Internet shopping mall management platform system

Abstract

The invention relates to the technical field of platform management, in particular to an internet shopping mall management platform system. The system comprises the following modules: a track acquisition module, a track clustering module, a cognitive optimization module and a conflict processing module, specifically, periodically acquiring user mouse coordinates, when the distance between a starting point and an ending point of a mouse track in a preset range is detected to be smaller than a first threshold value and the total length of a path is detected to be larger than a second threshold value, judging that the mouse track is a convolution track, outputting convolution trajectory data; the convolution trajectories of the multiple users in a preset time period are overlapped according to page positions, when the trajectory overlapping degree exceeds a preset overlapping threshold value, a trajectory cluster is formed, and the geometric center of the trajectory cluster is calculated based on convolution trajectory data to serve as a cognitive hotspot; and judging a group cognitive impairment point according to the number of the user sources of the cognitive hotspot. According to the method, the cognitive confusion moment of the user can be captured in real time through dynamic sampling frequency adjustment, and group cognitive impairment and individual preference are accurately distinguished through multi-user trajectory overlay analysis.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of platform management, and particularly relates to an Internet mall management platform system. BACKGROUND

[0002] The types and quantities of goods of the Internet mall platform are exponentially increasing, and it is becoming more and more difficult for users to find target goods in the mass of goods. The existing mall management platform mainly relies on search engines, recommendation algorithms and classification navigation to help users find goods, but these technical solutions have the following shortcomings: the traditional user behavior analysis technology is mainly based on click rate, dwell time, bounce rate and other macro indicators. These indicators can only reflect the final result of user behavior, and cannot capture the cognitive difficulties of users in the decision-making process. For example, when a user hesitates between two similar goods categories, the traditional analysis method can only record which option the user finally clicks, but cannot identify the confusion and hesitation experienced by the user in the selection process. The existing page optimization scheme mainly adopts A / B testing and other methods, which require a large number of samples and a long test period to draw a conclusion, and the optimization result is often based on statistical average value, and cannot accurately locate the specific cognitive obstacle point. Although the heat map tool can display the user's attention area, it cannot distinguish between normal browsing behavior and cognitive confusion behavior, resulting in a lack of pertinence in optimization. SUMMARY

[0003] Therefore, it is necessary to provide an Internet mall management platform system to solve at least one of the above technical problems.

[0004] To achieve the above-mentioned purpose, an Internet mall management platform system comprises the following modules: A trajectory acquisition module is configured to periodically acquire user mouse coordinates. When it is detected that the distance between the starting point and the ending point of the mouse trajectory in a preset range is less than a first threshold value and the total length of the path is greater than a second threshold value, it is determined that the trajectory is a spiral trajectory, and spiral trajectory data is output. A trajectory clustering module is configured to superimpose the spiral trajectories of multiple users in a preset time period according to page positions. When the trajectory overlap degree exceeds a preset overlap threshold, a trajectory cluster is formed. The geometric center of the trajectory cluster is calculated based on the spiral trajectory data as a cognitive hot area. A cognitive optimization module is configured to determine a group cognitive obstacle point according to the number of user sources of the cognitive hot area. The similarity of a pair of page elements in the group cognitive obstacle point is calculated. When the similarity exceeds a preset similarity threshold, a difference keyword is extracted to generate an optimization instruction. A conflict processing module is configured to receive the optimization instruction, execute an optimization strategy based on the cognitive conflict type according to the time distribution characteristics of the spiral trajectory, and feed back the trajectory change rate before and after optimization to the cognitive analysis module to form a closed-loop optimization.

[0005] The present application realizes accurate identification and dynamic optimization of page cognitive conflict and operation obstacles through real-time collection and analysis of user mouse revolving track. Firstly, based on analysis of displacement vector and track segment, the behavior of user wandering or repeatedly comparing in the page can be captured, so as to identify potential revolving track, break through the limitation of traditional click and dwell time statistics, and more intuitively reflect user confusion. Secondly, the track is mapped to a virtual grid and the spatial and timing similarity is calculated, so that the revolving behavior can be aggregated and analyzed to form a cognitive hot area and accurately locate confusing elements. Thirdly, through calculation of revolving center, radius and duration, the system quantifies the cognitive intensity of the hot area, and combines weighted center and outlier splitting to depict single-center or double-center hot area and reveal the cognitive distribution of user groups. Further, through analysis of the number of independent users, access frequency and dispersion, the group cognitive obstacle points are determined, and optimization instructions are automatically generated based on element similarity and difference attributes, and targeted improvement strategies are proposed, including immediate visual enhancement and delayed information supplement. After optimization, the system monitors the track change rate, compares the effects before and after optimization, automatically solidifies or reverts the strategy, and forms a data-driven closed-loop optimization mechanism. This method not only can find problems that traditional indicators cannot capture, but also can output executable improvement schemes, thereby significantly improving the clarity, efficiency and user experience of page interaction. BRIEF DESCRIPTION OF DRAWINGS

[0006] Other features, objects and advantages of the present application will become more apparent from the following detailed description of non-limiting embodiments made with reference to the accompanying drawings: Figure 1 A module content schematic diagram of an Internet mall management platform system of the present application; Figure 2 A multi-user track superposition and clustering schematic diagram of an embodiment of the present application; Figure 3 A user interface before and after optimization comparison diagram of an embodiment of the present application Figure 4 A cognitive hot area and obstacle point identification diagram of an embodiment of the present application. DETAILED DESCRIPTION

[0007] The technical method of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0008] In addition, the accompanying drawings are included to provide a further understanding of the present application, and are incorporated in and constitute a part of this specification. The drawings illustrate embodiments of the present application and, together with the description, serve to explain the principles of the present application. In the drawings:

[0009] It is to be understood that, although terms such as "first", "second", and so on can be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, without departing from the scope of the exemplary embodiments, a first element can be referred to as a second element, and similarly a second element can be referred to as a first element. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0010] To achieve the above object, please refer to Figures 1 to 4 The present application provides an Internet mall management platform system, comprising the following modules: S1: a trajectory acquisition module, configured to periodically acquire user mouse coordinates, and determine a spiral trajectory when detecting that the distance between the start point and the end point of the mouse trajectory within a preset range is less than a first threshold value and the total length of the path is greater than a second threshold value, and output spiral trajectory data; S2: a trajectory clustering module, configured to superimpose spiral trajectories of multiple users within a preset time period according to page positions, form a trajectory cluster when the trajectory overlap degree exceeds a preset overlap threshold, and calculate the geometric center of the trajectory cluster as a cognitive hot area based on the spiral trajectory data; S3: a cognitive optimization module, configured to determine a group cognitive barrier point according to the number of user sources of the cognitive hot area, calculate the similarity of page element pairs within the group cognitive barrier point, extract a difference keyword when the similarity exceeds a preset similarity threshold, and generate an optimization instruction; S4: a conflict processing module, configured to receive the optimization instruction, execute an optimization strategy based on the cognitive conflict type according to the time distribution characteristics of the spiral trajectory, and feed back the trajectory change rate before and after optimization to the cognitive analysis module to form a closed-loop optimization.

