Immersive script experience data processing system based on character selection

Abstract

The application provides an immersive script experience data processing system based on role selection, and relates to the technical field of data processing. The system is used for: constructing a behavior track under a role perspective, performing paragraph division and cluster reorganization on the behavior track, obtaining role paragraph data, and according to the node number span between the role paragraph data and a script preset node path, counting the number of jumps and the span difference, identifying the offset characteristics of the role behavior track, calculating a behavior jump factor, calculating the time interval sequence between consecutive nodes, performing timing stability analysis on the time interval sequence, calculating the synchronization deviation degree of the behavior timing, obtaining a timing offset factor, and performing deviation detection on each role number according to the behavior jump factor and the timing offset factor to obtain a behavior analysis result. The application can perform abnormal analysis on data generated by different role selections.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to an immersive script experience data processing system based on role selection. BACKGROUND

[0002] In the existing immersive teaching system of college classrooms, the script experience function allows students to log in to the platform by inputting the classroom verification code, and after selecting the group and specific role in the activity homepage, the students enter the script performance process. The system usually drives the performance process with a multi-node script structure, including plot fragments, question and answer interactions, clue distribution, plot branches, etc. The performance behavior of students is uploaded through interface clicks, option selections, etc. and forms process data. The system synchronously advances the script process based on an event trigger model. Although the system has a role selection function, at the data processing level, the role selection information and script performance process data may not be structurally bound, which may prevent data analysis in the role dimension from being able to perform behavior clustering, script feedback analysis, or bias identification processing.

[0003] For example, in the group interaction script of ideological and political classrooms, different students choose different roles to enter the question and answer section. The answer content or selection results submitted by each student are recorded by the system, but this data may not be structurally integrated with the role identity tag, which may prevent the system from identifying abnormal operations under the role perspective in the backtracking analysis. SUMMARY

[0004] The purpose of the present application is to provide an immersive script experience data processing system based on role selection, which aims to solve the problems mentioned in the background.

[0005] To solve the above technical problems, the technical solution of the present application is as follows:

[0006] The immersive script experience data processing system based on role selection comprises:

[0007] A trajectory module for obtaining operation records of users and performing node sequence extraction and behavior sorting on the operation records to construct behavior trajectories under the role perspective and obtain behavior trajectory data;

[0008] A clustering module for performing paragraph division and clustering reorganization on the behavior trajectories according to the changes in node numbers in the behavior trajectory data to obtain role paragraph data;

[0009] A jump analysis module for calculating the number of jumps and the span difference between the role paragraph data and the preset node path of the script according to the node number span to obtain jump index data;

[0010] A first calculation module for identifying the deviation characteristics of the role behavior trajectories and calculating the behavior jump factor according to the jump index data.

[0011] a time sequence analysis module, configured to calculate a time interval sequence between consecutive nodes according to the role paragraph data, and perform time sequence stabilization analysis on the time interval sequence to obtain time sequence index data;

[0012] a second calculation module, configured to calculate a synchronization deviation degree of the role behavior time sequence according to an interval difference sequence between the time sequence index data and a preset behavior time sequence of the script, and obtain a time sequence offset factor;

[0013] a behavior analysis module, configured to combine the behavior jump factor and the time sequence offset factor into a role behavior feature vector, and perform deviation detection on each role number according to the role behavior feature vector to obtain a behavior analysis result.

[0014] Further, the trajectory module comprises:

[0015] an operation collection unit, configured to acquire a click operation, an option selection and a submission behavior of a user in a script performing process, extract an operation record field, and obtain original operation data;

[0016] a node extraction unit, configured to identify a page jump path and a corresponding script node number according to the original operation data, arrange the node numbers according to behavior events, and obtain node sequence data;

[0017] a behavior sorting unit, configured to re-sort the node numbers according to the node sequence data, arrange the node operation chain in ascending order of time, and obtain sorted behavior data;

[0018] a role binding unit, configured to map and match the node numbers in the sorted behavior data with a role number selected by a current user, construct a behavior trajectory of the role, and obtain behavior trajectory data.

[0019] Further, the clustering module comprises:

[0020] a jump identification unit, configured to calculate a change amplitude between adjacent node numbers in the behavior trajectory data, identify a position point with a change amplitude exceeding a preset amplitude, determine all jump positions, and obtain jump marker data;

[0021] a paragraph division unit, configured to segment the behavior trajectory data according to the jump positions in the jump marker data, divide the behavior trajectory data into a plurality of paragraph structures, and obtain initial paragraph data;

[0022] a paragraph fusion unit, configured to merge and reconstruct paragraphs with crossed numbers in the initial paragraph data, generate paragraphs with continuous numbers and consistent behaviors, and obtain fused paragraph data;

[0023] A paragraph clustering unit is configured to cluster and divide the paragraphs in the fused paragraph data according to the node number distribution characteristics, identify different paragraph labels, and obtain role paragraph data.

[0024] Further, the jump analysis module comprises:

[0025] A path loading unit is configured to extract node numbers corresponding to the role numbers from the system, arrange the node numbers in an extraction order, and obtain a script preset node path.

[0026] A number comparison unit is configured to match the role paragraph data with the node number sequence of the script preset node path, calculate the number difference between adjacent node numbers, and obtain number difference data.

[0027] A jump identification unit is configured to mark positions where the number difference is greater than a preset continuous threshold in the number difference data, extract non-continuous number jump points, and obtain jump position data.

[0028] A span measurement unit is configured to calculate the actual number difference of each jump according to the jump position data, record the span values of all jump events, and obtain span record data.

[0029] A frequency statistics unit is configured to count the jump events in the jump position data, determine the number of jumps, and combine the number of jumps with the span record data to obtain jump index data.

[0030] Further, the first calculation module comprises:

[0031] A span extraction unit is configured to extract the span values corresponding to all jump events from the jump index data, calculate the average span value and the maximum span value, and obtain span feature data.

[0032] A trajectory length unit is configured to calculate a trajectory length value according to the total number of node numbers in the role paragraph data, and obtain path length data.

[0033] A density generation unit is configured to operate the number of jumps and the path length data, calculate the jump proportion per unit path length, and obtain jump density data.

[0034] A behavior jump factor unit is configured to calculate the behavior jump factor of the role according to the span feature data and the jump density data.

[0035] Further, the behavior jump factor unit comprises:

[0036] A behavior jump factor calculation unit is configured to calculate the jump density according to the number of jumps and the trajectory length, calculate the jump fluctuation coefficient according to the span variance and the span mean value of all jump events, and calculate the span offset coefficient according to the maximum span value and the average span value.

[0037] According to the span variance and the span mean, a log correction term is constructed, and the log correction term is combined with the jump density to identify unstable jump behavior, to obtain a jump log response term; according to the average span, the span variance and the trajectory length, an overall span behavior intensity is calculated, to obtain a span intensity term, according to the maximum span and the jump fluctuation coefficient, a span fluctuation penalty term is constructed, the span intensity term is fused with the span fluctuation penalty term, to obtain a span structure intensity term; according to the span offset coefficient and the jump fluctuation coefficient, a behavior structure skewness adjustment term is constructed;

[0038] The jump log response term, the span structure intensity term and the behavior structure skewness adjustment term are fused, to obtain a behavior jump factor.

