Space-time narrative and immersive interactive display method and system of evolution course of su and tang prefecture military system

Abstract

This disclosure provides a method and system for spatiotemporal narrative and immersive interactive display of the evolution of the Sui and Tang Dynasty's Fubing system. The method includes: acquiring the user's speed and dwell time during a timeline swipe, generating user behavior trajectory data; determining the user's attention distribution pattern for key turning points in the Fubing system based on the user behavior trajectory data; adaptively attenuating and visually marking the credibility weights of historical materials corresponding to the key turning points in the Fubing system based on the attention distribution pattern; and dynamically presenting the time-authority tension relationship of multiple historical materials in the spatiotemporal narrative interface according to the confidence visualization marking results. The solution of this disclosure can solve the problem that the time-authority tension of multiple historical materials is not made explicit through interactive feedback.

Description

Technical Field

[0001] This application relates to the field of human-computer interaction technology, specifically to a spatiotemporal narrative and immersive interactive display method and system for the evolution of the Sui and Tang dynasty military system. Background Technology

[0002] In the field of Sui and Tang dynasty military system research, scholars have long faced the challenge of inherent tensions in the temporal and authoritative dimensions of multiple historical sources (such as official histories, epitaphs, Dunhuang manuscripts, and the restored clauses of the Tiansheng edict). On the one hand, different historical sources often differ in the level of detail, dating, and stance regarding the same key turning point (such as the "Jiande Reform," the normalization of recruitment during the Yifeng era, and the edict abolishing the Zhechongfu in the tenth year of the Kaiyuan era). On the other hand, traditional static display methods (such as timelines, text lists, or static maps) cannot capture the dynamic changes in researchers' or learners' attention during the exploration process. This results in a lack of adaptive adjustment mechanisms for the credibility weights of historical sources, making it difficult to explicitly present the temporal-authoritative tensions between multiple historical sources. Users also find it difficult to intuitively perceive the impact of their own interactive behaviors (such as differences in scrolling speed and dwell time) on their understanding of historical sources. While existing digital humanities systems can provide visualization of spatiotemporal narratives, they have not yet solved the problem of how to dynamically adjust the confidence weights of different types of historical sources based on user behavior trajectory data and provide visual feedback. Summary of the Invention

[0003] In view of this, the present disclosure provides a method and system for spatiotemporal narrative and immersive interactive display of the evolution of the Sui and Tang Dynasty military system, which at least partially solves the problems existing in the prior art.

[0004] The spatiotemporal narrative and immersive interactive presentation methods for the evolution of the Sui and Tang dynasty's military system include:

[0005] The system acquires the user's speed and dwell time during the scrolling process on the timeline to generate user behavior trajectory data.

[0006] Based on the user behavior trajectory data, determine the user attention distribution pattern for key turning points in the Fubing system;

[0007] Based on the attention distribution pattern, the credibility weights of historical materials corresponding to key turning points in the Fubing system are adaptively decayed and confidence-visualized.

[0008] Based on the confidence visualization results, the time-authority tension relationship of multi-source historical materials is dynamically presented in the spatiotemporal narrative interface.

[0009] In one specific implementation, determining the user's attention distribution pattern regarding key turning points in the Fubing system based on the user behavior trajectory data includes:

[0010] Obtain the sliding speed V and residence duration T of the user in different intervals during the sliding process on the timeline;

[0011] Perform weighted calculation on V and T corresponding to each event node to obtain the attention weight W = α×V + β×T, where α and β are the normalization coefficients of the sliding speed and residence duration respectively;

[0012] Construct an attention distribution vector A(i) according to the W value, where A(i) represents the attention intensity of the i-th key event;

[0013] Use the sliding window average algorithm to smooth A(i) to reduce the impact of instantaneous outliers on the analysis.

[0014] In a specific embodiment, the adaptive attenuation and confidence visualization marking of the historical data credibility weight corresponding to the key turning events of the Fubing system based on the attention distribution pattern further includes:

[0015] Set an initial historical credibility weight C0 and assign different basic weight values according to the historical data type;

[0016] Construct an attenuation factor γ = (A_max - A(i)) / A_max according to the attention distribution vector A, where A_max is the maximum attention value;

[0017] Use the formula to adjust the weight to C_i = C0×(1 - γ^β), where β is the exponential attenuation parameter;

[0018] Map the result to the visualization marking, such as color transparency or label blinking frequency, to represent the trust level.

[0019] In a specific embodiment, the adaptive attenuation and confidence visualization marking of the historical data credibility weight corresponding to the key turning events of the Fubing system based on the attention distribution pattern includes:

[0020] Generate an event importance index E according to the sliding direction and speed change of the user on the timeline, defined as E = V×dT / dt, where dT / dt is the speed change rate;

[0021] Combine the attention distribution vector A and the event importance index E, and define a confidence threshold K = A_avg + λ×E_avg, where A_avg is the mean of A and E_avg is the mean of E;

[0022] Compare the historical credibility weight C_i with the threshold K. If C_i < K, enable the confidence attenuation mechanism;

[0023] Display the key historical data content with confidence decline in the form of dynamic highlighting and pop-up prompts through the graphical interface.

[0024] In one specific implementation, the adaptive decay and confidence visualization labeling of the historical credibility weights corresponding to key turning points in the Fubing system based on the attention distribution pattern further includes:

[0025] The dynamic window sliding method is used to analyze the trend of user attention change. The attention slope ΔA=(A_current-A_prev) / Δt is defined, where Δt is the length of the time window.

[0026] If ΔA is less than the threshold η, it is judged to be in the attention stabilization stage, and the confidence weight is allowed to remain at the original level;

[0027] If ΔA is greater than η, the decay mechanism is activated, and the weight decay rate is defined as ρ=(ΔA / η)^2;

[0028] The decay mechanism is applied to the confidence level of each historical material segment, and the weight changes are presented in real-time charts in the visualization.