[0011] Further, the trajectory acquisition module comprises: coarsely sampling the user mouse coordinates at a first sampling frequency, and recording the displacement vector between consecutive coordinate points; In an embodiment, the user mouse coordinates are coarsely sampled at a first sampling frequency, and the system periodically acquires the mouse position points of the user on the page and calculates the displacement vectors between the continuous coordinate points. In this way, a low-density trajectory sequence can be formed to preliminarily determine the mouse movement trend of the user.

[0012] For example, when the first sampling frequency is set to 20 times per second, if the user moves the mouse along a zigzag path on the page, the system will record several key coordinate points and calculate the displacement direction and length between adjacent points to obtain a sequence of displacement vectors.

[0013] When the sum of the direction angles of the three consecutive displacement vectors exceeds 180 degrees, the fine sampling mode is triggered, and the sampling frequency is increased to a second sampling frequency, wherein the second sampling frequency is greater than the first sampling frequency. In an embodiment, when the system detects that the sum of the direction angles of the three consecutive displacement vectors exceeds 180 degrees, it is determined that the user has a more complex mouse trajectory behavior, thereby triggering the fine sampling mode. In the fine sampling mode, the sampling frequency is increased from the first sampling frequency to the second sampling frequency to capture the trajectory details more carefully.

[0014] For example, when the user performs a fast circular drag on the page, the angles between the three consecutive displacement vectors may be 80 degrees, 70 degrees and 50 degrees, respectively, and the sum is 200 degrees, which is greater than 180 degrees. The system immediately switches to a fine sampling frequency of 100 times per second, so that the curved features of the trajectory can be accurately captured.

[0015] It should be noted that this determination method can effectively filter out normal straight lines or slow movements, thereby only increasing the resolution when complex trajectories appear, saving system resources.

[0016] In the fine sampling mode, the collected coordinate sequence is divided into sliding windows of length N, the start and end points of the trajectory in each window are determined, and the straight line distance between the start and end points is calculated. In an embodiment, in the fine sampling mode, the system divides the collected high-frequency coordinate sequence into sliding windows of length N for separate analysis. For each sliding window, the system extracts the start and end point coordinates and calculates the straight line distance between them to determine the overall shape of the trajectory segment.

[0017] For example, when N is set to 50, the start point of the coordinate sequence collected in a window is (100, 200) and the end point is (105, 195), and the system calculates the straight line distance between the start and end points to be about 7 pixels. In this way, it can quickly detect whether the trajectory segment exhibits a local closure or a swirling trend.

[0018] When the straight-line distance between the start point and the end point is less than the first threshold value, the path length verification is entered, the spiral trajectory is determined, and spiral trajectory data is output.

[0019] In an embodiment, when the straight-line distance between the start point and the end point of a certain sliding window is less than the first threshold value, the system enters the path length verification stage, that is, the actual path length of the trajectory is calculated. The actual path length is obtained by adding up the distances between all adjacent coordinate points in the window, and when the path length is greater than the second threshold value, it is determined that the segment is a spiral trajectory.

[0020] For example, when the first threshold value is set to 10 pixels and the second threshold value is set to 100 pixels, if the straight-line distance between the start point and the end point of a certain trajectory window is only 5 pixels, but the path cumulative length reaches 150 pixels, it indicates that the trajectory has multiple turns in a small range, and the system finally determines that it is a spiral trajectory and outputs the corresponding data.

[0021] Further, when the straight-line distance between the start point and the end point is less than the first threshold value, the path length verification is entered, the spiral trajectory is determined, and spiral trajectory data is output, specifically: The actual path total length of the trajectory segment satisfying the first threshold value condition is calculated, and the actual path total length is the sum of the distances between all adjacent coordinate points. When the path total length is greater than the second threshold value, it is determined to be a preliminary spiral trajectory. In an embodiment, for the trajectory segment satisfying the first threshold value condition, the actual path total length can be calculated by the point-by-point accumulation method. Specifically, the straight-line distances between adjacent coordinate points in the trajectory segment are added one by one until the entire trajectory segment is traversed, and the sum is the path total length. If the path total length exceeds the preset second threshold value, it is considered that the length of the trajectory segment is long enough to have the possibility of forming a spiral, and thus it is determined to be a preliminary spiral trajectory.

[0022] For example, although the trajectory drawn by the user on the screen is overall biased towards a circular arc, only when the total length of the circular arc turns exceeds the set reference value, it is considered to be a true preliminary spiral trajectory.

[0023] The number of swings of the preliminary spiral trajectory relative to the connecting line of the start point and each end point is calculated, and when the number of traversals of the trajectory on both sides of the connecting line exceeds the preset number, it is confirmed to be an effective spiral trajectory. In an embodiment, for the trajectory segment that has been determined to be a preliminary spiral trajectory, it can be further analyzed whether the trajectory swings back and forth on both sides of the reference straight line connecting the start point and the end point. The specific method is to observe the positions of the trajectory points relative to the reference straight line one by one, and record the distribution of the trajectory points on one side and the other side of the straight line. When it is found that the number of traversals of the trajectory points on both sides exceeds the set threshold value, it indicates that the trajectory has obvious reciprocating swing on both sides of the reference straight line, and such trajectory can be confirmed to be an effective spiral trajectory.

[0024] For example, if the mouse trajectory alternately crosses the reference line from left to right and from right to left for several times, it means that the trajectory indeed presents a spiral feature.

[0025] It should be noted that, in order to avoid false statistics caused by slight jitter, a certain offset range can be added when judging whether the trajectory crosses the straight line, and only when the trajectory obviously exceeds the range can it be counted as a crossing.