[0039] Further, the time sequence analysis module comprises:

[0040] A time extraction unit is configured to extract time points corresponding to each node from the character paragraph data, to obtain a standard time sequence;

[0041] An interval calculation unit is configured to perform difference operation on time points of adjacent nodes in the standard time sequence, to sequentially calculate time intervals between each jump, to obtain a time interval sequence;

[0042] A fluctuation measurement unit is configured to perform sliding window difference statistics on the time interval sequence, to identify a high fluctuation region, to obtain time fluctuation feature data;

[0043] A time sequence index unit is configured to calculate standard deviation and mean of the time interval according to the time fluctuation feature data, to obtain time sequence index data.

[0044] Further, the second calculation module comprises:

[0045] A time sequence loading unit is configured to extract standard time intervals between nodes in the script preset node path from the system, to constitute a script preset behavior time sequence;

[0046] An interval comparison unit is configured to compare the time sequence index data with the script preset behavior time sequence in terms of nodes, to calculate time interval difference values between each pair of nodes, to obtain interval difference sequence data;

[0047] A difference specification unit is configured to perform maximum value normalization processing on the interval difference sequence data, to obtain a normalized time difference sequence;

[0048] A fluctuation compensation unit is configured to perform amplification processing on the time interval difference values in the high fluctuation region according to the normalized time difference sequence, to obtain adjusted difference value data;

[0049] The time sequence offset factor unit is used for offset accumulation according to the adjustment difference data, identifying the synchronization deviation degree between the role behavior and the script, and calculating the time sequence offset factor of the role.

[0050] Further, the time sequence offset factor unit comprises:

[0051] The time sequence offset factor calculation unit is used for constructing a time interval difference absolute value sequence according to the time interval difference of all nodes, obtaining a median value; constructing a deviation enhancement term according to the ratio of the time offset value of each node to the median value, determining the relative offset degree of the current jump according to the ratio between the time interval difference and the maximum time interval difference, obtaining an amplitude normalization term, and obtaining a difference sum term according to the deviation enhancement term and the amplitude normalization term;

[0052] According to all time interval differences, the mean value and the standard deviation are calculated, and the overall instability degree of the behavior time sequence is further calculated, obtaining a fluctuation adjustment term; the difference value direction consistency rate is calculated according to the direction of each node time difference, obtaining a direction correction term;

[0053] The difference sum term, the fluctuation adjustment term and the direction correction term are fused to obtain the time sequence offset factor.

[0054] Further, the behavior analysis module comprises:

[0055] The factor combination unit is used for field splicing and vectorization processing of the behavior jump factor and the time sequence offset factor to obtain a role behavior feature vector.

[0056] The reference feature vector is constructed according to the preset node path of the script by the contrast construction unit.

[0057] The difference measurement unit is used for distance calculation, difference calculation and offset direction judgment of the role behavior feature vector and the reference feature vector to construct a behavior difference measurement value, obtaining offset vector data.

[0058] The judgment mapping unit is used for deviation classification and abnormal grading according to the offset vector data to obtain a behavior analysis result.

[0059] The above-mentioned scheme of the present application at least has the following beneficial effects:

[0060] The application realizes jump recognition and logical paragraph division of the script performance path by identifying the change trend of the node number in the role behavior track, not only can detect the mutation point between the node numbers, and generate jump mark data, but also further reconstructs the logical consistent behavior paragraph structure through paragraph fusion and number consistency analysis, the user operation data is recorded in linear time sequence, the traditional system does not perform semantic division on the behavior content or behavior stage in the script, and cannot reveal the operation behavior characteristics of students in different script stages, the system cuts the user behavior path into multiple structured paragraphs, and performs node feature clustering, so as to identify the behavior segment with semantic consistency or structural similarity in the role perspective, and provides data support for subsequent system intelligent script debugging.

[0061] The application corresponds one-to-one the user behavior track and the role perspective through node sequence extraction, operation behavior sorting and role number mapping, so that in the subsequent data analysis process, the operation path can be directly reconstructed from the role dimension, the behavior features are extracted, or the cross-role behavior comparison analysis is performed, the structural binding between the user behavior data and the selected role is realized, the traditional system cannot effectively distinguish whether the user behavior has role dependence, the application can accurately locate the behavior track of the specific role, so that more fine teaching management and script optimization process are realized.

[0062] The application extracts the node number structure of the preset path of the script, compares with the role paragraph path, calculates the number difference and measures the span, identifies the specific span and frequency of all script jump events, generates the jump factor, which can not only measure whether the behavior path is reasonable, but also judge whether the jump behavior has a potential abnormal trend based on the density and amplitude of the jump, the system can identify the script advancing path which does not conform to the logic in real time or retrospectively during the script experience process, and assist teachers to determine whether the students participate effectively and whether the key content nodes are skipped.

[0063] The application extracts the node time point sequence in each role behavior path, generates a time interval sequence through time difference operation, then detects the local fluctuation of the time interval using a sliding window mechanism, and then performs normalized comparison with the preset rhythm structure of the script to generate a time offset factor, which efficiently reveals whether the operation rhythm is stable and whether it is synchronized with the script rhythm, can automatically judge whether the behavior has reading or understanding missing risk, and provides a basis for teacher intervention, the application designs a complete behavior time sequence processing flow, supports modeling and offset analysis of the operation rhythm of the user in the script performance process.

[0064] The application fuses the behavior jump factor and the time sequence offset factor, constructs a role behavior feature vector, and judges the deviation degree and abnormal level of the role in the whole script execution process through the distance measurement, direction judgment and hierarchical mapping mechanism, presents the role behavior result as a clear vector structure, and intuitively displays the similarity and distance of the role behavior result and the standard behavior model, thereby effectively supporting the abnormal behavior identification, cheating behavior detection or personalized teaching strategy pushing in the script experience, and supporting the horizontal comparison across scripts and roles, having the application scene coverage capability, and providing a basis for a data-driven teaching system. BRIEF DESCRIPTION OF DRAWINGS

[0065] Figure 1 is a flow chart of a role selection-based immersive script experience data processing system provided by an embodiment of the application. DETAILED DESCRIPTION

[0066] Exemplary embodiments of the present disclosure will be described in greater detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be accurately conveyed to those skilled in the art.

[0067] As shown in Figure 1 , an embodiment of the application proposes a role selection-based immersive script experience data processing system, which comprises:

[0068] A trajectory module is configured to acquire operation records of a user, perform node sequence extraction and behavior sorting on the operation records, construct a behavior trajectory under a role perspective, and obtain behavior trajectory data.

[0069] A clustering module is configured to perform paragraph division and clustering reorganization on the behavior trajectory according to the change of node numbers in the behavior trajectory data, and obtain role paragraph data.

[0070] A jump analysis module is configured to calculate the number of jumps and the span difference according to the node number span between the role paragraph data and a preset node path of a script, and obtain jump index data.