[0029] In one specific implementation, the step of dynamically presenting the temporal authority tension relationship of multi-source historical materials in the spatiotemporal narrative interface based on the confidence visualization mark results further includes:

[0030] Establish a timeline coordinate system, with the horizontal axis representing the time range and the vertical axis representing the authorization level;

[0031] The historical credibility weight C_i and attention weight W_i are normalized to the [0,1] interval and used as the time point and access position of each historical material;

[0032] A visual distance index is constructed using the spatial mapping formula D=w1*C_i+w2*W_i, where w1 and w2 are weighting coefficients used to distinguish the relative influence of time and attention.

[0033] Historical materials are displayed in the form of nodes or broken lines. The greater the distance, the greater the difference in time and authority. Details can be expanded through user interaction.

[0034] In one specific implementation, the step of dynamically presenting the temporal authority tension relationship of multi-source historical materials in the spatiotemporal narrative interface based on the confidence visualization mark results further includes:

[0035] Establish independent time anchors for key events and set up dynamic triggering mechanisms;

[0036] When a user stays near the anchor point, the system automatically reads the historical weights and corresponding confidence scores of the surrounding historical materials and constructs a local tension matrix M=[C_i;W_i];

[0037] Introduce the time tension index T_E=∑|C_i-C_j| and the authority tension index A_E=∑|W_i-W_j|, where j represents adjacent historical data items;

[0038] The tension matrix is ​​dynamically refreshed on the interface and presented in the form of heatmaps, path lines, etc., to help users perceive the conflicts or connections between historical materials.

[0039] In one specific implementation, the step of dynamically presenting the temporal authority tension relationship of multi-source historical materials in the spatiotemporal narrative interface based on the confidence visualization mark results further includes:

[0040] Set up event-sensitive areas and determine whether to trigger the in-depth exploration mode based on the user's dwell time;

[0041] In the deep exploration mode, the system extracts text content from multiple historical sources and matches it with the current confidence score to output a credibility ranking;

[0042] Construct the text consistency ratio R = C_total / (C_total + E_total), where C_total is the total confidence score and E_total is the total error contribution.

[0043] The texts are arranged in a clustered manner, with those having high consistency in the core layer and those having lower consistency scattered on the periphery, and clickable to view detailed comparative analysis.

[0044] In one specific implementation, the step of dynamically presenting the temporal authority tension relationship of multi-source historical materials in the spatiotemporal narrative interface based on the confidence visualization mark results further includes:

[0045] Define the event influence function F=log(1+|ΔW_i|), where ΔW_i is the difference in user trust in the event before and after its history;

[0046] The historical credibility weight C_i is moved along the time axis to form a dynamic weight cloud map, which reflects the long-term evolution trend;

[0047] The personalized presentation parameter P is generated using the formula P=(C_i×α+ΔW_i×β) / (C_i+ΔW_i), where α is the weight bias term.

[0048] Historical data is dynamically sorted according to the P parameter, prioritizing the display of information that is of interest to users and has a high level of trust.

[0049] A spatiotemporal narrative and immersive interactive display system showcasing the evolution of the Sui and Tang dynasty's military system, including:

[0050] The data acquisition module obtains the user's speed and dwell time during the timeline scrolling process and generates user behavior trajectory data;

[0051] The determination module, based on the user behavior trajectory data, determines the user's attention distribution pattern on key turning points in the Fubing system;

[0052] The visualization module, based on the attention distribution pattern, adaptively decays and marks the confidence of historical materials corresponding to key turning points in the Fubing system with visualization.

[0053] The presentation module dynamically presents the time-authority tension relationship of multi-source historical materials in the spatiotemporal narrative interface based on the confidence visualization mark results.

[0054] This disclosure provides a method for spatiotemporal narrative and immersive interactive display of the evolution of the Sui and Tang Dynasty's Fubing system, including: acquiring the user's speed and dwell time during timeline scrolling to generate user behavior trajectory data; determining the user's attention distribution pattern on key turning points in the Fubing system based on the user behavior trajectory data; adaptively attenuating and visually marking the credibility weights of historical materials corresponding to key turning points in the Fubing system based on the attention distribution pattern; and dynamically presenting the time-authority tension relationship of multiple historical materials in the spatiotemporal narrative interface based on the confidence visualization marking results. Through the solution of this disclosure, the credibility weights of historical materials corresponding to key turning points in the Fubing system can be adaptively attenuated and visually marked based on the user's speed and dwell time during timeline scrolling, thus solving the problem that the time-authority tension of multiple historical materials is not made explicit by interactive feedback. Attached Figure Description

[0055] To more clearly illustrate the technical solutions of the exemplary embodiments of this disclosure, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this disclosure and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 It is a flowchart of the spatiotemporal narrative and immersive interactive display method of the evolution of the Sui and Tang Dynasty's Fubing system;

[0057] Figure 2 It is a framework diagram of the spatiotemporal narrative and immersive interactive display system of the evolution of the Sui and Tang Dynasty military system. Detailed Implementation

[0058] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the embodiments of this disclosure will be further described in detail below with reference to the accompanying drawings. The illustrative implementation methods and descriptions of the embodiments of this disclosure are only used to explain the embodiments of this disclosure and are not intended to limit the embodiments of this disclosure.

[0059] The following describes in detail the spatiotemporal narrative and immersive interactive display method for the evolution of the Sui and Tang Dynasty military system provided by this invention, with reference to specific embodiments. This method can run on terminal devices equipped with touchscreens or mouse interaction (such as tablets, interactive large screens, or virtual reality headsets). The system front-end uses HTML5 / WebGL technology to construct the timeline and spatiotemporal narrative interface, while the back-end uses Python or Node.js to process user behavior data and calculate historical data weights. The specific steps are as follows.

[0060] Step 1: Obtain the user's speed and dwell time during the timeline scrolling process to generate user behavior trajectory data.