[0026] The center point, radius and duration of the effective spiral trajectory are extracted and packaged with the coordinate sequence as spiral trajectory data.

[0027] In an embodiment, for the trajectory segment that has been confirmed as an effective spiral trajectory, key feature information such as the center point position of the spiral, the approximate radius of the spiral, and the duration of the spiral can be extracted. Specifically, the approximate center area around the trajectory is first calculated according to the overall distribution of the trajectory, and then the average distance from the trajectory points to the center area is estimated as the spiral radius, and the spiral duration is determined by recording the start and end times of the trajectory. Finally, these features are packaged together with the original trajectory coordinate data to form complete spiral trajectory data.

[0028] For example, when a user draws a circle on the screen, the center of the circle can be recorded to be approximately located in the middle of the screen, the radius is close to half of the user's mouse movement range, and the duration is two seconds. These information are output together with the trajectory coordinates as complete spiral trajectory data. Further, the trajectory clustering module includes: The page is divided into a virtual grid, and the size of the grid is dynamically determined according to the minimum distance of the page elements; In an embodiment, the page is divided into a virtual grid, and the size of the grid is dynamically determined according to the minimum distance of the interactive elements on the page, that is, when the page elements are dense, the grid is refined to distinguish adjacent elements, and when the element distance is large, a coarser grid is used to reduce the calculation amount.

[0029] For example, if the shortest distance between buttons or product cards on the page is small, the virtual grid will be correspondingly reduced to ensure that each grid contains as few elements or parts of elements as possible.

[0030] Each spiral trajectory is mapped on the virtual grid to record the grid sequence passed by the trajectory, and the trajectory density value of each grid is calculated, wherein the density value is equal to the number of trajectory segments passing through the grid; In an embodiment, each of the spiral trajectories is mapped onto a virtual grid and the sequence of grids that the trajectory passes through is recorded, while the number of trajectory segments that pass through each grid is counted to obtain the trajectory density value of the grid; here, "passing through" is counted once for each time the trajectory appears within the grid, and the density value only records the number of trajectory segments to reflect the strength of the area of interest.

[0031] For example, if the spiral trajectories of multiple users pass through the same grid at the same position in the page, the density of the grid will be higher, facilitating subsequent clustering judgment.

[0032] The spatial position superposition is performed on the spiral trajectories after grid mapping to obtain the spatial coincidence degree; In an embodiment, each of the spiral trajectories is mapped onto a virtual grid and the sequence of grids that the trajectory passes through is recorded, while the number of trajectory segments that pass through each grid is counted to obtain the trajectory density value of the grid; here, "passing through" is counted once for each time the trajectory appears within the grid, and the density value only records the number of trajectory segments to reflect the strength of the area of interest.

[0033] For example, if the spiral trajectories of multiple users pass through the same grid at the same position in the page, the density of the grid will be higher, facilitating subsequent clustering judgment.

[0034] The time pattern superposition is performed on the spiral trajectories after grid mapping to obtain the time sequence similarity; In an embodiment, the spatial position superposition is performed on the spiral trajectories after grid mapping to obtain the spatial coincidence degree, and the method is to extract the segments of the trajectories within each grid and compare the consistency of the entry points, exit points and overall directions of these segments within the grid; if the trajectory segments of different users are highly consistent in position and direction, it is considered that these segments have high spatial coincidence degree, thereby improving the spatial coincidence degree index of the grid.

[0035] For example, if the trajectory segments of two users in the same grid enter from the upper left corner of the grid and exit from the lower right corner, and the paths are approximately coincident, the spatial coincidence degree of the two segments is higher.

[0036] The product of the spatial coincidence degree and the time sequence similarity is taken as the trajectory overlap degree, and when the trajectory overlap degree exceeds a preset overlap threshold, the corresponding trajectories are marked as same-cluster trajectories; In an embodiment, the spatial coincidence degree of each grid is combined with its time sequence similarity (conceptually taking the product of the two or other synthetic measures) to obtain the trajectory overlap degree, which reflects the strength of the trajectory pairs that both coincide in space and appear synchronously in time at the same position; when the trajectory overlap degree is significantly higher than the surrounding level or exceeds a previously set judgment standard, the corresponding trajectories are marked as belonging to the same cluster.

[0037] For example, if a group of trajectories not only coincide in the same grid, but also all appear in the same time period after the page is loaded, their overlap degree will be high, and they will be classified as the same cluster trajectory.

[0038] The density peak search is performed on the same cluster trajectory, starting from the grid with the highest trajectory density value, and expanding outward to search adjacent density grids until the trajectory density value is lower than half of the average trajectory density value, forming a connected trajectory cluster. In an embodiment, the density peak search is performed on the trajectories marked as the same cluster to form a connected trajectory cluster, by first taking the grid with the highest trajectory density in the cluster as the starting point, and expanding layer by layer to the adjacent grids around it, only when the trajectory density of the adjacent grid is still significantly higher than the surrounding environment or comparable to the starting point density, it is included in the cluster; when the density of the extended grid decreases significantly and is lower than the average level of the surrounding environment, the expansion stops, thus obtaining a connected trajectory cluster that gradually decays outward from the high-density area.

[0039] For example, a certain central grid is repeatedly passed through by multiple trajectories, and the surrounding several annular zones also have a decreasing density distribution, so these annular zones will be continuously included until the density becomes sparse.

[0040] The geometric center of the trajectory cluster is calculated based on the spiral trajectory data as the cognitive hot zone.

[0041] In an embodiment, the geometric center of the trajectory cluster is calculated based on the obtained trajectory cluster and output as the cognitive hot zone, specifically, the center positions of each spiral trajectory in the cluster are summarized and weighted according to the significance of each trajectory in the cluster (such as the attention intensity represented by the spiral duration or spiral range), and finally a geometric center point representing the overall attention position of the cluster is obtained, and the center is output as the cognitive hot zone information together with the cluster boundary, typical trajectory, etc.

[0042] For example, if multiple trajectories in a cluster repeatedly revolve around a certain point in the middle of the page and have a long duration, the calculated geometric center will be located at the middle position and be identified as the cognitive hot zone.

