal configurations and node parameters that need to be adjusted, the lower the operation complexity.

[0034] Using the initial set of interaction paths as the initial population, iterative optimization is performed according to the fitness evaluation system to obtain the optimal interaction path and complete the planning of the interaction path. The process of iterative optimization is as follows: According to the fitness evaluation system, calculate the evaluation values of each initial interaction path, and sort the initial interaction paths in descending order of evaluation values according to the three evaluation items to obtain three sorted sets of initial interaction paths. For example, for the path smoothness evaluation item, sort the initial interaction paths in ascending order of the cumulative switching delay value to obtain a sorted set; for the interaction response volume evaluation item, sort in descending order of the total interaction flow to obtain another sorted set; for the operation complexity evaluation item, sort in ascending order of the number of terminal configurations and node parameters that need to be adjusted to obtain a third sorted set.

[0035] According to the evaluation values, select a number of initial interaction paths from the three sorted sets through probability to form three sub-populations in different directions. The selected probability is proportional to the evaluation value. For example, in the sorted set of path smoothness, the higher the evaluation value (i.e., the smaller the cumulative switching delay value) of the initial interaction path, the greater the probability of being selected. In this way, select the initial interaction paths that perform well in different evaluation items to form three sub-populations.

[0036] Within each sub-population, gene recombination, mutation adjustment, and individual learning are performed according to the genetic optimization algorithm, and unreachable path schemes are eliminated. Iterate a preset number of times respectively to form three sub-populations after iteration. Gene recombination refers to cross-combining the individuals in the sub-population to generate new individuals; mutation adjustment refers to randomly mutating the genes of individuals to increase the diversity of the population; individual learning refers to individuals improving their own fitness by learning the characteristics of other excellent individuals. During the iteration process, it is necessary to continuously eliminate unreachable path schemes, such as schemes with unreachable grid scenarios in the path.

[0037] Randomly select individuals from the three sub-populations according to the evaluation values to form a basic gene pool, and cross between the three sub-populations in pairs to form three groups of cross-sub-populations. For each group of cross-sub-populations, calculate the three evaluation values of each individual, and perform multi-objective sorting on the two evaluation items involved in the cross-sub-population. For example, the first group of cross-sub-populations involves the path smoothness evaluation item and the interaction response volume evaluation item, and perform multi-objective sorting on these two evaluation items to determine the superiority and inferiority order of the individuals.

[0038] Before the cross-sub-populations cross, a dynamic adjustment mechanism is used to select the cross objects, assign the selection probability according to the sorting order of the individuals to be crossed. The more backward the individual's sorting, the greater the selection probability value, and the more backward the position of the individual in the sorting, the higher the probability of being selected. This dynamic adjustment mechanism can increase the diversity of the population and prevent the algorithm from falling into local optimality.

[0039] According to the sorting results, gene recombination, mutation adjustment, and individual learning are carried out, and then the unreachable path solutions are eliminated. After the three cross-subpopulations iterate the preset number of times, the three subpopulations are combined together, and the weighted sum of the three evaluation items is obtained to get the comprehensive evaluation value. According to the comprehensive evaluation value, gene recombination, mutation adjustment, and individual learning are carried out, and after iterating the preset number of times, the individual with the largest evaluation value is obtained as the optimal interaction path. The weights of the weighted sum can be set according to actual needs. For example, the weight of the path fluency evaluation item can be set to 0.4, the weight of the interaction response volume evaluation item can be set to 0.3, and the weight of the operation complexity evaluation item can be set to 0.3, etc.

[0040] Multiple terminals collect according to this interaction path to obtain user interaction data. At the same time, an interaction benchmark is established based on the node status and terminal load data, and user interaction data is collected according to the interaction benchmark. The interaction benchmark can include the frequency of interaction, the data transmission rate, etc., to ensure that the collected user interaction data meets the requirements of the system. Embodiment 2:

[0041] This embodiment further defines the iterative optimization process on the basis of Embodiment 1.

[0042] According to the fitness evaluation system, the evaluation values of each initial interaction path are calculated. The fitness evaluation system includes a path fluency evaluation item, an interaction response volume evaluation item, and an operation complexity evaluation item. Among them, the path fluency evaluation item corresponds to the cumulative value of the switching delays between each scenario in the initial interaction path. The smaller this value is, the smoother the path switching is; the interaction response volume evaluation item is the total interaction traffic that each node can carry per unit time in the initial interaction path. The larger the value, the stronger the overall carrying capacity of the node; the operation complexity evaluation item is the number of terminal configurations and node parameters that need to be adjusted in the initial interaction path. The fewer the number, the lower the operation difficulty.