[0061] The system first presents a horizontal timeline on the display interface, with the horizontal axis covering the key evolutionary periods of the Sui and Tang Dynasty's Fubing system (e.g., 550-750 AD). Several key turning points in the Fubing system are marked on the timeline, including but not limited to the "Jiande Reform" (577 AD), the normalization of conscription during the Yifeng era (676-679 AD), and the edict abolishing the Zhechongfu in the tenth year of the Kaiyuan era (722 AD). Users can slide left and right along the timeline using their fingers (touchscreen) or drag with a mouse. The system records user interaction events in real time at a sampling frequency of once every 50 milliseconds. Each sampling acquires the current time position on the timeline (in AD years) and the timestamp of that moment (in system milliseconds). To calculate the sliding rate, the system records the time interval Δt (in seconds) between two consecutive samples and the change in timeline position Δx (in years). The instantaneous sliding rate V = |Δx / Δt|, in "years / second". Simultaneously, the system determines "staying" behavior by judging whether the user's swiping rate is lower than a preset threshold (e.g., V < 0.5 years / second) and the duration exceeds 0.3 seconds within any consecutive time period. The stay duration T is the total time (in seconds) elapsed from entering to leaving the stay state. Each swipe event and stay event is recorded as a behavior trajectory data item, including the timeline position, swipe rate V, stay duration T (T=0 if there is no stay), and the corresponding event node ID (if a key event node exists within 10 years of the current location, it is bound to that node). Finally, the system organizes the data from all sampling points into a user behavior trajectory dataset and stores it in an in-memory queue for further analysis.

[0062] Step 2: Based on the user behavior trajectory data, determine the user's attention distribution pattern on key turning points in the Fubing system.

[0063] After obtaining user behavior trajectory data, the system calculates the attention intensity for each preset key turning point event node (a total of N, for example, N=12 representative events). The specific method is as follows: First, extract the time interval [t_i-δ, t_i+δ] corresponding to each event node i, where δ is the event influence radius (for example, 5 years). Within this interval, retrieve all user behavior sampling points. For each sampling point, if its sliding rate V_j is lower than the dwell rate threshold (e.g., 0.5 years / second) and the corresponding dwell time T_j>0, then the sampling point is considered a valid dwell behavior; otherwise, it is considered a sliding behavior. Then, calculate the attention weight W_i=α×V_avg_i+β×T_total_i for event i, where V_avg_i is the average sliding rate of all sliding sampling points within the event interval (unit: years / second), T_total_i is the sum of all dwell times within the event interval (unit: seconds), and α and β are normalization coefficients. To ensure comparability between the two contributions, α is set to 1 / (V_max-V_min), and β is set to 1 / (T_max-T_min). V_max and T_max represent the maximum average rate and maximum total dwell time among all events, while V_min and T_min are the minimum values ​​(usually 0.1 and 0 to prevent division by zero). Through the above weighted calculation, the attention weight W_i for each event i is obtained. Furthermore, the W_i values ​​of all N events are arranged in event order to form an attention distribution vector A=[W_1,W_2,…,W_N]. To reduce the impact of transient outliers (such as accidental touches or rapid glancing) on ​​the analysis, the system uses a sliding window of length 3 to smooth vector A: for the i-th event, the smoothed attention intensity A_smooth(i)=(A(i-1)+A(i)+A(i+1)) / 3, and boundary events use the average of the available values ​​within the window. Ultimately, the smoothed A_smooth(i) represents the distribution pattern of user attention to key turning points in the Fubing system.

[0064] Step 3: Based on the attention distribution pattern, adaptively decay and confidence visualization mark the historical credibility weights corresponding to key turning points in the Fubing system.

[0065] The system pre-establishes a multi-source historical database for each key turning point event. The historical sources include official histories (such as the *Old Book of Tang*, *New Book of Tang*, and *Zizhi Tongjian*), epitaphs (such as the *Epitaph of the Late Zhang Jun of the Great Tang Dynasty*), Dunhuang manuscripts (such as the fragment of the *Kaiyuan Shuibu Shi*, p.2507), and restored clauses of the *Tiansheng Ling* (such as Article 21 of the *Tian Ling*). Each type of historical source corresponds to a basic credibility weight C0_base; for example, official histories are 0.9, epitaphs are 0.7, Dunhuang manuscripts are 0.6, and restored clauses of the *Tiansheng Ling* are 0.85. For each historical source k in event i, its initial credibility weight C0(i,k) = C0_base(type) × chronological correction coefficient (the closer the chronological period is to the year the event occurred, the closer the correction coefficient is to 1.0; otherwise, it decays linearly). Then, based on the attention distribution vector A_smooth obtained in step two, the adaptive decay factor for event i is calculated. Specifically, let A_max be the maximum value of the smoothed attention intensity among all events. Then, the decay factor γ_i for event i is (A_max - A_smooth(i)) / A_max. The value of γ_i ranges from [0,1]. When the user's attention intensity for event i is closer to the maximum value, γ_i approaches 0, indicating that no decay is needed; when the attention intensity is very low, γ_i approaches 1, indicating that significant decay is needed. Next, the credibility weight of each historical data k under event i is adjusted: C_adj(i,k) = C0(i,k) × (1 - γ_i^β), where β is the exponential decay parameter. In this embodiment, β = 1.5 is taken to enhance the decay sensitivity under low-attention events. The adjusted C_adj(i,k) is the credibility weight after adaptive decay. Next, the system uses a confidence visualization to mark the weight: for each event node and its associated historical materials, they are displayed on the interface as semi-transparent dots or cards. The transparency of the color is inversely proportional to C_adj(i,k) (the higher the weight, the less transparent). If C_adj(i,k) decreases by more than 30% relative to the initial value C0(i,k), a flashing yellow exclamation mark is displayed next to the historical material label, indicating that "the confidence level has decayed due to user interaction." Users can click the exclamation mark to see the reason for the decay (e.g., "You have paid less attention to this event, and the system has reduced the weight of non-official historical materials").

[0066] Step 4: Based on the confidence visualization results, dynamically present the time-authority tension relationship of multi-source historical materials in the spatiotemporal narrative interface.