[0043] Referring to Figure 2 , the trajectory clustering module superimposes and analyzes the spiral trajectories of multiple users in a preset time period according to the page position. As shown in Figure 2 , the system records the trajectory curves of 6 different users on a virtual grid of 20x15, represented by red, blue, green, orange, purple, and cyan solid lines respectively. Through spatial position superposition and time pattern superposition calculation, the system identifies multiple trajectory density areas: low-density areas are displayed with light blue transparent overlay, medium-density areas are displayed with yellow transparent overlay, and high-density areas are displayed with red transparent overlay. When the trajectory overlap degree exceeds the preset overlap threshold, the system automatically forms a trajectory cluster, as shown in Figure 2The area enclosed by the middle black dashed line circle is shown. Based on the spiral trajectory data, the system calculates the geometric center of the trajectory cluster (marked by the middle black solid circle) as the core position of the cognitive hot zone. The legend part details the corresponding trajectory color, density level division, and technical parameter settings. Figure 2

[0044] Further, the spiral trajectory after grid mapping is performed spatial position superposition, and the spatial coincidence degree is specifically: The part of the spiral trajectory after grid mapping in the grid is extracted as a trajectory segment, and the grid entry point and grid exit point coordinates of the segment are recorded; In an embodiment, the spiral trajectories mapped into the same grid are divided at the grid boundary, and the continuous coordinate sequence from the entry into the grid to the exit from the grid of each segment is extracted as a trajectory segment, and the grid entry point and grid exit point coordinates and necessary intermediate positions of the segment are recorded to completely describe the segment direction.

[0045] For example, if a mouse trajectory enters from the upper left of the grid and exits from the lower right, the grid entry point of the segment is the upper left intersection point, and the grid exit point is the lower right intersection point, and multiple sampling points inside the segment are saved together for subsequent comparison.

[0046] It should be noted that multiple entries and exits of the same user in the same grid should be recorded as independent segments in order to distinguish repeated behavior from single long stay.

[0047] The spatial deviation degree between trajectory segments of different users is calculated, wherein the spatial deviation degree is specifically the average distance between corresponding points of two segments; In an embodiment, the spatial deviation degree between trajectory segments of different users is calculated in pairs, which is to correspond two segments according to their advancing order (for example, corresponding point sequences are obtained by relative progress or equidistant resampling), measure the distance between corresponding points point by point, and take the average value as the spatial deviation degree of two segments, so as to reflect the overall deviation degree of two path segments in the grid.

[0048] For example, if two segments enter from the left side of the grid and pass through similar curves, and the distance between each set of corresponding points is generally small, then the spatial deviation degree of the two segments is also small.

[0049] It should be noted that when the length difference of the segments is large, the progress resampling method can be used to make the corresponding relationship comparable, so as to avoid the deviation caused by the difference in the number of points.

[0050] When the spatial deviation degree is less than a preset proportion of the diagonal length of the grid, it is determined that the two trajectory segments are spatially coincident; ​In an embodiment, the spatial deviation is combined with the grid scale to determine spatial coincidence, i.e. when the spatial deviation of two segments is less than a certain proportion of the grid diagonal length, it is determined that they are spatially coincident, indicating that the two trajectories point to the same location within the grid with a high possibility of having the same focus hotspot.

[0051] For example, when two segments are almost coincident within the grid range, and the average distance of the corresponding points is much smaller than the grid size, it is considered that they are spatially coincident.

[0052] The number of spatially coincident trajectory segment pairs in each grid is counted, and the number is divided by the total number of all possible trajectory pairs in the grid to obtain the spatial coincidence degree of the grid.

[0053] In an embodiment, the number of trajectory segment pairs determined to be spatially coincident in each grid is counted, and the number is divided by the total number of all possible trajectory segment pairs in the grid to obtain the spatial coincidence degree of the grid, which reflects the general degree of spatial coincidence of trajectories in the grid.

[0054] For example, if there are multiple segments from different users in a grid and most of the segments coincide with each other, the spatial coincidence degree of the grid will be close to a high value, indicating that it is a spatial focus hotspot.

[0055] Further, the time pattern superposition is performed on the spiral trajectories mapped to the grid to obtain the time sequence similarity, which is specifically: The time stamp of each spiral trajectory mapped to the grid is extracted, and the time stamp is converted into a relative time with respect to the page loading time; In an embodiment, the number of trajectory segment pairs determined to be spatially coincident in each grid is counted, and the number is divided by the total number of all possible trajectory segment pairs in the grid to obtain the spatial coincidence degree of the grid, which reflects the general degree of spatial coincidence of trajectories in the grid.

[0056] For example, if there are multiple segments from different users in a grid and most of the segments coincide with each other, the spatial coincidence degree of the grid will be close to a high value, indicating that it is a spatial focus hotspot.

[0057] The relative time is divided into equal length time slices, the number of trajectories appearing in each time slice is counted to form a time distribution vector; In an embodiment, the time stamp information of each spiral trajectory segment in the grid is obtained and uniformly converted into a relative time with respect to the page loading or page exposure, so as to eliminate the difference of absolute time of different sessions, and make the time information of all trajectories comparable in the same time scale.

[0058] For example, if the page is loaded at time zero, then the time of subsequent trajectories is expressed in seconds from this zero point, e.g. a segment occurring two seconds after loading is recorded as two seconds of relative time.

[0059] It should be noted that the use of relative time can more accurately compare different behavior stages of users in the life cycle of the page (e.g. the difference between just after loading and staying).

[0060] In an embodiment, the time distribution vectors of different trajectories in the same grid are compared for similarity, and the time sequence similarity is determined by measuring the consistency of the distribution patterns of the two vectors (i.e. whether their distribution peaks and troughs are synchronized); if the high-frequency occurrence segments of two trajectories are consistent in relative time, it is considered that they have high time sequence similarity.

[0061] For example, if multiple trajectories all occur in a short time after page loading, the time distribution vectors of these trajectories are highly consistent, and the time sequence similarity is large.

[0062] The cosine similarity of all trajectory pairs is averaged to obtain the time sequence similarity of the grid.

[0063] In an embodiment, the time sequence similarity of all trajectory pairs in the same grid is averaged as the overall time sequence similarity of the grid, representing the consistency of the trajectories in the time dimension at this position; the time sequence similarity can be used together with the spatial coincidence degree to measure the aggregation of trajectories in space-time.

[0064] For example, if most trajectory pairs in the grid show the feature of synchronous occurrence in time, the time sequence similarity of the grid will tend to a higher value.

[0065] It should be noted that when the number of trajectories in the grid is small, the reliability of the average value will decrease, and should be evaluated in combination with the sample size or confidence index.