[0043] Based on these three evaluation items, the initial interaction paths are respectively sorted from high to low in terms of evaluation values to obtain three sorting sets of the initial interaction paths. For example, for the path fluency evaluation item, all initial interaction paths are sorted from small to large according to the cumulative value of the switching delays. The path with a smaller cumulative value of the switching delays is ranked higher in the sorting set; for the interaction response volume evaluation item, it is sorted from large to small according to the total interaction traffic that each node can carry per unit time. The path with a larger total traffic is ranked higher; the operation complexity evaluation item is sorted from small to large according to the number of terminal configurations and node parameters that need to be adjusted. The fewer the number, the higher the ranking.

[0044] According to the evaluation values, a number of initial interaction paths are probabilistically selected from the three sorted sets respectively to form three sub-populations in different directions. The probability of being selected here is proportional to the evaluation value, that is, the higher the evaluation value of a path, the greater the probability of being selected in the corresponding sorted set. For example, in the sorted set of path fluency, the paths ranked higher have a smaller cumulative switching delay value and a higher evaluation value, so the probability of being selected is higher than that of the paths ranked lower. Through this probability selection mechanism, the paths that perform excellently in different evaluation dimensions can be separately selected into the corresponding sub-populations, thus forming three sub-populations with different optimization directions.

[0045] Within each sub-population, gene recombination, mutation adjustment, and individual learning are carried out according to the genetic optimization algorithm, and the unreachable path schemes are eliminated. During the gene recombination process, the individual paths in the sub-population are cross-combined to exchange their gene segments (i.e., the sequence of scene nodes in the path), thus generating new path individuals; mutation adjustment is to randomly replace or adjust some scene nodes in the individual path to increase the diversity of the population and avoid the algorithm falling into local optimality; individual learning is to let each path individual refer to the path structure of the excellent individuals in the sub-population to optimize its own node sequence. During this process, the reachability of the path needs to be continuously checked. If there are scene nodes in the path that cannot be reached from the starting point, or the end point of the path cannot reach the target position, then this path scheme is eliminated. Each sub-population is iterated a preset number of times (such as 100 times or 200 times, and the preset number can be set according to actual needs) to form three iterated sub-populations.

[0046] Individuals are randomly selected from the three sub-populations according to the evaluation values to form a basic gene pool. This random selection is not completely equiprobable, but is a probabilistic selection based on the size of the evaluation value. The individuals with higher evaluation values have a relatively greater probability of being selected. After forming the basic gene pool, pairwise crossover is performed between the three sub-populations to form three groups of crossover sub-populations. For example, sub-population A and sub-population B are crossed to form the first group of crossover sub-populations, sub-population B and sub-population C are crossed to form the second group, and sub-population C and sub-population A are crossed to form the third group.

[0047] For each group of crossover sub-populations, calculate the three evaluation values of each individual, and then perform multi-objective sorting on the two evaluation items involved in the crossover sub-population. For example, the first group of crossover sub-populations is formed by crossing sub-population A and sub-population B. Sub-population A mainly optimizes path fluency, and sub-population B mainly optimizes the amount of interaction response. Therefore, this group of crossover sub-populations involves the path fluency evaluation item and the interaction response amount evaluation item, and multi-objective sorting needs to be performed on these two evaluation items. When performing multi-objective sorting, it is necessary to comprehensively consider the values of the two evaluation items to determine the relative superiority and inferiority order of each individual, rather than sorting only based on a single evaluation item.

[0048] Before the crossover of the cross-subpopulations, a dynamic adjustment mechanism is adopted to select the crossover objects. Specifically, the selection probability is assigned according to the order of the individuals to be crossed in the ranking. The later the individual ranks, the larger the selection probability value, that is, the later the position of the individual in the ranking, the higher the probability of being selected as a crossover object. For example, in the multi-objective ranking of a certain group of cross-subpopulations, the probability of the individual ranked 10th being selected is higher than that of the individual ranked 1st. This mechanism can prevent the algorithm from converging prematurely to the local optimal solution. By increasing the crossover probability of the individuals with lower ranks, more diverse gene combinations are introduced, thereby enhancing the global search ability of the algorithm.

[0049] According to the ranking results, genetic recombination, mutation adjustment, and individual learning are performed on the cross-subpopulations. It is also necessary to eliminate the unreachable path schemes. The methods of genetic recombination and mutation adjustment are similar to the operations within the subpopulations. However, since it involves the gene combinations of two different subpopulations, new path structures may be generated in the path individuals after crossover. Individual learning is to optimize its own structure by referring to the path characteristics of the excellent individuals in the cross-subpopulations. After the three groups of cross-subpopulations are iterated for the preset number of times, the three subpopulations are combined together.

[0050] After combination, the weighted sum of the three evaluation items is obtained to get the comprehensive evaluation value. When performing the weighted sum, the weight of each evaluation item can be set according to the requirements of the actual application scenario. For example, the weight of the path smoothness evaluation item is 0.4, the weight of the interaction response quantity evaluation item is 0.3, and the weight of the operation complexity evaluation item is 0.3. Through the weighted sum, the evaluation values in three different dimensions are converted into a comprehensive evaluation value for unified comparison of all individuals.