[0067] After completing the adaptive decay and labeling of historical source credibility weights, the system needs to dynamically present the time-authority tension relationship between multiple historical sources in the spatiotemporal narrative interface. The so-called "time-authority tension" refers to the degree of contradiction or difference between different historical sources when recording the same event, caused by factors such as the time of writing, the author's perspective, and the source of excavation. The specific implementation method for this step is as follows: First, establish a two-dimensional coordinate system in the interface, with the horizontal axis representing time (year in the Gregorian calendar) and the vertical axis representing the authority level (obtained by weighting the historical source type and its dynamic credibility weight C_adj(i,k)). For each key turning point i, the system maps each associated historical document k to a point. The x-axis of this point represents the year of the event recorded in the historical document (if the year is not explicitly recorded, the standard year of the event is used), and the y-axis represents the normalized authority score: Score_authority(i,k) = 0.4 × C_adj(i,k) + 0.3 × type_weight + 0.3 × citation_freq. Here, type_weight represents the inherent authority of the historical document type (official history = 0.9, epitaphs = 0.7, Dunhuang manuscripts = 0.5, Tiansheng edict = 0.85), and citation_freq is the normalized value of the frequency of citations of this historical document in academia (pre-obtained from statistics from CNKI or CSSCI). Then, for different historical document points under the same event, the system calculates the pairwise Euclidean distance d = sqrt((Δt)^2 + (ΔScore)^2), where Δt is the time difference (in years), and ΔScore is the difference in authority scores. The average of all distances is used as the "Time-Authority Tension Index" T_i for the event. If T_i is greater than a preset threshold (e.g., 0.35), the event is considered to have significant historical tension. In terms of visualization, the system uses the event node as the center, radiating outwards to connect various historical data points. The thickness and color of the connecting lines represent the magnitude of the tension: the greater the tension, the thicker the line and the color gradually changes from blue to red. Simultaneously, when the user hovers the mouse over any historical data point, a detailed comparison card pops up, showing the percentage difference between that historical data and other historical data related to the same event in terms of time record and authority rating. Furthermore, the system dynamically generates a "tension heatmap" overlaid on the timeline, with the horizontal axis representing time and the vertical axis representing tension intensity. Users can intuitively see which historical periods (e.g., from the Yifeng era to the Kaiyuan era) saw the most intense conflicts among multiple historical sources. Finally, users can directly jump to the corresponding event node by clicking on the highlighted area on the heatmap, automatically triggering a recalculation of the attention distribution in step two, achieving a closed-loop interactive feedback.

[0068] Through the coordinated work of the above four steps, this invention can adaptively adjust the confidence weight of different historical materials according to the user's scrolling speed and dwell time on the timeline, and present the time-authority tension between multiple historical materials to the user in a visual way. This solves the technical problem in the prior art that the differences in the authority of historical materials are not made explicit by interactive feedback, and significantly improves the exploration efficiency and cognitive depth of the evolution of the Fubing system in digital humanities research.

[0069] In a preferred implementation of step two, which determines the user's attention distribution pattern for key turning points in the Fubing system based on user behavior trajectory data, the system further employs a weighted calculation and sliding window smoothing method to improve the accuracy of the attention distribution pattern. Specifically, the system first extracts all sampling points within the time interval [t_i-δ, t_i+δ] corresponding to each key turning point node i from the user behavior trajectory data, and obtains the sliding rate V (unit: years / second) and dwell time T (unit: seconds) at each sampling point. The sliding rate V is obtained by dividing the time axis position difference between two consecutive sampling points by the time interval; the dwell time T is obtained by accumulating the time span of a sequence of sampling points continuously below the rate threshold (e.g., 0.5 years / second). Then, for event i, the system calculates the average sliding rate V_avg and the total dwell time T_sum of all sampling points within that interval. To avoid weighting distortion caused by the inconsistency in the dimensions of V_avg and T_sum, the system pre-calculates the maximum average sliding rate V_max and the maximum total dwell time T_max for all events, and sets normalization coefficients α=1 / (V_max+ε) and β=1 / (T_max+ε), where ε=0.001 to prevent division by zero. Therefore, the attention weight W_i for event i is W_i=α×V_avg+β×T_sum. When a user dwells for a longer period (e.g., V_avg=0.2 years / second, T_sum=8 seconds) at a slower pace, W_i will have a higher value; conversely, a faster passage (V_avg=10 years / second, T_sum=0 seconds) will result in a lower W_i. Arranging the W_i values ​​of all N events in chronological order yields the original attention distribution vector A_raw=[W_1,W_2,…,W_N]. Due to potential momentary jitter or accidental touches during user operation, the system uses a sliding window of length 3 to smooth A_raw: for i=2 to N-1, the smoothed attention intensity A(i) = (A_raw(i-1) + A_raw(i) + A_raw(i+1)) / 3; for i=1, A(1) = A_raw(1); for i=N, A(N) = A_raw(N). The smoothed A(i) is then output as the final attention distribution pattern to subsequent steps. This scheme effectively filters out noise, making the attention distribution more realistically reflect the user's level of attention to key turning points in the Fubing system.