[0066] Further, the method for calculating the geometric center of the trajectory cluster based on the spiral trajectory data as the cognitive hot area comprises: extracting the spiral center point, spiral radius and spiral duration of the spiral trajectory data in the trajectory cluster, and calculating the cognitive intensity value of each trajectory, wherein the cognitive intensity value is equal to the product of the spiral duration and the spiral radius; In an embodiment, the key information of each spiral trajectory is extracted from the trajectory cluster, including the center position of the spiral, the approximate radius of the spiral, and the duration of the spiral, and the cognitive intensity of each trajectory is determined accordingly. Specifically, the "duration" and "spiral range" can be combined as an index for measuring the importance of the trajectory in user cognition, i.e. the longer the duration and the larger the spiral range, the higher the cognitive intensity of the trajectory.

[0067] For example, if a trajectory stays at the same location for a long time and covers a large range, the cognitive strength of this trajectory is significantly higher than that of a short and small range.

[0068] The initial geometric center is obtained by weighted average of all the cyclone centers with the cognitive strength as the weight; In an embodiment, the initial geometric center is obtained by weighted average of all the cyclone centers in the cluster with the cognitive strength as the weight. That is, each cyclone center contributes different weight to the overall center position according to its cognitive strength, and the trajectory with higher strength has more influence on the initial geometric center, so that the final center can better reflect the most focused area in the cluster.

[0069] For example, when most trajectories in the cluster are gathered at the upper part of the page, but one of them has significantly longer duration and larger range, the initial center obtained by weighted average is closer to the position around which this trajectory revolves.

[0070] The distances from the initial geometric center to each cyclone center are calculated, and when more than a preset number of distances are greater than twice the corresponding cyclone radius, the outlying center is determined, and the center splitting is performed to separate a preset proportion of the cyclone trajectories farthest from the initial geometric center, and the weighted geometric centers of the main cluster and the separated cluster are calculated to obtain the cognitive hot area of the double center.

[0071] In an embodiment, the distances between each cyclone center and the initial geometric center in the cluster are evaluated based on the initial geometric center to determine whether there is an outlying center and to perform center splitting accordingly. Specifically, the distance from each cyclone center to the initial center is calculated and compared with the cyclone range of the cyclone itself, and when a certain number of cyclone centers are found to be significantly away from the initial center and their discrete distances are much larger than their own cyclone ranges, these outlying centers can be determined and separated from the original cluster to form one or more separated clusters; then the weighted geometric centers of the main cluster and the separated cluster are recalculated respectively, thereby obtaining the cognitive hot area representation of the double center (or multiple center).

[0072] For example, if most trajectories revolve around a point on the left side of the page, but a small number of trajectories are concentrated in the lower right corner and the distance from the left center is much larger than their respective cyclone ranges, the trajectories in the lower right corner can be separated to form a second hot area, and the center positions of the two hot areas are given respectively.

[0073] It should be noted that the judgment of center splitting should be based on the overall situation of the dispersion degree in the cluster (such as the proportion of discrete points and the relative size of the cyclone range), and the same confidence evaluation can be applied to the separated cluster after splitting to determine whether to finally retain the sub-cluster.

[0074] Further, the cognitive optimization module comprises: receiving the cognitive hot zone and generating a user ID list of the hot zone, counting the number of independent users and the number of repeated visits of the users; In an embodiment, the system receives the cognitive hot zone information output from the trajectory clustering module, and generates a user ID list of the hot zone based on the trajectory data in the hot zone, and further counts the number of independent users in the list and the repeated visits of the users.

[0075] For example, if multiple visits from the same user in the hot zone are identified as several trajectories, the system will merge these entries into the same user and record the number of repeated visits, thereby distinguishing between new visitors and returning visitors.

[0076] calculating a user dispersion index, wherein when the user dispersion is greater than a preset dispersion threshold and the number of independent users exceeds a preset number of users, the cognitive hot zone is determined to be a group cognitive obstacle point; In an embodiment, the system calculates a user dispersion index based on the distribution of users in the hot zone, which reflects whether the user's focus points in the hot zone are concentrated in a few locations or dispersed in multiple locations; when the hot zone has a large number of independent users and a high dispersion, the system will determine the hot zone to be a group cognitive obstacle point.

[0077] For example, if the users in the hot zone are not all concentrated in the same small range, but repeatedly revolve around multiple page elements, and these behaviors come from a large number of different users, it indicates that there may be group cognitive confusion at this place, which should be identified as an obstacle point.

[0078] It should be noted that the dispersion judgment should be considered together with the number of independent users - the dispersion conclusion is not reliable when the sample is too small, so the group problem should be confirmed when the sample size is sufficient.

[0079] determining all the interactive elements covered by the group cognitive obstacle point through coordinate matching, and extracting the two elements closest to the center of the hot zone as a core element pair; In an embodiment, for the selected core element pair, the system calculates its multi-dimensional similarity, mainly including text similarity (comparing the semantic or keyword overlap degree of element title and description), visual similarity (comparing appearance characteristics such as size, color matching, position relationship, icon and visual weight), and functional similarity (comparing click behavior, jump destination or interaction response type) to comprehensively evaluate the distinguishability of the two elements in user cognition.

[0080] For example, if the titles of the two elements use similar words, the button colors and sizes are similar, and similar information is popped up after clicking, the three-dimensional similarity will be high, indicating that the user may have difficulty distinguishing between the two.

[0081] The multi-dimension similarity of the pair of core elements is calculated, including text similarity, visual similarity and function similarity; When the multi-dimension similarity exceeds a preset similarity threshold, difference keywords are extracted to generate an optimization instruction.

[0082] In an embodiment, when the multi-dimension similarity indicates that the pair of core elements are highly similar in cognition and may cause confusion, the system further extracts difference keywords between the two elements and generates a standardized optimization instruction, which includes target element identification, representative difference keywords and a recommended display or interaction adjustment scheme (such as enhancing visual differentiation, supplementing contrast information or strengthening function labels), so that the conflict processing module implements A / B type optimization according to the instruction and monitors the effect.

[0083] For example, if the titles of two product cards are similar but the specifications are different, the system can extract "specifications / size" as the difference keyword and recommend highlighting the specification contrast or adding a function icon in the display.

[0084] In particular, the multi-dimension similarity calculation is as follows: The multi-dimension similarity of the pair of core elements is calculated, including text similarity, visual similarity and function similarity, wherein the text similarity is obtained by calculating the proportion of repeated characters in the element title and description text, the visual similarity is obtained by comparing the size, color and position relationship of the elements, and the function similarity is obtained by analyzing the click response type of the elements.