[0051] According to the comprehensive evaluation value, genetic recombination, mutation adjustment, and individual learning are performed on the combined population and iterated for the preset number of times. During the iteration process, the path structure of the individuals is continuously optimized. Finally, the individual with the largest evaluation value is obtained, and this individual is the optimal interaction path. The entire iterative optimization process gradually improves the comprehensive performance of the interaction path through steps such as multi-objective ranking, dynamic crossover selection, and genetic operations, ensuring that the planned path achieves the optimal balance in terms of path smoothness, interaction response quantity, and operation complexity. Example 3:

[0052] This example elaborates that when constructing an interaction model, it is necessary to screen the matching interaction pairs to eliminate the mis-matched interaction pairs to ensure the accuracy of the subsequent construction of the interaction model.

[0053] The matching interaction pairs are obtained through steps such as acquiring user interaction data collected by various terminals, extracting behavioral features, extracting feature descriptors, assigning temporal tags, and finding matching nodes. However, in this process, due to factors such as terminal data collection errors, behavioral feature extraction biases, or limitations of matching algorithms, some interaction pairs may be mis-matched. Therefore, a screening mechanism is needed to verify and validate them.

[0054] The screening operation adopts the method of two-way verification and uniqueness verification. Two-way verification is to verify information from two directions of the interaction pair to ensure the consistency and accuracy of the interaction relationship. For example, for a group of interaction pairs obtained by matching the behavioral features of a touch terminal and a voice terminal, it is necessary to verify whether the feature descriptors corresponding to the touch behavior and the feature descriptors corresponding to the voice behavior match each other in terms of temporal tags and attribute sequences. Specifically, starting from the behavioral features of the touch terminal, according to its feature descriptor and temporal tag, check whether there is a corresponding matching node in the voice terminal to form a forward verification; then starting from the behavioral features of the voice terminal, also according to its feature descriptor and temporal tag, check whether there is a corresponding matching node in the touch terminal to form a reverse verification. Only when both the forward verification and the reverse verification pass, the interaction pair is considered to be consistent in two-way information, otherwise it is determined as a mis-matched interaction pair and excluded.

[0055] Uniqueness verification is used to exclude duplicate or conflicting interaction pairs to ensure the uniqueness and validity of the interaction pairs. In a multi-terminal interaction scenario, there may be a situation where the same behavioral feature is repeatedly matched by multiple terminals, or different terminal behavioral features are mis-matched to form duplicate interaction pairs. For example, a somatosensory interaction terminal and a touch interaction terminal may simultaneously respond to a certain action of the user. If the matching algorithm is not precise enough, two groups of interaction pairs with similar content but slightly different temporal tags may be generated. Such interaction pairs belong to duplicate interaction pairs. Uniqueness verification needs to traverse and check all the obtained interaction pairs, and judge whether there are duplicates or conflicts according to the feature descriptor and temporal tag. Specifically, for each interaction pair, compare its feature descriptor (i.e., the attribute sequence describing the behavioral feature) and temporal tag with all other interaction pairs. If there is an interaction pair with highly similar feature descriptors and temporal tags within the allowable error range, it is determined as a duplicate interaction pair, and one group is retained (such as the group with more accurate temporal tags or more complete feature descriptors), and other duplicate groups are excluded. For conflicting interaction pairs, that is, interaction pairs with mutually contradictory feature descriptors or obvious logical errors in temporal tags, they are directly excluded.

[0056] During the two-way verification process, detailed verification rules need to be established. Taking the interaction pair between a touch terminal and a voice terminal as an example, the feature descriptors of touch behaviors may include attribute sequences such as touch position, touch force, and touch duration, and the feature descriptors of voice behaviors may include attribute sequences such as voice keywords, volume, intonation, and voice duration. During two-way verification, it is necessary to not only check whether the attribute sequences of touch behaviors and voice behaviors are semantically and logically related. For example, whether the behavior of touching a certain exhibit model corresponds to the voice behavior of saying the name of the exhibit, but also ensure that the timing tags of the two are reasonable on the time axis, that is, the time interval between the occurrence of touch behavior and voice behavior is within a preset reasonable range (such as within 0.5 seconds to 2 seconds). If the time interval is too long or too short and exceeds the reasonable range, it is considered that the two-way verification fails.

[0057] Uniqueness verification also needs to consider the differences and matching tolerances of the behavior characteristics of different terminals. Since the data collection methods and precisions of different terminals are different, for example, there are differences in the precision of the action data collected by the somatosensory interaction terminal and the touch data of the touch interaction terminal. When comparing feature descriptors, it is not required to be exactly the same, but a reasonable matching tolerance should be set. For example, for the position coordinates of somatosensory actions and touch position coordinates, a certain coordinate range (such as allowing a 5% error in the X, Y, and Z axis directions) is allowed to be regarded as similar features; for the recognition of voice keywords, a certain voice recognition error range (such as the phoneme matching degree of keywords reaching more than 80%) is allowed to be regarded as the same feature. By setting the matching tolerance, it can not only effectively exclude completely duplicate or conflicting interaction pairs, but also avoid the mis-elimination of reasonable interaction pairs caused by terminal data differences.