[0070] In a preferred implementation of the adaptive decay and confidence visualization labeling of historical material credibility weights based on attention distribution patterns in step three, the system introduces an exponential decay model based on attention intensity and a visualization mapping rule. First, the system presets an initial historical credibility weight C0 for each type of historical material: 0.9 for official histories (such as the *Old Book of Tang* and *New Book of Tang*), 0.7 for epitaphs, 0.6 for Dunhuang manuscripts (such as P.2507), and 0.85 for the restoration clauses of the *Tiansheng Edict*. For each key turning point event i and its associated k-th historical material, the initial weight C0(i,k) is multiplied by a chronological correction factor (the absolute value of the difference between the standard year of the event and the year the historical material was written; if less than 20 years, the correction factor is 1.0, decreasing by 0.05 for every additional 10 years, with a minimum of 0.6). Then, based on the obtained attention distribution vector A (after smoothing), the decay factor γ_i = (A_max - A(i)) / A_max for event i is calculated, where A_max is the maximum value of A(i) among all events. γ_i∈[0,1], the larger A(i) is (the more attention users pay), the closer γ_i is to 0, and the smaller the decay; the smaller A(i) is, the closer γ_i is to 1, and the more drastic the decay. Next, the adjusted credibility weight is calculated using the exponential decay formula: C_adj(i,k)=C0(i,k)×(1-γ_i^β), where β is the exponential decay parameter, and in this embodiment, β is taken as 1.5. For example, if an event A(i)=0.2 and A_max=0.8, then γ_i=(0.8-0.2) / 0.8=0.75, γ_i^β=0.75^1.5≈0.65, then C_adj=C0×(1-0.65)=0.35×C0, that is, the weight decreases by about 65%. After calculation, the system maps C_adj to visual markers: next to event nodes on the timeline, a semi-circular list of historical materials is displayed, with the transparency of the bubbles inversely proportional to C_adj (completely opaque when C_adj=0.9, and 70% transparent when C_adj=0.3). Simultaneously, if C_adj decreases by more than 30% compared to C0, a flashing orange dot appears in the upper right corner of the bubble, accompanied by the text "Confidence Decay". Users can click on the bubble to view a bar chart comparing the original C0 with the current C_adj. This scheme achieves an adaptive response of historical material credibility weights to user attention, enhancing the explicitness of interactive feedback.

[0071] In a more preferred embodiment, an event importance index based on the user's sliding direction and speed change is introduced, and a confidence threshold is set accordingly to enable the attenuation mechanism. The specific steps are as follows: The system monitors the user's sliding behavior on the timeline in real time, recording the sliding rate V(t) and the sliding direction (positive is to the right / towards the future, negative is to the left / towards the past) at each moment. For each key turning event i, within its corresponding time interval, calculate the rate of change of speed dT / dt (that is, the difference in rates between two adjacent sampling points divided by the time interval, unit: year / second²). Then define the event importance index E_i = V_avg_i × (dT / dt)_max, where V_avg_i is the average sliding rate within this event interval, and (dT / dt)_max is the maximum absolute value of the rate of change of speed within this event interval. When the user suddenly decelerates or slides in the opposite direction near an event, (dT / dt)_max is larger, resulting in an increase in E_i, indicating that this event may have higher research value. Next, the system calculates the average value of the attention distribution of all events A_avg = (1 / N)∑A(i) and the average value of the importance index E_avg = (1 / N)∑E_i, and sets the confidence threshold K = A_avg + λ × E_avg, where λ is the adjustment coefficient, and in this embodiment, λ = 0.5. For each historical material k in event i, compare its adjusted credibility weight C_adj(i,k) with K: If C_adj(i,k) < K, it is determined that this historical material needs to enable the confidence attenuation mechanism (that is, further reduce its displayed confidence). The specific operation of the attenuation mechanism is: Multiply C_adj(i,k) by the quadratic attenuation factor σ = 0.8 to obtain the final display weight C_final = C_adj(i,k) × 0.8, and inform the user in the form of dynamic highlighting (such as flashing red border) and pop-up prompt ("This historical material is marked as low confidence due to your important attention behavior to this event. Please check other sources") in the interface. If C_adj(i,k) ≥ K, keep C_adj unchanged. Through the above method, the system can identify the "important events" implied by the user's variable-speed sliding behavior, and specifically reduce the confidence mark of historical materials related to this event but with originally low credibility, so as to more accurately reflect the user's potential research intention.

[0072] In another embodiment, a dynamic window sliding analysis method is further employed to monitor the changing trend of user attention in real time and adaptively adjust the decay rate according to the attention slope, while presenting the weight changes in a real-time chart. Specifically, the system sets a fixed-length time window L (e.g., L = 30 seconds), which slides forward continuously with the user's interaction time. Within each window, the user's attention intensity A(i) (smoothed) for each event node is recorded. For the currently focused event i, the system calculates its attention slope ΔA = (A_current - A_prev) / Δt, where A_current is the average attention intensity of event i in the current window, A_prev is the average attention intensity of the same event in the previous window, and Δt is the time difference between the start times of the two windows (in seconds). The sign of ΔA indicates whether attention is increasing or decreasing, and its absolute value represents the rate of change. The system presets an attention stability threshold η = 0.05 (dimensionless, because A is already in the [0,1] interval). If |ΔA| < η, the system is considered to be in the attention stabilization phase. In this phase, the system allows the confidence weights of all historical materials associated with the event to remain at their original levels (i.e., no additional decay occurs, and C_final remains C_adj(i,k)). If |ΔA| ≥ η, the decay mechanism is activated, and the weight decay rate ρ = (|ΔA| / η)^2 is defined. For example, when |ΔA| = 0.1, ρ = (0.1 / 0.05)^2 = 4; when |ΔA| = 0.2, ρ = 16. The larger the decay rate ρ, the faster the confidence weights decrease. Subsequently, the system applies decay to each historical material k under event i: C_final(i,k) = C_adj(i,k) × (1 - ρ × Δt_adjust), where Δt_adjust is the duration after entering the decay state (unit: seconds, maximum limit of 0.2 seconds to prevent zeroing). To ensure a good user experience, the system generates a real-time line chart in the right sidebar of the interface. The horizontal axis represents time (seconds), and the vertical axis represents the weight values ​​of various historical materials under the current focused event (multiple lines represent official histories, epitaphs, Dunhuang manuscripts, etc.). When a change in user attention causes attenuation to begin, the line chart will decline in real time, and a text prompt such as "Attention slope ΔA=0.12, attenuation rate ρ=5.76" will be displayed below the chart. Users can click on any point on the chart to lock the weight distribution at that moment for detailed analysis. This solution not only achieves a sensitive response to changes in attention but also visually presents the dynamic process of weights, significantly improving the system's interactive transparency and educational research value.