[0071] A first calculation module is configured to identify the offset feature of the role behavior trajectory according to the jump index data, and calculate a behavior jump factor.

[0072] A time sequence analysis module is configured to calculate a time interval sequence between consecutive nodes according to the role paragraph data, and perform time sequence stability analysis on the time interval sequence, and obtain time sequence index data.

[0073] The second calculation module is configured to calculate a synchronization deviation degree of the role behavior time sequence according to an interval difference sequence between the time sequence index data and a preset behavior time sequence of the script, and obtain a time sequence deviation factor.

[0074] The behavior analysis module is configured to combine the behavior jump factor and the time sequence deviation factor into a role behavior feature vector, and perform deviation detection on each role number according to the role behavior feature vector to obtain a behavior analysis result.

[0075] In the embodiment of the present application, the trajectory module is configured to obtain operation records of the user, and perform node sequence extraction and behavior sorting on the operation records to construct a behavior trajectory under a role perspective, and obtain behavior trajectory data, so as to realize structural binding of user behavior and role identity, and ensure that the behavior path can be tracked in the role as the main dimension in subsequent analysis; the clustering module is configured to perform paragraph division and clustering reorganization on the behavior trajectory according to the change of the node number in the behavior trajectory data, and obtain role paragraph data, so as to divide the script operation behavior into stages with clear structure and coherent logic; and the jump analysis module is configured to calculate the number of jumps and the span difference value according to the node number span between the role paragraph data and a preset node path of the script, and obtain jump index data, so as to accurately identify the nonlinear jump behavior of the user in the script process, quantify the stability and deviation trend of the behavior path, and provide a data basis for abnormal path detection and script optimization suggestion.

[0076] The first calculation module is configured to identify the deviation feature of the role behavior trajectory according to the jump index data, calculate the behavior jump factor, comprehensively reflect the non-continuous degree and volatility of the role behavior path, effectively identify the abnormal path behavior, and provide a fine behavior structure analysis basis for the system; the time sequence analysis module is configured to calculate the time interval sequence between the continuous nodes according to the role paragraph data, and perform time sequence stability analysis on the time interval sequence to obtain time sequence index data, model the user operation behavior from the time dimension, and effectively reveal the behaviors such as skipping and jumping of the user; the second calculation module is configured to calculate the synchronization deviation degree of the role behavior time sequence according to the interval difference sequence between the time sequence index data and the preset behavior time sequence of the script, and obtain the time sequence deviation factor, so as to realize accurate measurement of the operation rhythm deviation from the script rhythm, and strengthen the perception ability of the system to the behavior rhythm control; and the behavior analysis module is configured to combine the behavior jump factor and the time sequence deviation factor into a role behavior feature vector, and perform deviation detection on each role number according to the role behavior feature vector to obtain a behavior analysis result, so as to realize high-accuracy role behavior abnormality identification and support individual behavior feedback.

[0077] In a preferred embodiment of the present application, the trajectory module comprises:

[0078] The operation collection unit is configured to acquire click operations, option selections and submission behaviors of a user in a script performing process, extract operation record fields and obtain original operation data;

[0079] The node extraction unit is configured to identify page jump paths and corresponding script node numbers according to the original operation data, arrange the node numbers according to behavior events and obtain node sequence data.

[0080] The behavior sorting unit is configured to reorder the node numbers according to the node sequence data, arrange the node operation chain in ascending order of time and obtain sorted behavior data.

[0081] The role binding unit is configured to map and match the node numbers in the sorted behavior data with a role number selected by a current user, construct a behavior track of the role and obtain behavior track data.

[0082] In the embodiment of the application, the operation collection unit is configured to acquire click operations, option selections and submission behaviors of a user in a script performing process, extract operation record fields and obtain original operation data, so as to ensure that subsequent behavior track analysis has complete data basis; the node extraction unit is configured to identify page jump paths and corresponding script node numbers according to the original operation data, arrange the node numbers according to behavior events and obtain node sequence data, so as to realize mapping from user operation behaviors to script structure numbers and provide an explicit structure reference for subsequent behavior track construction; the behavior sorting unit is configured to reorder the node numbers according to the node sequence data, arrange the node operation chain in ascending order of time and obtain sorted behavior data, so as to accurately restore actual behavior tracks of the user in a script performing process, eliminate order disorder problems caused by system delay, cache and the like and provide a reliable basis for behavior mode recognition and jump path analysis; and the role binding unit is configured to map and match the node numbers in the sorted behavior data with a role number selected by a current user, construct a behavior track of the role and obtain behavior track data, so as to realize strong binding between behavior paths and role identities and provide accurate data basis for subsequent operations.

[0083] The node extraction unit is configured to identify page jump paths and corresponding script node numbers according to the original operation data, arrange the node numbers according to behavior events and obtain node sequence data, and specifically includes:

[0084] The original operation data includes a user identifier, an operation timestamp, a page identifier, an operation type field, a jump link field and carried operation parameter information; in a system running process, a user performs script experience operations through a client interface, for example, clicks an option button, enters a next plot node, submits an answer result and the like, and these operations are recorded in the original operation data in a structured form.

[0085] Firstly, the system extracts the page identifier and the jump link field from the page, wherein the page identifier reflects the current script page number of the user, and the jump link field indicates the new page number or target node path jumped to after the operation behavior. When the user operation generates a jump behavior, the system first judges whether the jump is in line with the legal jump relationship in the script structure path, and extracts the current page number and the target page number. If the jump relationship is legal, the system takes the target page number as the script node number reached by this operation, and records the number into the node sequence draft; if the jump relationship cannot find the corresponding path in the preset script structure, the system marks it as an abnormal jump node.

[0086] The behavior event refers to an effective interaction of the user in the script performance process, for example, an option click or a submission operation, and each event corresponds to a unique timestamp and a page jump process. The system takes the event occurrence time as the main index, rearranges the node number sequence in time sequence, and merges, removes duplicates and cleanses the node numbers contained in the event. For example, if an behavior event contains multiple page jumps (such as a confirmation button jump followed by loading a story display page), the system will take the final target node as the main node number, retain the key path node number of the event, ensure that the node sequence reflects the user's main behavior path, and obtain the node sequence data.

[0087] In a preferred embodiment of the present application, the clustering module comprises:

[0088] The jump recognition unit is configured to calculate the change amplitude between adjacent node numbers in the behavior trajectory data, identify the position points with the change amplitude exceeding a preset amplitude, determine all jump positions, and obtain jump mark data.

[0089] The paragraph division unit is configured to divide the behavior trajectory data into a plurality of paragraph structures according to the jump positions in the jump mark data, and obtain initial paragraph data.

[0090] The paragraph fusion unit is configured to merge and reconstruct the paragraphs with crossed numbers in the initial paragraph data, generate paragraphs with continuous numbers and consistent behaviors, and obtain fused paragraph data.

[0091] The paragraph clustering unit is configured to cluster and divide the paragraphs in the fused paragraph data according to the node number distribution characteristics, identify different paragraph labels, and obtain role paragraph data.