[0085] Referring to Figure 4 , the cognitive optimization module receives cognitive hot zone information and performs obstacle point analysis. As shown in Figure 4 , the system identifies two main cognitive hot zones in the simulated e-commerce page: hot zone 1 covers the product comparison area and is displayed in red semi-transparency, with data annotation showing that 23 independent users and 156 repeated visits have a dispersion of 0.78, and the obstacle level is high; hot zone 2 covers the search area and is displayed in yellow semi-transparency, with 45 independent users and 89 repeated visits, a dispersion of 0.52, and an obstacle level of medium. The system determines the interactive elements covered by the cognitive hot zones through coordinate matching and marks them with a blue dashed line frame. When the user dispersion is greater than a preset dispersion threshold (0.6) and the number of independent users exceeds a preset number of users (20), the system determines that the cognitive hot zone is a group cognitive obstacle point and marks it with a red triangular warning icon.

[0086] In particular, the extraction of difference keywords and the generation of optimization instructions when the multi-dimension similarity exceeds a preset similarity threshold are as follows: The DOM node ID of the pair of core elements is obtained as the target element identification; In an embodiment, the system first matches the coordinate mapping with the page rendering structure, identifies all the interactive elements within the hot area range, and reads the DOM node IDs of these elements as the unique identifiers of the target elements for subsequent operations.

[0087] For example, when the hot area covers the product picture and the purchase button, the system records the corresponding DOM node IDs as the target identifiers.

[0088] Compare the attributes of the core element pairs to obtain the difference attributes, and divide the difference attributes into three categories: numerical type, descriptive type, and identification type; In an embodiment, the system compares the attributes of the core element pairs one by one to find the differences, and divides the difference attributes into three categories according to their properties: numerical type (such as price, quantity, specification, and other quantifiable information), descriptive type (such as title, description text, or function description, and other semantic text), and identification type (such as color, icon, style, or visual identifier).

[0089] For example, if the prices of the two elements are different, it is of the numerical type; if the description text is different, it is of the descriptive type; and if the button color or icon is similar, it is of the identification type.

[0090] Perform dynamic verification on each difference attribute, add temporary markers at the attribute position, and measure the change in user's orbit trajectory before and after the markers are added; In an embodiment, dynamic verification is performed on each difference attribute by adding non-intrusive temporary markers or enhancing the display (such as highlighting the price range, adding contrast labels, or temporarily bolding / bordering) at the attribute position for a short time, and comparing the orbit trajectory indicators of the target area before and after the markers are added to evaluate the actual impact of the attribute.

[0091] For example, a "price range comparison" label can be temporarily added at the price, and then the number and duration of the orbit trajectory around the element can be observed to determine whether they have decreased.

[0092] Select the difference attribute that causes the greatest reduction in the orbit trajectory as the core difference based on the change in the user's orbit trajectory, and extract the difference keyword; In an embodiment, the system compares the contributions of each difference attribute to the reduction of the orbit trajectory based on the results of dynamic verification, selects the difference attribute that causes the most improvement in the orbit trajectory as the core difference, and extracts the keyword that can clearly express the difference point from the attribute as the difference keyword for subsequent instruction generation.

[0093] For example, if the orbit significantly decreases after temporary supplementation of "specification display", then "specification comparison" is extracted as the difference keyword.

[0094] Determine display strategy according to difference type, add range comparison for numerical type, add function icon for description type, and strengthen visual distinction for identification type; In an embodiment, determine specific display strategy according to difference type to guide optimization landing: for numerical difference, prefer to use range or numerical comparison display to highlight the difference; for description difference, prefer to use function icon or supplement short comparison text to clarify function or use; for identification difference, prefer to use visual distinction means (such as enhanced contrast, border, size or obvious icon) to improve distinguishability.

[0095] For example, price difference can display "from X-Y", function difference can add icon and attach short description, and visual similarity can be distinguished by color contrast and border.

[0096] Encapsulate target element identification, difference keywords and display strategy as optimization instructions.

[0097] In an embodiment, the system combines target element identification, determined difference keywords and recommended display strategy to form standardized optimization instructions for conflict processing module or front-end iteration execution; the optimization instructions can also include experimental suggestions (such as shunt display strategy, monitor which revolving indicators, evaluation period and rollback conditions) to close loop verification and solidification.

[0098] For example, the final instruction will explain "for the element with DOM X, highlight the'specification comparison' keyword and add a function icon, and within two weeks, if the number of rotations changes significantly, do not roll back".

[0099] It should be noted that the optimization instructions should be issued in an executable and easy-to-rollback form, and the actual effect should be verified in a small range before deployment in production environment.

[0100] Further, the conflict processing module includes: Receive optimization instructions and parse target element identification and difference keywords; In an embodiment, the conflict processing module receives optimization instructions issued from the cognitive optimization module and parses the target element identification and difference keywords in the instructions, and then locates the corresponding element instance in the current page context for subsequent operation.

[0101] For example, if the target identification in the instruction is the DOM node id of a purchase button and the difference keyword is "specification comparison", the system will locate the button element in the page and take the element as the target of optimization operation.

[0102] Statistical revolving trajectory time distribution relative to page loading time, judge cognitive conflict type according to the time period of trajectory concentration, divided into immediate cognitive conflict and delayed cognitive conflict; In an embodiment, the system analyzes and calculates the time distribution of the spiral trajectories in the target area relative to the time when the page is loaded, and determines the type of cognitive conflict by mapping the time of trajectory occurrence to the relative time axis of the page life cycle and observing the time period when the trajectories are densely present.

[0103] For example, if most of the spiral trajectories are concentrated in a short time after the page is loaded, it can be determined as an immediate cognitive conflict; if the trajectories are repeatedly present between elements after the user has browsed for a period of time, it can be determined as a delayed cognitive conflict.

[0104] For immediate cognitive conflicts, visual difference enhancement is performed, including increasing color contrast between elements, increasing element spacing, and adding distinctive borders; In an embodiment, for cases determined as immediate cognitive conflicts, the system preferentially performs visual difference enhancement strategies to improve the distinguishability between elements through a series of non-destructive display adjustments, such as enhancing color contrast between elements, appropriately increasing element spacing or inner and outer margins, adding distinctive borders or outlines to key elements, adjusting font weight or size to highlight differences, etc.