[0073] In another embodiment, the system can first establish a two-dimensional timeline coordinate system: the horizontal axis represents the time range, covering the key evolutionary period of the Sui and Tang dynasty's military system, for example, from 618 AD (the first year of Emperor Gaozu's Wude reign) to 756 AD (the fifteenth year of Emperor Xuanzong's Tianbao reign). The horizontal axis is marked in years and supports scaling and translation. The vertical axis represents the authorization level, with values ​​normalized to [0,1], where 0 represents the lowest authority (such as private notes or later legends) and 1 represents the highest authority (such as official history and verified by multiple sources). For each key turning point (such as the "Jiande Reform", the normalization of recruitment during the Yifeng era, and the edict abolishing the Zhechong Prefecture in the tenth year of the Kaiyuan era), the system maps its associated multi-source historical materials (official history, epitaphs, Dunhuang documents, and the restored clauses of the Tiansheng edict) to a point in the coordinate system. The x-coordinate of each point represents the year of the event recorded in the historical material (or the standard year of the event if no specific year is specified). The y-coordinate is a weighted composite of two parts: historical credibility weight C_i (i.e., C_final, which takes into account user attention decay) and attention weight W_i (from the original attention weights above, unsmoothed). Both are normalized to the interval [0,1], and the y-coordinate position of the point is calculated using the spatial mapping formula D=w1×C_i_norm+w2×W_i_norm, where w1 and w2 are weighting coefficients. In this implementation, w1=0.7 and w2=0.3 are chosen to emphasize that the credibility of the historical material itself is more important than user attention. The larger the D value, the higher the position of the historical material in the time-authority space. Then, the system calculates the Euclidean distance between any two historical material points under the same event. The greater the distance, the greater the difference in time and authority. In terms of visualization, the system uses semi-transparent polylines to connect all historical data points related to the same event. The color gradient of the polyline represents changes in distance (the color is redder where the distance is greater). Users can click on the line connecting any two points to bring up a comparison card, displaying the time difference, the difference in authority score, and the reasons for the difference (e.g., "Official history records it as 722 years, the epitaph records it as 725 years, a difference of 3 years; the difference in authority score is 0.3"). Furthermore, the system allows users to dynamically adjust the values ​​of w1 and w2 using sliders, changing the composition of the vertical axis in real time, thereby exploring the tension relationship under different weightings. This approach transforms the abstract tension of historical data into an interactive two-dimensional spatial distance, greatly enhancing users' ability to perceive conflicts between multiple historical sources.

[0074] In another implementation, independent time anchors and dynamic triggering mechanisms are established for key events, and local tension matrices and tension indices are introduced for visualization. Specifically, the system pre-selects the most representative key events from the evolution of the Fubing system, such as the Jiande Reform (577 AD), the normalization of conscription during the Yifeng era (median value 678 AD), and the edict abolishing the Zhechongfu in the tenth year of the Kaiyuan era (722 AD), and sets an independent time anchor for each event. The anchor is marked with a gold diamond icon on the timeline and is bound to a sensing area with a radius of 15 pixels. When the user slides on the timeline and hovers the cursor or finger near the anchor (within 10 pixels) for more than 0.8 seconds, the system automatically triggers local tension analysis. Upon triggering, the system reads all historical data items within a ±20-year time range surrounding the event corresponding to the anchor point. It obtains the historical credibility weight C_i (from yy's C_final) and attention weight W_i (from yy's original weight) for each item, constructing a local tension matrix M=[C_i;W_i], where M is a 2xm matrix and m is the number of historical data items. Then, it calculates the time tension index T_E=∑{i=1}^{m-1}∑{j=i+1}^{m}|C_i-C_j|, which is the sum of the absolute differences in credibility weights between all pairs of historical data; and the authority tension index A_E=∑{i=1}^{m-1}∑{j=i+1}^{m}|W_i-W_j|, which is the sum of the absolute differences in attention weights between all pairs of historical data. The larger the values ​​of T_E and A_E, the more severe the discrepancy in credibility and user attention among historical data items near the event. On the interface, the system dynamically refreshes the local tension matrix in the form of a heatmap: the horizontal and vertical axes of the heatmap are both historical document numbers, and the color depth of each cell represents the magnitude of |C_i-C_j| (the darker the red, the greater the difference). The historical documents are connected by path lines (Sankey diagram style), with the line width proportional to |W_i-W_j|. When a user clicks on any cell in the heatmap, the system automatically highlights the corresponding two historical documents and displays their detailed text snippets. In addition, the system displays a radar chart next to the anchor point, showing the percentile ranking of T_E and A_E relative to the historical average, such as "Events with a time tension index exceeding 92% and events with an authority tension index exceeding 68%." This approach allows users to quickly locate the key events with the most intense historical document conflicts and to deeply analyze the specific sources of these conflicts.

[0075] Furthermore, in another embodiment, an event-sensitive area and a deep exploration mode are further set up to assist users in identifying credible historical materials through text consistency ratio and clustering. Specifically, the system delineates a circular sensitive area (with a radius of 30 pixels on the timeline, corresponding to approximately 15 years) for each key event node. When a user stays in this area for more than 2 seconds (without clicking), the system automatically determines that it has entered the deep exploration mode. After entering this mode, the system extracts all original text fragments (each fragment not exceeding 200 characters) related to the current event from local or cloud-based historical databases, including original historical texts, epitaph interpretations, Dunhuang manuscript transcriptions, and annotations of the restored clauses of the "Tiansheng Edict," etc. For each historical material k, the system matches its current confidence score (i.e., C_final, normalized to Confidence_k) with the key statements in the historical material text (such as the year of the event, the issuer, and the content summary). The specific matching method is as follows: a pre-trained BERT model is used to calculate the semantic similarity between the text statement and the standard event description to obtain the text consistency score Consistency_k. Then, the text consistency ratio R is calculated as R = (∑Confidence_k) / (∑Confidence_k + ∑(1-Consistency_k)), where the second term in the denominator is the sum of error contributions (i.e., the degree of text inconsistency). R ranges from [0,1]. A value closer to 1 indicates a high degree of consistency between the high-confidence historical material and the text, resulting in high overall credibility; a value closer to 0 indicates that the text content of most high-confidence historical materials contradicts each other. Subsequently, the system sorts all historical materials in descending order according to their current confidence score Confidence_k and uses a force-oriented clustering algorithm to present them on the interface: nodes with high consistency (Confidence_k > 0.7 and Consistency_k > 0.8) are clustered in the core layer (central circular area), nodes with medium consistency (0.4~0.7) are distributed in the middle ring, and nodes with low consistency (<0.4) are scattered in the outer ring. Nodes are connected by dashed lines to indicate references or mutual exclusion relationships. Users can click on any node, and the system will pop up a detailed comparative analysis panel, displaying a table comparing the historical material with other core-level historical materials on key sentences, and highlighting divergent words and phrases. This solution helps users automatically filter out the most trustworthy historical materials through an in-depth exploration mode, while intuitively presenting the textual consistency levels between different historical materials.