[0092] In the embodiment of the present application, the jump identification unit is configured to calculate the change amplitude between adjacent node numbers in the behavior trajectory data, identify position points with a change amplitude exceeding a preset amplitude, determine all jump positions, and obtain jump marker data, so as to effectively identify the possible nonlinear jump behavior of the user in the script experience process or the boundary of different logical segments in the script structure, thereby constructing an initial logical segmentation basis. The paragraph division unit is configured to segment the behavior trajectory data according to the jump positions in the jump marker data, divide the behavior trajectory data into a plurality of paragraph structures, and obtain initial paragraph data, so as to effectively improve the granularity control of the behavior data. The paragraph fusion unit is configured to merge and reconstruct the paragraphs with crossed numbers in the initial paragraph data, generate paragraphs with continuous numbers and consistent behaviors, and obtain fusion paragraph data, so as to clean up redundancies, recombine boundaries, make the paragraph structure more stable and logical, and thereby improve the accuracy of the paragraph label generation. The paragraph clustering unit is configured to cluster and divide the paragraphs in the fusion paragraph data according to the node number distribution characteristics, identify different paragraph labels, and obtain role paragraph data, so as to enhance the abstraction ability of the system to the user behavior structure and realize the conversion from low-level operation data to high-level semantic structure.

[0093] The jump identification unit is configured to calculate the change amplitude between adjacent node numbers in the behavior trajectory data, identify position points with a change amplitude exceeding a preset amplitude, determine all jump positions, and obtain jump marker data, and specifically includes the following steps.

[0094] First, the node number sequence is extracted from the behavior trajectory data and arranged in time sequence. Then, the system calculates the difference between each pair of adjacent nodes in the node sequence to obtain a node number change amplitude sequence. To determine which changes belong to jump behavior, the system sets a predefined jump threshold, which is usually set by the script designer according to the average span of node numbers or the characteristics of the script structure, or obtained through system adaptive calculation, for example, set to the mean plus twice the standard deviation of the node number difference. Then the system traverses the difference sequence. For each difference, if it is greater than the jump threshold, it is determined that the point is a behavior jump point. The index position of the jump point represents that there is a significant number span between the behavior node of the user at this position and the previous node. The system records the position of the point as a jump position, and finally groups all node index positions that meet the jump conditions to obtain jump marker data.

[0095] The paragraph fusion unit is configured to merge and reconstruct the paragraphs with crossed numbers in the initial paragraph data, generate paragraphs with continuous numbers and consistent behaviors, and obtain fusion paragraph data, and specifically includes the following steps.

[0096] The system first reads the starting node number and the ending node number of all paragraphs in the initial paragraph data, sorts all paragraphs in ascending order according to the node number, and forms a paragraph number interval list. The system checks whether there is a number crossing relationship between adjacent paragraphs in sequence, and the judgment standard is: if the ending node number of the current paragraph is greater than the starting node number of the next paragraph, it means that there is a number crossing, and it is marked as an overlapping paragraph. For the paragraph group marked as overlapping, the system merges its number set to form a new node number set, and reconstructs the behavior sequence based on the appearance order of these numbers in the original behavior track. During the reconstruction process, the system will calculate the node number density (number of numbers per unit length), time span, behavior consistency (such as operation type distribution) and other indicators of the merged paragraph, to ensure that the merged result still maintains the behavior consistency in semantics. If the merged paragraph is significantly abnormal in structure or behavior characteristics, the system will retain the original paragraph and mark it as non-fusible. Finally, the system outputs all fused paragraphs or confirmed retained paragraphs to form fused paragraph data.

[0097] The paragraph clustering unit is configured to cluster and divide the paragraphs in the fused paragraph data according to node number distribution characteristics, identify different paragraph labels, and obtain role paragraph data, and specifically includes:

[0098] First, the core feature dimensions of each fused paragraph are extracted, including but not limited to: node number mean, node number variance, paragraph length (i.e. node number), node number span, node number density (number of access nodes per unit time), operation rhythm stability (such as standard deviation of node dwell time), etc. The system vectorizes these feature values to form a paragraph representation matrix, where each element is a vector containing the above feature dimensions. Then the system uses the K-means algorithm for preliminary classification. The system dynamically determines the optimal number of clusters according to the silhouette coefficient to ensure the separation and compactness of the clustering results. During the clustering process, the system classifies paragraphs with high similarity into the same class, and assigns a paragraph label to each clustering result, such as "smooth plot segment", "operation concentrated segment", "high-frequency jump segment", "backtracking segment", etc. These labels can be generated based on the clustering center feature vector through rule definition. After clustering, the system outputs the cluster number, paragraph label and feature vector of each paragraph to form a role paragraph data set.

[0099] In a preferred embodiment of the present application, the jump analysis module includes:

[0100] The path loading unit is configured to extract the node numbers corresponding to the role numbers from the system, and arrange the node numbers in the extraction order to obtain a preset node path of the script;

[0101] A number comparison unit is configured to match the character paragraph data with the node number sequence of the preset node path of the script, and calculate the number difference between adjacent node numbers to obtain number difference data;

[0102] A jump identification unit is configured to mark positions where the number difference is greater than a preset continuous threshold in the number difference data, extract non-continuous number jump points, and obtain jump position data.

[0103] A span measurement unit is configured to calculate the actual number difference of each jump according to the jump position data, record the span value of all jump events, and obtain span record data.

[0104] A frequency statistics unit is configured to count the jump events in the jump position data, determine the number of jumps, and combine the number of jumps with the span record data to obtain jump index data.

[0105] In the embodiment of the present application, the path loading unit is configured to extract the node numbers corresponding to the character numbers from the system, arrange the node numbers according to the extraction order, and obtain the preset node path of the script, so as to provide a reference benchmark for jump analysis and avoid relative errors in behavior difference judgment; the number comparison unit is configured to match the character paragraph data with the node number sequence of the preset node path of the script, and calculate the number difference between adjacent node numbers to obtain number difference data, so as to convert the script path difference into a measurable jump index and provide an explicit difference reference for subsequent jump identification; the jump identification unit is configured to mark positions where the number difference is greater than a preset continuous threshold in the number difference data, extract non-continuous number jump points, and obtain jump position data, so as to effectively detect node segments that obviously deviate from the script logical sequence in the user behavior path; the span measurement unit is configured to calculate the actual number difference of each jump according to the jump position data, record the span value of all jump events, and obtain span record data, so as to quantify the severity of each jump; and the frequency statistics unit is configured to count the jump events in the jump position data, determine the number of jumps, and combine the number of jumps with the span record data to obtain jump index data, so as to form a clear user behavior feature label and provide a data basis for subsequent use.

[0106] The span measurement unit is configured to calculate the actual number difference of each jump according to the jump position data, record the span value of all jump events, and obtain span record data, and specifically includes:

[0107] First, the jump position data output by the jump recognition unit is taken as the basis, which is a set of information containing jump event node number pairs, denoted as a jump position set. After obtaining the jump position data, the system traverses each jump record in the data set in turn, calculates the number difference for each pair of numbers, and obtains the result by using the absolute value operation method, which is the span value corresponding to the jump event. This calculation method considers the forward and reverse jump situations of the node number, ensuring that whether the behavior is forward jumping or backtracking, the span size between nodes can be reflected by a non-negative value. The system writes the result into the span record data in real time after calculating the span value of each jump.