[0058] After completing the two-way verification and uniqueness verification, all the interaction pairs that pass the screening will be used to construct an interaction model. Using the screened interaction pairs as anchor points, an interaction model is obtained through continuous matching. Specifically, starting from a certain interaction pair, according to its feature descriptor and timing tag, other adjacent or related interaction pairs are searched for, and the interaction network is gradually expanded to form an interaction model covering a certain interaction scenario. During the expansion process, the feature information of the screened interaction pairs also needs to be referred to to ensure that the newly included interaction pairs are consistent with the existing interaction pairs in logic and timing.

[0059] When constructing an interaction model using the method based on scenario segmentation, based on the sparse scenarios obtained from the behavior feature matching between multiple terminals (that is, the scenarios covered by the interaction pairs obtained after the initial screening may be relatively scattered), a complete or partial interaction path model covering the target area is obtained through expansion and screening. During the expansion process, based on the screened interaction pairs, according to the feature descriptors of the interaction pairs and the scenario information of the target area, possible interaction extension points are searched for and new interaction pairs are added; the screening process is to simply verify the expanded interaction pairs again to ensure the accuracy of the newly added interaction pairs and avoid introducing new mis-matched pairs.

[0060] Through the screening operation of bidirectional verification and uniqueness verification of the matched interaction pairs, the accuracy of the interaction pairs can be effectively improved, thereby ensuring the reliability and effectiveness of the interaction model. The accurate interaction model provides a solid foundation for subsequent scenario encapsulation and generation of standard interaction packages, enabling the interaction system of the virtual exhibition hall to more accurately respond to the user's interaction behavior and improve the user experience. The entire screening process runs through the early stage of the construction of the interaction model and is one of the key links to ensure the normal operation of the interaction system. Through strict verification and validation mechanisms, the impact of mis-matched interaction pairs is minimized, laying a foundation for the smooth progress of the subsequent interaction process. Example 4:

[0061] This example relates to an interaction system for a virtual exhibition hall. This system applies the above interaction method, and its core component is an interaction controller. The interaction controller undertakes key tasks such as data acquisition, feature extraction, interaction pair matching, interaction model construction, and scenario encapsulation in the system. The following elaborates on its implementation method in detail with specific examples.

[0062] Taking an art virtual exhibition hall as an example, users interact with the virtual exhibition hall through a touch interaction terminal (such as a tablet computer), a voice interaction terminal (such as a smart voice assistant), and a somatosensory interaction terminal (such as a Kinect device). The interaction controller first needs to acquire the user interaction data collected by various terminals. When a user browses a painting in the virtual exhibition hall, the touch interaction terminal collects touch data of the user clicking on the painting to zoom in and view, including the touch position, touch duration, etc.; the voice interaction terminal collects voice data of the user saying "introduce this painting", including the voice waveform, keywords, etc.; the somatosensory interaction terminal collects motion data of the user approaching the painting, including the movement trajectory, body movements, etc. The interaction controller receives these interaction data from different terminals in real time, providing a basis for subsequent processing.

[0063] Next, the interaction controller extracts behavioral features from the acquired user interaction data respectively. For touch data, behavioral features such as the virtual exhibition hall coordinates corresponding to the touch position and the touch operation type (such as click, drag) are extracted; for voice data, keywords (such as "introduction", "painting") and voice intonation are extracted through speech recognition technology as behavioral features; for somatosensory data, behavioral features such as the speed of the user's movement, direction, and amplitude of limb movements are extracted. For example, in the scenario where the user clicks on a painting and says "introduction", the behavioral feature of the touch terminal is a click operation at the virtual coordinates (100, 200), the behavioral feature of the voice terminal is a voice command containing the keyword "introduction" with an upward intonation, and the behavioral feature of the somatosensory terminal is the action of the user moving 0.5 meters in the direction of the painting.

[0064] After the behavioral feature extraction, the interaction controller extracts feature descriptors at the behavioral features, and the feature descriptor is an attribute sequence describing the behavioral features. For touch behavioral features, its feature descriptor may be "{coordinates: 100, 200; operation type: click; timestamp: 2025-06-20 10:00:01}"; the feature descriptor of voice behavioral features may be "{keywords: introduction, painting; intonation: upward; timestamp: 2025-06-20 10:00:02}"; the feature descriptor of somatosensory behavioral features may be "{movement direction: forward; movement distance: 0.5 meters; timestamp: 2025-06-20 10:00:00.5}". At the same time, a timing label is assigned to the behavioral features. The timing label is determined based on the time order of data acquisition and is used to identify the temporal sequence relationship of the behavioral features. For example, the timestamp of the above somatosensory data is 10:00:00.5, the touch data is 10:00:01, and the voice data is 10:00:02, forming a chronological order.