[0076] Furthermore, this embodiment of the invention further defines an event influence function and generates a dynamic weight cloud map and personalized presentation parameters to achieve adaptive ranking of historical materials for different users. Specifically, the system monitors in real time the change in user trust in each event i, ΔW_i, defined as the change in the user's average confidence score for the historical materials related to the event in the current session: ΔW_i = (current average C_final) - (initial average C0), where the initial average C0 is the average base credibility at the beginning of the session. A positive ΔW_i indicates that the user's overall trust in the historical materials of the event has increased after interaction, while a negative ΔW_i indicates a decrease in trust. Then, the event influence function F = log(1 + |ΔW_i|) is defined, which compresses the change in trust to a logarithmic scale to avoid the influence of extreme values. When the absolute value of ΔW_i is large, F increases accordingly, indicating that the event has a significant impact on the user's perception. Next, the system moves the historical credibility weight C_i (the average C_final of all historical data under that event) of each event i along the time axis (for example, when the user swipes from left to right, the system updates the C_i value in the current window at a rate of 10 frames per second), forming a dynamic weight cloud map. The cloud map uses the time axis as the horizontal axis and the C_i value as the vertical axis, using a semi-transparent colored point cloud to represent the weight distribution of each event, with the color gradually changing from blue (low weight) to red (high weight), and the size of the points is proportional to F. Users can see that as their interaction behavior changes, the weight cloud points of certain events gradually become redder and larger, intuitively reflecting their increasing influence. In addition, the system uses the formula P=(C_i×α+ΔW_i×β) / (C_i+ΔW_i+ε) to generate personalized presentation parameters P, where α is the weight bias term (α=0.8 in this embodiment), β is the trust change sensitivity coefficient (β=0.5), and ε=0.001 to prevent division by zero. The value of P is experimentally controlled within the range of [0.2, 1.2]. A larger P indicates that the historical material is more worthy of priority display under the current user model. Finally, the system dynamically sorts all historical materials for each event according to the P value from largest to smallest, and displays the top three historical materials with the highest P values ​​as cards at the top of the event details panel, with a green "Recommended for Priority Reading" label. After each user swipes or stays on the screen, the system recalculates ΔW_i and P, and updates the sorting in real time, thus achieving personalized historical material recommendations for each user. This solution visualizes the user's own trust changes through influence functions and dynamic cloud graphs, and optimizes the information presentation order accordingly, significantly improving the intelligence and user stickiness of the immersive interactive display system.

[0077] Furthermore, this embodiment of the invention further collects the frequency of changes in the user's swiping direction and constructs a behavior vector. High-frequency feature components are extracted using wavelet transform to generate an interest index, thereby more precisely characterizing the attention distribution. Specifically, the system records the direction and speed changes of each swipe operation in real time with a sampling period of 50 milliseconds. Let the swipe speed vector at sampling time t be (v_x(t), v_y(t)), where v_x(t) is the speed in the horizontal direction of the time axis (positive values ​​indicate swiping to the right / future, negative values ​​indicate swiping to the left / past), and v_y(t) is the speed in the vertical direction (vertical swiping on a touchscreen typically does not change the time axis position, but can be used to detect gesture intent). The system defines the behavior vector H=(ΔV_x,ΔV_y), where ΔV_x=v_x(t)-v_x(t-Δt) is the horizontal speed change, ΔV_y=v_y(t)-v_y(t-Δt) is the vertical speed change, and Δt is the sampling interval. H reflects the jerk of the user's swipe behavior per unit time. Next, the system performs wavelet transform on the horizontal velocity change sequence ΔV_x in H, using the Daubechies-4 wavelet as the basis function, and decomposes it into three layers. Components related to attention changes are extracted from the high-frequency detail coefficients d1 and d2 obtained from the wavelet transform, as these high-frequency coefficients correspond to behaviors such as rapid directional reversals and sudden acceleration / deceleration during user swiping, which often occur when the user develops interest in an event. Specifically, the system calculates the high-frequency feature component H_ω = sqrt(mean(d1^2) + mean(d2^2)), which is the root mean square of the detail coefficients at each layer. Then, the interest index I is defined as I = ∑(H_ω*sin(ω*t)), where ω is the frequency term, taken as the center frequency corresponding to the high-frequency band in the wavelet transform (e.g., ω = 2π × 5Hz), and t is the current session time. This formula highlights the periodic fluctuations in attention over time by sinusoidally weighted summing of the high-frequency components. Finally, the system combines the interest index I with the proportion of time a user spends on event i (T_ratio, i.e., the time spent on that event divided by the total time spent on it) to generate a weighted attention score S_i = I × T_ratio. Normalizing S_i replaces or modifies the attention weight W_i, resulting in a more refined attention distribution pattern. For example, when a user frequently and rapidly swipes back and forth (ΔV_x is large) and spends a long time near an event, both I and T_ratio are high, and S_i increases significantly, indicating that the user shows strong curiosity and a deep intention to explore the event. This scheme extracts micro-behavioral features through wavelet transform, enabling the system to capture attention fluctuation details that traditional rate-stay models cannot reflect.