[0105] For example, if two buttons are similar in color and size, causing confusion, the color of one of the buttons can be adjusted to a higher contrast and a distinct border can be added to quickly distinguish between them.

[0106] For delayed cognitive conflicts, information supplement strategies are performed, and when the mouse is detected moving back and forth between elements, contrast information is dynamically displayed; In an embodiment, for cases determined as delayed cognitive conflicts, the system uses information supplement strategies to dynamically display contrast information or supplementary explanations when the user's mouse is detected moving back and forth between two or more elements, thereby helping the user quickly distinguish the differences. The implementation can be to temporarily pop up a comparison panel, display parallel key information prompts between elements, or hover a short text / icon over the element to emphasize the differences.

[0107] For example, when the user moves the mouse back and forth between product specifications and purchase options, the page can pop up a small specification comparison card between the two, intuitively displaying the differences.

[0108] Monitor new spiral trajectories in the target area after optimization, and calculate the trajectory change rate before and after optimization; In an embodiment, after implementing optimization measures, the system will continuously monitor new spiral trajectories in the target area and compare them with the trajectory indicators before optimization to calculate the trajectory change rate before and after optimization as a key indicator of effectiveness; this change rate can be expressed based on the relative changes in the number, average duration, or density of spiral trajectories.

[0109] For example, if the average number of rotations per session of the target area before optimization is four, and it is reduced to one after optimization, the number of rotations is reduced by 75%, which is the change rate.

[0110] The optimization strategy is solidified when the trajectory change rate is higher than the preset improvement threshold, otherwise, the fallback is executed and the alternative strategy is tried; In an embodiment, the system makes a processing decision on the optimization strategy according to the calculated trajectory change rate: if the change rate exceeds the established improvement standard, the optimization strategy is solidified as a long-term display (for example, the style change is written into the production style sheet and the delivery is expanded), otherwise, the fallback is executed and the alternative strategy prepared in advance is tried or the next round of dynamic verification is entered.

[0111] For example, when the visual difference enhancement significantly reduces the rotation trajectory and the user conversion is not negatively affected, the system merges the change into the normal style; if there is no improvement, the change is withdrawn and another information supplement scheme is implemented for retest The trajectory change rate and optimization parameters are fed back to the cognitive analysis module to adjust the threshold of cognitive barrier determination, forming a closed-loop optimization.

[0112] In an embodiment, the system feeds back the trajectory change rate, sample size, optimization parameters, and confidence to the cognitive analysis module, which is used to adjust the threshold and rules of cognitive barrier determination and improve the identification of hot areas and optimization recommendations in the next round, thereby forming a closed-loop optimization.

[0113] For example, if multiple optimizations show that a certain type of visual enhancement is generally effective, the analysis module can reduce the sensitivity to this type of problem in subsequent hot area determination or directly preferentially recommend the corresponding strategy.

[0114] Referring to Figure 3 , the conflict processing module implements the optimization strategy and verifies the effect. For example Figure 3 As shown on the left side "before optimization", the original interface uses the same blue style for the two buttons, and the user produces a clear rotation trajectory, with a rotation number of 4.2 times per session and a duration of 8.5 seconds. After detecting the real-time cognitive conflict pattern, the system executes the visual difference enhancement strategy. For example Figure 3 As shown on the right side "after optimization", the main button remains blue and increases the thick border to strengthen, and the secondary button is changed to orange to enhance the distinction. After the optimization is implemented, the user trajectory becomes a clear green straight line path, with a rotation number of 1.1 times per session and a duration of 2.3 seconds. The system calculates that the trajectory change rate is -73%, which exceeds the preset improvement threshold, so the optimization strategy is solidified.

[0115] Therefore, the embodiments should be regarded, at any point, as being exemplary and not limiting, the scope of the application being defined by the appended claims and not by the above description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.

[0116] The foregoing is considered as illustrative only of the principles of the application. Numerous modifications and changes will readily occur to those skilled in the art, and the generic principles defined herein can be applied to other embodiments without departing from the spirit or scope of the application. Therefore, the application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An internet e-commerce platform system, characterized in that, Includes the following modules: The trajectory acquisition module is used to periodically collect the user's mouse coordinates. When it detects that the distance between the start and end points of the mouse trajectory within a preset range is less than a first threshold while the total path length is greater than a second threshold, it is determined to be a spiral trajectory and the spiral trajectory data is output. The trajectory clustering module is used to superimpose the circumference trajectories of multiple users within a preset time period according to the page position. When the trajectory overlap exceeds the preset overlap threshold, a trajectory cluster is formed. The geometric center of the trajectory cluster is calculated based on the circumference trajectory data as the cognitive hot zone. The cognitive optimization module is used to determine the cognitive barrier points of a group based on the number of users from the cognitive hot zone. Calculate the similarity of page element pairs within the group's cognitive impairment points, and extract the difference keywords to generate optimization instructions when the similarity exceeds the preset similarity threshold; The conflict handling module receives optimization instructions, executes optimization strategies based on cognitive conflict types according to the time distribution characteristics of the spiral trajectory, and feeds back the trajectory change rate before and after optimization to the cognitive analysis module to form a closed-loop optimization.

2. The internet e-commerce management platform system according to claim 1, characterized in that, The trajectory acquisition module includes: The user's mouse coordinates are coarsely sampled at the first sampling frequency, and the displacement vector between consecutive coordinate points is recorded. When the sum of the directional angles of three consecutive displacement vectors exceeds 180 degrees, the fine sampling mode is triggered, and the sampling frequency is increased to the second sampling frequency, which is greater than the first sampling frequency. In the fine sampling mode, the collected coordinate sequence is divided into a sliding window of length N, the start and end points of the trajectory within each window are determined, and the straight-line distance between the start and end points is calculated. When the straight-line distance between the start and end points is less than the first threshold, the path length verification is initiated, and the trajectory is determined to be a loop trajectory, and the loop trajectory data is output.

3. The internet e-commerce management platform system according to claim 2, characterized in that, When the straight-line distance between the start and end points is less than the first threshold, the path length verification is initiated, and the trajectory is determined to be a loop trajectory. The loop trajectory data is output as follows: For trajectory segments that meet the first threshold condition, calculate their actual total path length. The actual total path length is the sum of the distances between all adjacent coordinate points. When the total path length is greater than the second threshold, it is determined to be a preliminary spiral trajectory. Calculate the number of swings of the preliminary loop trajectory relative to the line connecting the starting point and the end point. When the number of times the trajectory crosses the line on both sides of the line exceeds the preset number, it is confirmed as a valid loop trajectory. Extract the rotation center point, rotation radius, and rotation duration of the effective rotation trajectory, and encapsulate them with the coordinate sequence to form rotation trajectory data.