[0108] In a preferred embodiment of the present application, the first calculation module comprises:

[0109] The span extraction unit is configured to extract the span values corresponding to all jump events from the jump indicator data, and calculate the average span value and the maximum span value to obtain span feature data.

[0110] The trajectory length unit is configured to calculate the trajectory length value according to the total number of node numbers in the role paragraph data to obtain path length data.

[0111] The density generation unit is configured to calculate the jump proportion per unit path length by operating the jump frequency and the path length data to obtain jump density data.

[0112] The behavior jump factor unit is configured to calculate the behavior jump factor of the role according to the span feature data and the jump density data.

[0113] In the embodiment of the present application, the span extraction unit is configured to extract the span values corresponding to all jump events from the jump indicator data, and calculate the average span value and the maximum span value to obtain span feature data, which effectively identifies the severity of structural changes in the behavior path of the script; the trajectory length unit is configured to calculate the trajectory length value according to the total number of node numbers in the role paragraph data to obtain path length data, which eliminates the calculation bias caused by different behavior lengths of different roles; the density generation unit is configured to calculate the jump proportion per unit path length by operating the jump frequency and the path length data to obtain jump density data, which reflects the continuity of the user behavior path; and the behavior jump factor unit is configured to calculate the behavior jump factor of the role according to the span feature data and the jump density data, which provides data support for subsequent intelligent discrimination and strategy adjustment of the system.

[0114] The trajectory length unit is configured to calculate the trajectory length value according to the total number of node numbers in the role paragraph data to obtain path length data, and specifically comprises:

[0115] First, the role paragraph data is parsed, and the node number sequence in each paragraph is extracted. To ensure the accuracy of path statistics, the track length unit will remove duplicate nodes, ensuring that the same nodes that appear repeatedly in multiple paragraphs are not counted repeatedly in the length statistics process, thereby eliminating the interference effects caused by node loops, reconstruction or clustering errors. Then, all the de-duplicated node numbers are combined in order according to the order of node appearance to form a complete behavior path chain. The system counts the number of nodes in the chain, i.e., records the number of independent nodes accessed by the user from the start node to the end node of the script. This statistical result is the total path length value of the behavior trajectory. In the scenario where the node number has a jump but is not repeatedly accessed, this statistical process will not exclude the jump section, and the path length still reflects the range of node access coverage, rather than physical continuity. The path length value is output as path length data in a structured data form.

[0116] In a preferred embodiment of the present application, the behavior jump factor unit comprises:

[0117] A behavior jump factor calculation unit is configured to calculate a jump density according to the number of jumps and the track length, calculate a jump fluctuation coefficient according to the span variance and the span mean value of all jump events, and calculate a span offset coefficient according to the maximum span value and the average span value.

[0118] According to the span variance and the span mean value, a logarithmic correction term is constructed, and the logarithmic correction term is combined with the jump density to identify unstable jump behavior, obtaining a jump logarithmic response term. According to the average span, the span variance and the track length, the overall span behavior intensity is calculated to obtain a span intensity term. According to the maximum span and the jump fluctuation coefficient, a span fluctuation penalty term is constructed. The span intensity term and the span fluctuation penalty term are fused to obtain a span structure intensity term. According to the span offset coefficient and the jump fluctuation coefficient, a behavior structure skewness adjustment term is constructed.

[0119] The jump logarithmic response term, the span structure intensity term and the behavior structure skewness adjustment term are fused to obtain the behavior jump factor.

[0120] In the embodiment of the present application, the behavior jump factor calculation unit is used to calculate a jump density according to the jump number and the track length, measure the concentration degree of the jump operation in the role behavior path; calculate a jump fluctuation coefficient according to the span variance and the span mean value of all jump events, measure the instability degree of the jump span; calculate a span offset coefficient according to the maximum span value and the average span value, measure the deviation degree between the maximum span and the average span; construct a logarithmic correction term according to the span variance and the span mean value, and combine the logarithmic correction term with the jump density to identify the unstable jump behavior, obtain a jump logarithmic response term, and amplify the behavior response risk under high jump fluctuation; calculate an overall span behavior intensity according to the average span, the span variance and the track length, obtain a span intensity term, construct a span fluctuation penalty term according to the maximum span and the jump fluctuation coefficient, and fuse the span intensity term with the span fluctuation penalty term to obtain a span structure intensity term, measure the overall influence degree of the jump behavior in the script path; construct a behavior structure skewness adjustment term according to the span offset coefficient and the jump fluctuation coefficient, and reveal whether the jump behavior exists a highly asymmetric deviation trend; fuse the jump logarithmic response term, the span structure intensity term and the behavior structure skewness adjustment term to obtain the behavior jump factor, and distinguish the normal push path from the abnormal deduction path.

[0121] The calculation formula of the behavior jump factor is:

[0122] ,

[0123] Among them, is the behavior jump factor of the role, is the index of the role, is the jump density, , is the jump number, is the track length value, is the maximum span value in all jump events, is the average span value of all jump events, is the span value variance of all jump events, is the span offset coefficient, , is the jump fluctuation coefficient, .

[0124] Among them, is the logarithmic correction term, is the jump logarithmic response term; is the span intensity term, is the span fluctuation penalty term, is the span structure intensity term; is the behavior structure skewness adjustment term. ​

[0125] In a preferred embodiment of the present application, the time sequence analysis module comprises:

[0126] a time extraction unit configured to extract time points corresponding to each node from the character paragraph data to obtain a standard time sequence;

[0127] an interval calculation unit configured to perform difference operation on the time points of adjacent nodes in the standard time sequence to sequentially calculate time intervals between each jump to obtain a time interval sequence;

[0128] a fluctuation measurement unit configured to perform sliding window difference statistics on the time interval sequence to identify a high fluctuation region to obtain time fluctuation feature data;

[0129] a time sequence index unit configured to calculate a standard deviation and a mean of the time intervals according to the time fluctuation feature data to obtain time sequence index data.

[0130] In the embodiment of the present application, the time extraction unit is configured to extract time points corresponding to each node from the character paragraph data to obtain a standard time sequence, which maps the user's operation behavior to an accurate time axis to provide a basis for dynamic behavior modeling; the interval calculation unit is configured to perform difference operation on the time points of adjacent nodes in the standard time sequence to sequentially calculate time intervals between each jump to obtain a time interval sequence, which reveals the fast and slow rhythm and time delay characteristics of the operation of the character in different script segments; the fluctuation measurement unit is configured to perform sliding window difference statistics on the time interval sequence to identify a high fluctuation region to obtain time fluctuation feature data, which detects a region with local intense fluctuation in the time sequence; and the time sequence index unit is configured to calculate a standard deviation and a mean of the time intervals according to the time fluctuation feature data to obtain time sequence index data, which realizes structured modeling of time sequence characteristics to provide a basis for subsequent character abnormal behavior recognition.