[0065] The interaction controller searches for matching nodes according to the feature descriptors to obtain matching interaction pairs. For example, in the feature descriptor of the touch behavior, "coordinates 100, 200" corresponds to the position of a certain painting in the virtual exhibition hall, and in the feature descriptor of the voice behavior, "introduce painting" corresponds to an introduction instruction for the painting. Their timestamps are within a reasonable time range (such as within 1 second), and they are semantically related. The interaction controller uses a preset matching algorithm to search for nodes in the interaction node library of the virtual exhibition hall that match these feature descriptors and finds that the interaction node of the painting contains the response logic for both touch click and voice introduction. Therefore, the touch behavior and the voice behavior are matched as a group of interaction pairs. Similarly, the action of the user moving towards the painting in the somatosensory behavior is continuous in time with the touch click action, and the moving direction points to the position of the painting. The interaction controller can match the somatosensory behavior with the touch behavior as another group of interaction pairs, forming a combination of multi-terminal interaction pairs.

[0066] After the interaction controller obtains the interaction pairs, it uses the obtained interaction pairs that match each other as anchor points to obtain the interaction model through continuous matching. Still taking the scenario of a user browsing a painting as an example, the initially matched interaction pairs are (touch click, voice introduction) and (gesture movement, touch click). The interaction controller takes these interaction pairs as starting points to find other interaction behaviors related to them. For example, after hearing the introduction of the painting, the user may issue an instruction of "zoom in on the details" through the voice terminal, and the touch terminal performs a drag operation at the same time. The interaction controller continuously matches this new set of voice and touch behaviors with the previous interaction pairs, confirms that it belongs to the extended behavior under the same interaction scenario, and thus incorporates the new interaction pairs into the interaction model, gradually expanding the interaction network.

[0067] When constructing the interaction model, a method based on scenario segmentation is adopted. Based on the sparse scenarios obtained by matching the behavior characteristics between multiple terminals, a complete or partial interaction path model covering the target area is obtained through expansion and screening. For example, the initially matched interaction pairs only cover the behaviors of the user approaching the painting, clicking on the painting, and requesting an introduction, forming sparse scenario segments. The interaction controller expands possible interaction behaviors, such as the user zooming in on the details, rotating the painting, etc., according to the topological structure of the virtual exhibition hall and the target area (i.e., the display area of this painting), and filters out reasonable interaction paths to form a complete interaction path model from the user entering this area to completing the interaction.

[0068] The interaction controller adopts hierarchical scenario encapsulation and generates standard interaction packages through two levels of scenario encapsulation. The first level encapsulates from the original interaction space to the intermediate scenario expressed in an available protocol. The original interaction space is the various interaction behavior data generated by the user through multiple terminals. The interaction controller converts this data into intermediate scenario data that conforms to a specific protocol (such as JSON format). For example, behaviors such as touch clicks and voice instructions are converted into the format specified by the protocol, including fields such as action type, parameters, and timestamp. The second level encapsulates the data of the intermediate scenario into the target space that needs to be interacted with. The target space is the specific interaction environment of the virtual exhibition hall. The interaction controller maps the intermediate scenario data to the models in the target space (such as the 3D model of the painting, display racks, etc.) to generate standard interaction packages. The standard interaction packages contain the information of all interaction pairs, the interaction path model, and the mapping relationship with the target space, which can be called by the rendering engine and interaction logic module of the virtual exhibition hall to realize the response to the user's interaction behaviors, such as displaying the painting introduction text, zooming in on the painting details, etc.

[0069] Throughout the process, the interaction controller coordinates the processing of interaction data across multiple terminals, ensuring that the behavioral characteristics of different terminals are accurately extracted and matched. The constructed interaction model can truly reflect the user's interaction intent. The standardized interaction packages generated through hierarchical encapsulation provide a unified interface and data format for the interaction in the virtual exhibition hall, enabling the system to respond to user operations efficiently and accurately, and enhancing the interaction experience in the virtual exhibition hall. For example, when a user approaches a painting by somatic movement, clicks on the painting for selection via touch, and requests an introduction by voice simultaneously, the interaction controller can quickly process these three types of data, match the interaction pairs, construct the interaction model, and generate a standardized interaction package to drive the virtual exhibition hall to perform operations such as moving the perspective, highlighting the painting, and playing the introduction voice simultaneously, achieving seamless connection of multimodal interaction. Embodiment 5:

[0070] In this embodiment, in addition to the functions described in Embodiment 4, the interaction controller in the virtual exhibition hall interaction system is further used to obtain the exhibition hall topology data and generate an interaction path, and multiple terminals collect user interaction data according to this path. The following will be described in detail in combination with a specific application scenario of a virtual exhibition hall for scientific and technological products.