4. The internet e-commerce management platform system according to claim 3, characterized in that, The trajectory clustering module includes: The page is divided into a virtual grid, and the grid size is dynamically determined based on the minimum spacing between page elements; Each spiral trajectory is mapped onto a virtual grid to record the grid sequence through which the trajectory passes, and the trajectory density value of each grid is calculated, where the density value is equal to the number of trajectory segments passing through that grid. Spatial position superposition is performed on the spiral trajectory after mesh mapping to obtain spatial overlap. Temporal pattern superposition is performed on the spiral trajectory after grid mapping to obtain temporal similarity; The product of spatial overlap and temporal similarity is used as the trajectory overlap. When the trajectory overlap exceeds the preset overlap threshold, the corresponding trajectory is marked as a cluster trajectory. Perform density peak search on trajectories in the same cluster, starting from the grid with the highest trajectory density value and expanding outward to search adjacent density grids until the trajectory density value is less than half of the average trajectory density value, thus forming a connected trajectory cluster; The geometric center of the trajectory cluster is calculated based on the spiral trajectory data as a cognitive hot zone.

5. The internet e-commerce management platform system according to claim 4, characterized in that, The spatial position superposition of the mesh-mapped spiral trajectory to obtain the spatial overlap is specifically as follows: Extract the portion of the spiral trajectory within the grid after grid mapping as a trajectory segment, and record the coordinates of the entry and exit points of the segment; Calculate the spatial deviation between different user trajectory segments, where the spatial deviation is specifically the average distance between corresponding points of two segments; When the spatial deviation is less than a preset proportion of the grid diagonal length, the two trajectory segments are determined to be spatially overlapping. Count the number of spatially overlapping trajectory segments within each grid, and divide this number by the total number of all possible trajectory pairs within that grid to obtain the spatial overlap degree of that grid.

6. The internet e-commerce management platform system according to claim 5, characterized in that, The process of performing time pattern superposition on the mesh-mapped spiral trajectory to obtain temporal similarity is as follows: Extract the timestamp of each spiral trajectory after grid mapping and convert the timestamp into a relative time relative to the page load time; The relative time is divided into equal time slices, and the number of trajectories appearing in each time slice is counted to form a time distribution vector. For different trajectories within the same grid, calculate the cosine similarity of their time distribution vectors; The temporal similarity of the grid is obtained by averaging the cosine similarities of all trajectory pairs.

7. The internet e-commerce management platform system according to claim 6, characterized in that, The geometric center of the trajectory cluster calculated based on the spiral trajectory data, as the cognitive hot zone, includes: Extract the center point of rotation, radius of rotation, and duration of rotation from the rotation trajectory data in the trajectory cluster, and calculate the cognitive intensity value for each trajectory, where the cognitive intensity value is equal to the product of the duration of rotation and the radius of rotation; Using cognitive intensity values ​​as weights, a weighted average is taken from all the centroids to obtain the initial geometric center. Calculate the distance from the initial geometric center to each spiral center point. When the distance exceeds a preset number and is greater than twice the corresponding spiral radius, it is determined to be an outlier center. Perform center splitting to separate the spiral trajectories that are farthest from the initial geometric center by a preset proportion. Calculate the weighted geometric centers of the main cluster and the separated clusters respectively to obtain the cognitive hot zone of the dual centers.

8. The internet e-commerce management platform system according to claim 7, characterized in that, The cognitive optimization module includes: Receive the cognitive hotspot and generate a list of user IDs for that hotspot, and count the number of unique users and the number of times a user visits repeatedly; Calculate the user dispersion index. When the user dispersion is greater than the preset dispersion threshold and the number of independent users exceeds the preset number of users, the cognitive hot zone is determined to be a group cognitive barrier point. By using coordinate matching, all interactive elements covered by the collective cognitive impairment points are determined, and the two elements closest to the center of the hot zone are extracted as the core element pair. Calculate the multi-dimensional similarity of core element pairs, including textual similarity, visual similarity, and functional similarity; When the multidimensional similarity exceeds the preset similarity threshold, the difference keywords are extracted to generate optimization instructions.

9. The internet e-commerce management platform system according to claim 8, characterized in that, The specific steps for extracting difference keywords and generating optimization instructions when the multidimensional similarity exceeds a preset similarity threshold are as follows: Obtain the DOM node ID of the core element pair as the target element identifier; By comparing the attributes of the core element pairs, we obtain the difference attributes and categorize them into three types: numerical, descriptive, and identifier. Perform dynamic validation for each difference attribute, add a temporary marker at the attribute location, and measure the change in the user's circumduction trajectory before and after the marker is added; Based on the change in the user's circling trajectory, the attribute that causes the largest reduction in the circling trajectory is selected as the core difference, and difference keywords are extracted. The display strategy is determined based on the type of difference: numerical differences are compared with ranges, descriptive differences are added with function icons, and symbolic differences are enhanced with visual distinctions. Encapsulate target element identifiers, differentiating keywords, and display strategies into optimization instructions.

10. The internet e-commerce management platform system according to claim 9, characterized in that, The conflict resolution module includes: Receive optimization instructions and parse target element identifiers and difference keywords; The time distribution of the circular trajectory relative to the page loading time is statistically analyzed, and the type of cognitive conflict is determined based on the time period in which the trajectory clusters appear, which is divided into immediate cognitive conflict and delayed cognitive conflict. Visual differentiation enhancements are applied to address immediate cognitive conflicts, including increasing color contrast between elements, increasing spacing between elements, and adding distinctive borders. To address delayed cognitive conflict, an information supplementation strategy is implemented, dynamically displaying comparative information when the mouse is detected moving back and forth between elements. After optimization, monitor the new circling trajectory of the target area and calculate the trajectory change rate before and after optimization; When the trajectory change rate is higher than the preset improvement threshold, the optimization strategy is solidified; otherwise, a rollback is executed and alternative strategies are tried. The trajectory change rate and optimization parameters are fed back to the cognitive analysis module to adjust the obstacle judgment threshold, forming a closed-loop optimization.

Images