[0131] The fluctuation measurement unit is configured to perform sliding window difference statistics on the time interval sequence to identify a high fluctuation region to obtain time fluctuation feature data, and specifically comprises:

[0132] First, the time interval sequence is preprocessed to ensure that the time interval values are all non-negative, and all time interval values are normalized to unify the scale dimension to avoid dimensional interference on subsequent fluctuation detection caused by operation time length difference between nodes. Then, the time interval sequence is traversed according to a set sliding window width (for example, 3, 5 or 7), and for each time interval subsequence in the sliding window, the variance value of the subsequence is calculated to reflect the uniformity of the user's operation rhythm in the current time window. The greater the variance, the more intense the operation rhythm change of the user in the paragraph, and there are behaviors such as abnormal pause and sudden jump.

[0133] After the traversal of the sliding window is completed, the system will obtain a set of local variance sequences that are equal or approximately equal in length to the original time interval sequence, referred to as a local fluctuation curve. The system sets a fluctuation threshold, and when the local variance corresponding to a certain sliding window exceeds the fluctuation threshold, it is considered to be a high fluctuation region, and the fluctuation threshold is a multiple of the global variance, such as 1.5 times. If multiple adjacent windows are in a high fluctuation state, they can be combined into an integral fluctuation segment; if an isolated window meets the threshold but differs greatly from the previous and subsequent windows, it can be screened out by calculating the covariance with adjacent windows to eliminate accidental false alarm regions. Finally, all time intervals determined to be high fluctuation are marked to obtain time fluctuation feature data.

[0134] In a preferred embodiment of the present application, the second calculation module comprises:

[0135] a time sequence loading unit for extracting standard time intervals between nodes in the script preset node path from the system to form a script preset behavior time sequence;

[0136] an interval comparison unit for comparing the time sequence index data with the script preset behavior time sequence node by node, calculating the time interval difference between each pair of nodes, and obtaining interval difference sequence data;

[0137] a difference specification unit for performing maximum value normalization processing on the interval difference sequence data to obtain normalized time difference sequence;

[0138] a fluctuation compensation unit for performing amplification processing on the time interval difference in the high fluctuation region according to the normalized time difference sequence to obtain adjusted difference value data;

[0139] a time sequence offset factor unit for performing offset accumulation according to the adjusted difference value data to identify the degree of synchronization deviation between the role behavior and the script, and calculate the time sequence offset factor of the role.

[0140] In the embodiment of the present application, the time sequence loading unit is used to extract standard time intervals between nodes in the script preset node path from the system to form a script preset behavior time sequence, ensuring that the system has a reference rhythm model, so that the subsequent analysis of the actual operation rhythm of the user has a unified reference. The interval comparison unit is used to compare the time sequence index data with the script preset behavior time sequence node by node, calculate the time interval difference between each pair of nodes, and obtain interval difference sequence data, establish a rhythm deviation mapping relationship between the actual behavior of the role and the preset standard of the script, so that the system can identify the difference in operation rhythm at each node level.

[0141] The difference specification unit is configured to perform maximum value normalization on the interval difference sequence data to obtain a normalized time difference sequence, so as to eliminate the incomparability caused by the difference between the script nodes or the absolute magnitude difference of the operation duration; the fluctuation compensation unit is configured to perform amplification processing on the time interval difference in the high fluctuation region according to the normalized time difference sequence to obtain adjusted difference data, so as to effectively amplify the influence of the region with a dramatic rhythm fluctuation in the overall behavior evaluation; and the time sequence offset factor unit is configured to perform offset accumulation according to the adjusted difference data, identify the degree of synchronization deviation between the role behavior and the script, calculate a time sequence offset factor of the role, and reflect the real execution state of the user behavior rhythm, so as to provide data support for the subsequent intelligent discrimination and strategy adjustment of the system.

[0142] The fluctuation compensation unit is configured to perform amplification processing on the time interval difference in the high fluctuation region according to the normalized time difference sequence to obtain adjusted difference data, and specifically includes:

[0143] The system performs dynamic amplification processing on the time interval difference in the high fluctuation region, and assigns a dynamic adjustment factor to each time interval item marked as high fluctuation. The value of the adjustment factor can be calculated according to the standard deviation size of the window in which the time point is located and the center deviation degree of the position in the window. Two factors are given priority: first, the greater the fluctuation intensity, the higher the corresponding adjustment factor should be; and second, the time point in the fluctuation center should obtain a higher weight, so the adjustment factor The calculation formula is: wherein, is a global adjustment coefficient, is the standard deviation in the current window, is the maximum value of the standard deviations of all windows, is the index of the current time interval item, is the index of the window center, is the window width. The system applies the dynamic adjustment factor to each item of the normalized time difference sequence to calculate the adjusted time interval difference, thereby forming the adjusted difference data. For the time interval item not marked as the high fluctuation region, the adjustment factor is 1, and the original normalized value remains unchanged. Finally, all the adjusted time interval difference data are the adjusted difference data.

[0144] In a preferred embodiment of the present application, the time sequence offset factor unit includes:

[0145] The time sequence offset factor calculation unit is used for constructing a time interval difference absolute value sequence according to the time interval difference values of all nodes, obtaining a median value, avoiding the overall offset misjudgment problem caused by individual abnormal values, and providing a more stable reference baseline for subsequent deviation degree measurement; constructing a deviation degree enhancement term according to the ratio of the time offset value of each node to the median value, amplifying the weight of the behavior node with a larger deviation degree, determining the relative offset degree of the current jump according to the ratio between the time interval difference value and the maximum time interval difference value, obtaining an amplitude normalization term, providing a data basis for the horizontal comparison analysis of the offset degree of multiple roles, obtaining a difference value sum term according to the deviation degree enhancement term and the amplitude normalization term, providing support for behavior visualization sorting and script optimization target priority setting of the system; calculating the mean value and the standard deviation according to all time interval difference values, and further calculating the overall instability degree of the behavior time sequence, obtaining a fluctuation adjustment term, judging whether the behavior execution is stable; calculating the difference value direction consistency rate according to the direction of each node time difference value, obtaining a direction correction term, distinguishing the role behavior which is stable but overall offset; fusing the difference value sum term, the fluctuation adjustment term and the direction correction term, obtaining the time sequence offset factor, representing the stability and synchronization degree of the role behavior in the time rhythm.

[0146] The mean value and the standard deviation are calculated according to all time interval difference values, and the overall instability degree of the behavior time sequence is further calculated, obtaining a fluctuation adjustment term; the difference value direction consistency rate is calculated according to the direction of each node time difference value, obtaining a direction correction term.

[0147] The difference value sum term, the fluctuation adjustment term and the direction correction term are fused, obtaining the time sequence offset factor.