[0071] Suppose that after a user enters the virtual exhibition hall, they hope to browse the exhibition area of the latest released intelligent devices of a certain brand. The interaction controller obtains the exhibition hall topology data required for interaction, which includes the target scene, node status, and terminal load data of the area to be interacted with. Among them, the target scene is the 3D model and layout of the intelligent device exhibition area, including the display positions of each exhibit and interaction nodes (such as click to view details, drag and rotate the model, etc.); the node status indicates the operating conditions of each interaction node. For example, whether the 3D model loading node of Exhibit A is currently available, and whether there is a situation of excessive load on the video playback node of Exhibit B; the terminal load data records the real-time loads of touch interaction terminals, voice interaction terminals, and somatic interaction terminals, such as the CPU occupancy rate and memory usage of a certain touch terminal.

[0072] The interaction controller constructs an interaction area through boundary recognition based on the target scene. The target scene of the intelligent device exhibition area is processed by 3D modeling software, and the boundary recognition algorithm (such as a geometric shape-based recognition method) is used to determine the range of the interaction area. For example, the surrounding walls and booth edges of the exhibition area are used as boundaries to form a closed interaction area. Then, a two-dimensional exhibition hall topology map is constructed for the target scene to be interacted with, and the three-dimensional exhibition area model is projected onto a two-dimensional plane, marking the positions of each exhibit, the coordinates of interaction nodes, and information such as the channel directions. According to the two-dimensional exhibition hall topology map, the target scene to be interacted with is divided into several grid scenes with a preset interval as the side length. The preset interval is set to 1 meter, so that the entire exhibition area is divided into multiple 1-meter × 1-meter grids, and each grid corresponds to a specific area in the virtual exhibition hall.

[0073] The interaction controller classifies the two-dimensional grid scenario according to the target scenario, node status, and terminal load data. The classification includes three types: the core interaction area, the ordinary scene area, and the blocked interaction area, and priority labels are assigned to the core interaction area. In the intelligent device exhibition area, the area where the latest smartwatch is displayed is classified as the core interaction area because of the high user interaction demand, good interaction node status of the exhibit, and low terminal load; the area where accessories are displayed has less interaction demand and is classified as the ordinary scene area; and a certain exhibit near the entrance is classified as the blocked interaction area because of the malfunction of the interaction node and the excessive corresponding terminal load. The core interaction area is labeled with the highest priority to be considered first in the interaction path planning.

[0074] After the classification is completed, the interaction controller randomly generates a number of initial interaction paths to form an initial interaction path set. For example, the initial path 1 is: Entrance → Core Interaction Area of Smartwatch → Ordinary Scene Area of Headphone → Exit; the initial path 2 is: Entrance → Ordinary Scene Area of Smart Speaker → Core Interaction Area of Smartwatch → Exit; the initial path 3 is: Entrance → Blocked Interaction Area → Core Interaction Area of Smartwatch → Exit (this path will be optimized in the iteration because it includes the blocked interaction area). Through the random generation method, multiple different initial paths are obtained, providing a basis for subsequent optimization.

[0075] Next, the interaction controller constructs a fitness evaluation system, which includes a path smoothness evaluation item, an interaction response volume evaluation item, and an operation complexity evaluation item. Taking the initial path 1 as an example, the path smoothness evaluation item is the cumulative value of the switching delays between scenarios. Suppose the switching delay from the entrance to the core interaction area of the smartwatch is 0.3 seconds, the switching delay from the core interaction area to the ordinary scene area of the headphone is 0.2 seconds, and the switching delay from the ordinary scene area to the exit is 0.1 seconds, and the cumulative value is 0.6 seconds; the interaction response volume evaluation item is the total interaction traffic that can be carried by each node per unit time. The nodes in the core interaction area of the smartwatch can carry a traffic of 50 times per second, and the nodes in the ordinary scene area of the headphone can carry a traffic of 20 times per second, and the total is 70 times per second; the operation complexity evaluation item is the number of terminal configurations and node parameters that need to be adjusted. In this path, there is no need to adjust the terminal configuration or node parameters, and the number is 0.

[0076] The interaction controller uses the initial set of interaction paths as the initial population and performs iterative optimization according to the fitness evaluation system. First, calculate the evaluation values of each initial interaction path and sort them separately according to three evaluation items. For example, in the path fluency evaluation item, the cumulative switching delay value of the initial path 2 is 0.5 seconds, which is better than 0.6 seconds of path 1, and the cumulative switching delay value of path 3 is 1.2 seconds because it contains a blocked interaction area, and the sorting is path 2 > path 1 > path 3; in the interaction response volume evaluation item, the sum of path 1 is 70 times / second, and the sum of the node traffic that can be carried in the ordinary scene area and the core interaction area of the smart speaker in path 2 is 60 times / second, and the sorting is path 1 > path 2 > path 3; in the operation complexity evaluation item, the numbers of both path 1 and path 2 are 0, and the number of path 3 is 1 because it may need to adjust node parameters to bypass the blocked interaction area, and the sorting is path 1 = path 2 > path 3.

[0077] Select a number of initial interaction paths from the three sorted sets according to probability to form three sub-populations. Suppose in the path fluency sorted set, path 2 has the highest evaluation value and the highest probability of being selected, followed by path 1, and path 3 has the lowest probability; in the interaction response volume sorted set, path 1 has the highest probability of being selected. Within each sub-population, gene recombination, mutation adjustment, and individual learning are performed according to the genetic optimization algorithm, and infeasible path solutions are eliminated. For example, path 3 is eliminated because it contains a blocked interaction area, resulting in some nodes being unreachable. After iterating a preset number of times, three iterated sub-populations are formed.

[0078] Randomly select individuals from the three sub-populations to form a basic gene pool, and cross between the sub-populations in pairs to form three groups of cross sub-populations. For each group of cross sub-populations, calculate the three evaluation values of the individuals, perform multi-objective sorting, and use a dynamic adjustment mechanism to select cross objects. The lower the individual ranking, the higher the probability of being selected. For example, in a certain cross sub-population, the individuals with a lower ranking are more likely to be selected during crossing to increase the population diversity. After crossing, gene recombination, mutation adjustment, and individual learning are performed, and infeasible solutions are eliminated, and the iteration is performed for a preset number of times.

[0079] Combine the three sub-populations, perform weighted summation on the three evaluation items to obtain a comprehensive evaluation value. Suppose the weights are 0.4 for path fluency, 0.4 for interaction response volume, and 0.2 for operation complexity. Calculate the individual with the largest evaluation value as the optimal interaction path. For example, the optimized path is: entrance → core interaction area of smart watch → ordinary scene area of smart earphone → ordinary scene area of smart speaker → exit, and this path reaches a balance in terms of fluency, response volume, and operation complexity.

[0080] Multiple terminals collect user interaction data according to this interaction path. When the user browses along the optimal path, the touch terminal collects the touch data of the user clicking on the smart watch model to view details, the voice terminal collects the voice data of the user saying "Play the headphone function introduction", and the somatosensory terminal collects the trajectory data of the user moving within the exhibition area. At the same time, the interaction controller establishes an interaction benchmark based on the node status and terminal load data. For example, it sets the upper limit of the interaction frequency in the core interaction area to 30 times per minute and the terminal data transmission rate to not less than 10 Mbps, ensuring that the collected user interaction data meets the benchmark requirements and providing accurate data for subsequent behavior feature extraction and interaction pair matching.

[0081] Through the processing of the exhibition hall topology data and the generation of the interaction path by the interaction controller, combined with the data collection by multiple terminals according to the path, the virtual exhibition hall can guide users to efficiently browse the core interaction area, reasonably allocate terminal resources, and improve the overall interaction experience. For example, in the above-mentioned smart device exhibition area, the optimal interaction path avoids blocked areas and preferentially guides users to the core exhibits. The terminals collect data according to the path and the benchmark, enabling the system to accurately respond to user operations and achieve smooth multi-modal interaction.

[0082] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device.

[0083] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An interaction method for a virtual exhibition hall, characterized in that, It includes the following: Obtain user interaction data collected by multiple terminals, extract behavior features respectively according to the user interaction data, and obtain interaction pairs through matching based on the extracted behavior features; Among them, the multiple terminals include a touch interaction terminal, a voice interaction terminal, and a somatosensory interaction terminal; Among them, obtaining interaction pairs through matching based on the extracted behavior features includes: extracting a feature descriptor at the behavior feature, assigning a time sequence label to the behavior feature, and finding a matching node according to the feature descriptor, so as to obtain a matching interaction pair; where the feature descriptor is an attribute sequence describing the behavior feature; Construct an interaction model according to the user interaction data and the interaction pairs, and perform scenario encapsulation on the interaction model according to the user interaction data to generate a standard interaction package, including: Use the interaction pairs obtained by mutual matching as anchor points to obtain an interaction model through continuous matching. Adopt a method based on scenario segmentation. Based on the sparse scenarios obtained by behavior feature matching between multiple terminals, obtain a complete or partial interaction path model covering the target area through extension and screening; Adopt hierarchical scenario encapsulation, and perform scenario encapsulation through two levels. The first level encapsulates from the original interaction space to an intermediate scenario that can be expressed by available protocols; the second level encapsulates the data of the intermediate scenario into the target space that needs to be interacted to generate a standard interaction package.