[0148] In the embodiment of the application, the time sequence offset factor calculation unit is used for constructing a time interval difference absolute value sequence according to the time interval difference values of all nodes, obtaining a median value, avoiding the overall offset misjudgment problem caused by individual abnormal values, and providing a more stable reference baseline for subsequent deviation degree measurement; constructing a deviation degree enhancement term according to the ratio of the time offset value of each node to the median value, amplifying the weight of the behavior node with a larger deviation degree, determining the relative offset degree of the current jump according to the ratio between the time interval difference value and the maximum time interval difference value, obtaining an amplitude normalization term, providing a data basis for the horizontal comparison analysis of the offset degree of multiple roles, obtaining a difference value sum term according to the deviation degree enhancement term and the amplitude normalization term, providing support for behavior visualization sorting and script optimization target priority setting of the system; calculating the mean value and the standard deviation according to all time interval difference values, and further calculating the overall instability degree of the behavior time sequence, obtaining a fluctuation adjustment term, judging whether the behavior execution is stable; calculating the difference value direction consistency rate according to the direction of each node time difference value, obtaining a direction correction term, distinguishing the role behavior which is stable but overall offset; fusing the difference value sum term, the fluctuation adjustment term and the direction correction term, obtaining the time sequence offset factor, representing the stability and synchronization degree of the role behavior in the time rhythm.

[0149] The calculation formula of the time sequence offset factor is:

[0150] ,

[0151] Wherein, is the time sequence offset factor of the role, is the index of the role, is the index of the node in the behavior track, is the total number of nodes, is the first is the second the time interval difference value of the i th node, , the standard time interval of the i th node in the script preset node path, the time interval of the i th node in the timing index data, the standard time interval of the i th node in the script preset node path, the median value of the time interval difference value of all nodes, the maximum value of the time interval difference value of all nodes, the standard deviation of the time interval difference value, , , the mean value of the time interval difference value, , the difference value direction consistency rate, , , is a conditional judgment function, when is true, then , otherwise 0.

[0152] wherein, is a deviation enhancement term, is an amplitude normalization term, is a difference sum term; is a fluctuation adjustment term; is a direction correction term.

[0153] wherein, is the median value of the time interval difference value of all nodes. When is odd, is the time interval difference value of the i th node; when is even, is the time interval difference value of the i th node and the i th node. is half of the sum of the time interval difference value of the i th node and the i th node. In a preferred embodiment of the present application, the behavior analysis module comprises:

[0154] a factor combination unit for field splicing and vector processing of the behavior jump factor and the timing offset factor to obtain a role behavior feature vector;

[0155] a reference construction unit for constructing a reference feature vector according to the script preset node path;

[0156] a difference measurement unit for distance calculation, difference calculation and offset direction judgment of the role behavior feature vector and the reference feature vector to construct a behavior difference measurement value and obtain a deviation vector data;

[0157] a difference measurement unit for distance calculation, difference calculation and offset direction judgment of the role behavior feature vector and the reference feature vector to construct a behavior difference measurement value and obtain a deviation vector data;

[0158] ​A judgment mapping unit is configured to perform deviation classification and abnormality grading according to the deviation vector data, and obtain the behavior analysis result.

[0159] In the embodiment of the present application, the factor combination unit is configured to perform field splicing and vectorization processing on the behavior jump factor and the time sequence offset factor, to obtain the role behavior feature vector, and to unify the structure of the discrete calculation result, so as to provide a mathematical basis and a measurable standard for subsequent difference judgment and scoring system construction; the reference feature vector is constructed according to the preset node path of the script by the reference construction unit, so as to ensure that the system has a unified target model when judging the behavior difference; the difference measurement unit is configured to perform distance calculation, difference calculation and offset direction judgment on the role behavior feature vector and the reference feature vector, to construct the behavior difference measurement value, to obtain the deviation vector data, and to realize accurate comparison between the role behavior and the standard model; the judgment mapping unit is configured to perform deviation classification and abnormality grading according to the deviation vector data, to obtain the behavior analysis result, and to realize evaluation and abnormality identification of the role behavior deviation degree.

[0160] The factor combination unit is configured to perform field splicing and vectorization processing on the behavior jump factor and the time sequence offset factor, to obtain the role behavior feature vector, and specifically includes:

[0161] The behavior jump factor represents the jump amplitude and frequency characteristics of the role on the node number path, and the time sequence offset factor represents the deviation degree and rhythm stability of the role on the node time interval. First, the system establishes a feature structure body according to the role number, reads two factor data corresponding to each role number, and splices the two factors into a unified structure feature vector according to a preset field order by using a field splicing operation. The feature vector is represented by a floating point number array and maintains dimensional consistency in the vector space, so as to facilitate subsequent comparison processing. In the splicing operation, the system sets the behavior jump factor corresponding field in the front half and the time sequence offset factor in the back half of the combined vector. Then, the system performs normalization processing on each component of the feature vector, eliminates the scale difference between the behavior jump factor and the time sequence offset factor in the order of magnitude, and makes the two factors comparable, to obtain the role behavior feature vector.

[0162] The judgment mapping unit is configured to perform deviation classification and abnormality grading according to the deviation vector data, and obtain the behavior analysis result, and specifically includes:

[0163] Firstly, the system divides the deviation intensity into multiple intervals according to the hierarchical judgment rule of the configuration setting, for example, setting 0.0-0.2 as a normal interval, 0.2-0.4 as a slight deviation, 0.4-0.6 as a moderate deviation, and 0.6 or above as a serious deviation. The system corresponds the deviation intensity value of each role to a specific interval to obtain the deviation level of the role in the process of executing the script behavior. If the deviation value of a role is higher than the preset maximum deviation threshold, the system marks it as an abnormal role and stores it in the abnormal behavior list.

[0164] Then, the system further subdivides the deviation type in combination with the deviation direction information, for example: if the behavior jump factor deviation value is much larger than the time shift factor deviation value, it is identified as a structure jump type deviation; if the time shift factor deviation value is dominant, it is identified as a rhythm imbalance type deviation; if both are significantly deviated, it is determined as a global behavior anomaly. Finally, the system packs the role number, deviation level, deviation type and related vector value to generate an analysis result record to obtain the behavior analysis result.

[0165] The above is the preferred embodiment of the present application. It should be noted that for those skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, which should also be considered within the scope of protection of the present application.

Claims

1. A role selection based immersive script experience data processing system, characterized in that, The system comprises: a trajectory module for obtaining operation records of a user and performing node sequence extraction and behavior sorting on the operation records to build behavior trajectories under a role perspective and obtain behavior trajectory data; a clustering module for performing paragraph division and clustering reorganization on the behavior trajectories according to changes in node numbers in the behavior trajectory data to obtain role paragraph data; a jump analysis module for counting the number of jumps and span differences according to the node number span between the role paragraph data and a preset node path of a script to obtain jump index data; a first calculation module for identifying the deviation characteristics of role behavior trajectories according to the jump index data and calculating a behavior jump factor; a time sequence analysis module for calculating the time interval sequence between consecutive nodes according to the role paragraph data and performing time sequence stability analysis on the time interval sequence to obtain time sequence index data; a second calculation module for calculating the degree of synchronization deviation of role behavior time sequences according to the interval difference sequence between the time sequence index data and a preset behavior time sequence of a script to obtain a time sequence deviation factor; a behavior analysis module for combining the behavior jump factor and the time sequence deviation factor into a role behavior feature vector and performing deviation detection on each role number according to the role behavior feature vector to obtain a behavior analysis result.