2. The interactive method of a virtual exhibition hall according to claim 1, characterized in that: It also includes the following: Obtain the exhibition hall topology data required for interaction, and generate an interaction path according to the exhibition hall topology data; multiple terminals collect user interaction data according to the interaction path; The exhibition hall topology data includes the target scenario, node status, and terminal load data of the area required for interaction. Generating an interaction path according to the exhibition hall topology data includes the following: Perform boundary recognition according to the target scenario to construct an interaction area, and plan an interaction path according to the interaction area; And establish an interaction benchmark according to the node status and terminal load data, and collect user interaction data according to the interaction benchmark.

3. The interactive method of a virtual exhibition hall according to claim 2, characterized in that: Planning an interaction path according to the interaction area includes: Construct a two-dimensional exhibition hall topology map in the target scenario to be interacted, and divide the target scenario to be interacted into several grid scenarios with a preset interval as the side length according to the two-dimensional exhibition hall topology map; Classify the two-dimensional grid scenarios according to the target scenario, node status, and terminal load data. The classification includes three types: a core interaction area, a general scenario area, and a blocked interaction area, and mark the priority of the core interaction area; Randomly generate a number of initial interaction paths to form an initial interaction path set; Construct a fitness evaluation system, and the fitness evaluation system includes a path fluency evaluation item, an interaction response volume evaluation item, and an operation complexity evaluation item; The path fluency evaluation item is the accumulated value of the switching delay between each scenario in the initial interaction path; The interaction response volume evaluation item is the total interaction traffic that can be carried by each node per unit time in the initial interaction path; The operation complexity evaluation item is the number of terminal configurations and node parameters that need to be adjusted in the initial interaction path; Use the initial interaction path set as the initial population, and perform iterative optimization according to the fitness evaluation system to obtain the optimal interaction path and complete the planning of the interaction path.

4. The interactive method of a virtual exhibition hall according to claim 3, wherein: Using the initial interaction path set as the initial population and performing iterative optimization according to the fitness evaluation system includes: According to the fitness evaluation system, calculate the evaluation values of each initial interaction path, and sort the initial interaction paths in descending order of evaluation values according to the three evaluation items respectively to obtain three sorted sets of initial interaction paths; According to the evaluation values, select a number of initial interaction paths from the three sorted sets by probability respectively to form three sub-populations in different directions, and the selected probability is proportional to the evaluation value; Within each sub-population, perform gene recombination, mutation adjustment and individual learning according to the genetic optimization algorithm, and eliminate the unreachable path solutions, and iterate a preset number of times respectively to form three iterated sub-populations; Randomly select individuals from the three sub-populations according to the evaluation values to form a basic gene pool, and cross between the three sub-populations in pairs to form three groups of cross sub-populations. For each group of cross sub-populations, calculate the three evaluation values of each individual, perform multi-objective sorting on the two evaluation values involved in the cross sub-population, and perform gene recombination, mutation adjustment and individual learning according to the sorting results, and then eliminate the unreachable path solutions; After the three groups of cross sub-populations are iterated a preset number of times, combine the three sub-populations together, perform weighted summation on the three evaluation items to obtain a comprehensive evaluation value, perform gene recombination, mutation adjustment and individual learning according to the comprehensive evaluation value, and after iterating a preset number of times, obtain the individual with the largest evaluation value as the optimal interaction path.

5. The interactive method of a virtual exhibition hall according to claim 4, characterized in that: In the multi-objective sorting of the two evaluation values involved in the cross sub-population and the gene recombination, mutation adjustment and individual learning according to the sorting results, before the cross sub-population crosses, a dynamic adjustment mechanism is used to select the cross object, and the selection probability of the cross object is related to the position of the individual in the sorting. The more backward the position of the individual in the sorting, the higher the probability of being selected.

6. The interactive method of a virtual exhibition hall according to claim 5, wherein: It further includes: Assign the selection probability according to the sorting order of the individuals to be crossed. The more backward the individual sorting, the larger the selection probability value.

7. The interactive method of a virtual exhibition hall according to claim 1, characterized in that: It further includes the following content: Screen the matching interaction pairs, screen out the mis-matched interaction pairs, and construct the interaction model according to the screened interaction pairs when constructing the interaction model.

8. The interactive method of a virtual exhibition hall according to claim 7, characterized in that: The screening of the matching interaction pairs is: using two-way verification and uniqueness verification to eliminate the wrongly matched interaction pairs.

9. An interaction system for a virtual exhibition hall, which is applied to an interaction method for a virtual exhibition hall according to any one of the above claims 1-8, including an interaction controller, characterized in that: The interaction controller is used to obtain user interaction data collected by a variety of terminals, extract behavior features respectively according to the user interaction data, and obtain interaction pairs according to the extracted behavior features; The interaction controller is further used to construct an interaction model according to the user interaction data and the interaction pairs, and perform scene encapsulation on the interaction model according to the user interaction data to generate a standard interaction package.

10. An interactive system for a virtual exhibition hall according to claim 9, characterized in that: The interaction controller is further used to generate an interaction path according to the exhibition hall topology data, and a variety of terminals collect user interaction data according to the interaction path.

Images