2. The immersive scripted experience data processing system based on character selection of claim 1, wherein, The trajectory module comprises: an operation collection unit for obtaining click operations, option selections and submission behaviors of a user during the progress of a script, extracting operation record fields and obtaining original operation data; a node extraction unit for identifying page jump paths and corresponding script node numbers according to the original operation data and arranging the node numbers according to behavior events to obtain node sequence data; a behavior sorting unit for reordering the node numbers according to the node sequence data, arranging the node operation chain in ascending order of time to obtain sorted behavior data; a role binding unit for mapping and matching the node numbers in the sorted behavior data with the role number selected by a current user to build the behavior trajectory of the role and obtain behavior trajectory data.

3. The immersive scripted experience data processing system based on character selection of claim 2, wherein, The clustering module comprises: a jump identification unit for calculating the change amplitude between adjacent node numbers in the behavior trajectory data, identifying position points with a change amplitude exceeding a preset amplitude, determining all jump positions and obtaining jump marker data; a paragraph division unit for segmenting the behavior trajectory data according to the jump positions in the jump marker data to divide the behavior trajectory data into a plurality of paragraph structures and obtain initial paragraph data; a paragraph fusion unit for merging and reconstructing paragraphs with crossed numbers in the initial paragraph data to generate paragraphs with continuous numbers and consistent behaviors and obtain fused paragraph data; a paragraph clustering unit for clustering and dividing the paragraphs in the fused paragraph data according to the node number distribution characteristics to identify different paragraph labels and obtain role paragraph data.

4. The immersive scripted experience data processing system based on character selection of claim 3, wherein, The jump analysis module comprises: a path loading unit for extracting node numbers corresponding to a role number from a system and arranging the node numbers in extraction order to obtain a preset node path of a script; A number comparison unit is configured to match the character paragraph data with the node number sequence of the script preset node path, and calculate the number difference between adjacent node numbers to obtain number difference data; A jump identification unit is configured to mark positions where the number difference is greater than a preset continuous threshold in the number difference data, extract non-continuous number jump points, and obtain jump position data; A span measurement unit is configured to calculate the actual number difference of each jump according to the jump position data, record the span value of all jump events, and obtain span record data; A frequency statistics unit is configured to count the jump events in the jump position data, determine the number of jumps, and combine the number of jumps with the span record data to obtain jump index data.

5. The immersive scripted experience data processing system based on character selection of claim 4, wherein, The first calculation module includes: A span extraction unit is configured to extract the span value corresponding to all jump events from the jump index data, and calculate the average span value and the maximum span value to obtain span feature data; A trajectory length unit is configured to calculate the trajectory length value according to the total number of node numbers in the character paragraph data to obtain path length data; A density generation unit is configured to calculate the jump proportion per unit path length by operating the number of jumps and the path length data to obtain jump density data; An action jump factor unit is configured to calculate the action jump factor of the character according to the span feature data and the jump density data.

6. The immersive scripted experience data processing system based on character selection of claim 5, wherein, The action jump factor unit includes: An action jump factor calculation unit is configured to calculate the jump density according to the number of jumps and the trajectory length, calculate the jump fluctuation coefficient according to the span variance and the span mean value of all jump events, and calculate the span offset coefficient according to the maximum span value and the average span value; A log correction term is constructed according to the span variance and the span mean value, and the log correction term is combined with the jump density to identify unstable jump behaviors to obtain a jump log response term; the overall span behavior intensity is calculated according to the average span, the span variance and the trajectory length to obtain a span intensity term, a span fluctuation penalty term is constructed according to the maximum span and the jump fluctuation coefficient, and the span intensity term is fused with the span fluctuation penalty term to obtain a span structure intensity term; an action structure skewness adjustment term is constructed according to the span offset coefficient and the jump fluctuation coefficient; The jump log response term, the span structure intensity term and the action structure skewness adjustment term are fused to obtain the action jump factor.

7. The immersive scripted experience data processing system based on character selection of claim 6, wherein, The time sequence analysis module includes: A time extraction unit is configured to extract the time points corresponding to each node from the character paragraph data to obtain a standard time sequence; An interval calculation unit is configured to perform difference operation on the time points of adjacent nodes in the standard time sequence, and sequentially calculate the time interval between each jump to obtain a time interval sequence; A fluctuation measurement unit is configured to perform sliding window difference statistics on the time interval sequence to identify a high fluctuation region to obtain time fluctuation feature data; A time sequence index unit is configured to calculate the standard deviation and the mean value of the time interval according to the time fluctuation feature data to obtain time sequence index data.

8. The immersive scripted experience data processing system based on character selection of claim 7, wherein, The second calculation module includes: A time sequence loading unit is configured to extract the standard time interval between each node in the script preset node path from the system to construct a script preset action time sequence; The interval comparison unit is configured to compare the timing index data with the preset behavior timing of the script, calculate the time interval difference between each pair of nodes, and obtain interval difference sequence data. The difference specification unit is configured to perform maximum value normalization on the interval difference sequence data to obtain normalized time difference sequence data. The fluctuation compensation unit is configured to perform amplification processing on the time interval difference in a high fluctuation region according to the normalized time difference sequence to obtain adjusted difference data. The timing offset factor unit is configured to perform offset accumulation according to the adjusted difference data, identify the degree of synchronization deviation between the role behavior and the script, and calculate the timing offset factor of the role.

9. The immersive scripted experience data processing system based on character selection of claim 8, wherein, The timing offset factor unit includes: The timing offset factor calculation unit is configured to construct a time interval difference absolute value sequence according to the time interval difference of all nodes to obtain a median value, construct a deviation enhancement term according to the ratio of the time offset value of each node to the median value, determine the relative offset degree of the current jump according to the ratio between the time interval difference and the maximum time interval difference to obtain an amplitude normalization term, and obtain a difference total term according to the deviation enhancement term and the amplitude normalization term. The timing offset factor calculation unit is configured to calculate the mean and standard deviation of all time interval differences, further calculate the overall instability degree of the behavior timing, obtain a fluctuation adjustment term, calculate the difference value direction consistency rate according to the direction of each node time difference, and obtain a direction correction term. The timing offset factor is obtained by fusing the difference total term, the fluctuation adjustment term, and the direction correction term.

10. The immersive scripted experience data processing system based on character selection of claim 9, wherein, The behavior analysis module includes: The factor combination unit is configured to perform field splicing and vectorization processing on the behavior jump factor and the timing offset factor to obtain a role behavior feature vector. The reference feature vector is constructed according to the preset node path of the script. The difference measurement unit is configured to calculate the distance, difference, and offset direction of the role behavior feature vector and the reference feature vector to construct a behavior difference measurement value and obtain deviation vector data. The judgment mapping unit is configured to perform deviation classification and abnormal grading according to the deviation vector data to obtain a behavior analysis result.

Images