[0078] Furthermore, this embodiment of the invention establishes a spatiotemporal coupling matrix between the historical event timestamp sequence and the user's swiping trajectory. It calculates event attention scores using exponential decay similarity and employs a clustering algorithm to group key turning points according to attention distribution, thereby optimizing subsequent resource allocation and display strategies. The specific steps are as follows: First, the system extracts the timestamp sequence T_seq=[t1,t2,…,tn] of all preset key turning points, where ti is the standard Gregorian year of the i-th event (e.g., the Jiande Reorganization is in 577). Simultaneously, it records the user's swiping trajectory G=[g1,g2,…,gm] throughout the entire interactive session, where each gj is a swiping trajectory point, containing its position coordinates on the timeline (represented by year) and the corresponding timestamp (number of milliseconds the system has run). To measure the spatiotemporal correlation between each event *ti* and the sliding trajectory, the system constructs a spatiotemporal coupling matrix *S_c*, whose element *s_ij* = e^{λ|(tj-g_i)|}, where *λ* is a decay constant (λ = -0.05 in this embodiment), and *|tj-g_i|* represents the absolute value of the difference between the event year and the year of the sliding trajectory point. This exponential function ensures that *s_ij* approaches 1 as the sliding trajectory point is closer to the event year; the farther away the point is, the faster *s_ij* approaches 0. Then, for each event *i*, the sum of its similarity to all sliding trajectory points is calculated as *S_c_total(i)* = ∑_{j=1}^{m}e^{λ|(ti-g_j)|}. To account for the impact of sliding rate on attention, the system further defines the event attention score A_i = (S_c_total(i) / max(S_c_total)) × (1 + α * ΔV_i), where max(S_c_total) is the maximum sum of similarities among all events, α is the attention coefficient when the sliding rate reaches the threshold (in this implementation, α = 0.3, which takes effect when the average sliding rate near the event exceeds 2 years / second), and ΔV_i is the difference between the average sliding rate within 10 years near the event and the global average rate (after normalization). This formula allows events that are frequently slid through (high S_c_total) and accompanied by rapid sliding behavior (high ΔV_i) to receive higher attention scores. Finally, the system combines the A_i of each event i with features such as attention weight W_i and interest index I_i into a multi-dimensional feature vector F_i = [A_i, W_i, I_i, ti]. The K-means clustering algorithm (K=3) was used to divide all key turning points into three groups based on attention distribution patterns: high-attention group (users frequently approached and stayed for long periods), medium-attention group (users occasionally stopped or quickly glanced at), and low-attention group (almost never touched). After clustering, the system dynamically adjusted resource allocation based on the grouping results: for events in the high-attention group, more detailed historical texts, high-resolution map layers, and 3D scene models were preloaded; for events in the low-attention group, only thumbnail information was loaded to conserve system resources.Meanwhile, different colored background bars mark each group interval on the timeline interface, allowing users to filter and display only events in the high-interest group with a single click, thus focusing on historical nodes that truly interest them. This solution, through spatiotemporal coupling matrices and cluster analysis, achieves the discovery and proactive adaptation to users' implicit attention patterns, significantly improving the efficiency of interactive display on large-scale historical data.

[0079] The above describes embodiments of the method of the present invention. Furthermore, as... Figure 2 As shown, the present invention also provides a spatiotemporal narrative and immersive interactive display system 200 for the evolution of the Sui and Tang Dynasty military system, comprising:

[0080] The data acquisition module 201 acquires the user's speed and dwell time during the timeline scrolling process and generates user behavior trajectory data;

[0081] The determination module 202 determines the user's attention distribution pattern on key turning points of the Fubing system based on the user behavior trajectory data.

[0082] The visualization module 203, based on the attention distribution pattern, adaptively decays and marks the credibility weights of historical materials corresponding to key turning points in the Fubing system with confidence visualization.

[0083] The presentation module 204 dynamically presents the time-authority tension relationship of multi-source historical materials in the spatiotemporal narrative interface based on the confidence visualization mark results.

[0084] The functions of each module of the spatiotemporal narrative and immersive interactive display system 200 for the evolution of the Sui and Tang Dynasty military system of the present invention have been described above with reference to the method embodiments, and will not be repeated here.

[0085] The above description is the preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this invention, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A spatiotemporal narrative and immersive interactive display method for the evolution of the Sui and Tang dynasty military system, characterized by: Including: Obtain the rate and dwell time of the user during the sliding process on the timeline, and generate user behavior trajectory data; Based on the user behavior trajectory data, determine the attention distribution pattern of the user towards the key turning events of the Fubing System; Based on the attention distribution pattern, perform adaptive attenuation and confidence visualization marking on the credibility weights of historical materials corresponding to the key turning events of the Fubing System; According to the confidence visualization marking results, dynamically present the time-authority tension relationship of multi-source historical materials in the spatio-temporal narrative interface.

2. The method for spatiotemporal narrative and immersive interactive display of the evolution of the Sui and Tang Dynasty military system according to claim 1, characterized in that, The determining the attention distribution pattern of the user towards the key turning events of the Fubing System based on the user behavior trajectory data includes: Obtain the sliding rate V and dwell time T of the user in different intervals during the sliding process on the timeline; Perform weighted calculation on V and T corresponding to each event node to obtain the attention weight W = α×V + β×T, where α and β are the normalization coefficients of the sliding rate and dwell time respectively; Construct an attention distribution vector A(i) according to the W value, where A(i) represents the attention intensity of the i-th key event; Use the sliding window average algorithm to smooth A(i) to reduce the impact of instantaneous outliers on the analysis.

3. The method for spatiotemporal narrative and immersive interactive display of the evolution of the Sui and Tang Dynasty military system according to claim 2, characterized in that, The performing adaptive attenuation and confidence visualization marking on the credibility weights of historical materials corresponding to the key turning events of the Fubing System based on the attention distribution pattern includes: Set the initial historical credibility weight C0, and assign different basic weight values according to the type of historical materials; According to the attention distribution vector A, construct the attenuation factor γ = (A_max - A(i)) / A_max, where A_max is the maximum attention value; The adjusted weight using the formula is C_i = C0×(1 - γ^β), where β is the exponential decay parameter; Map the result to the visualization marking, such as color transparency or label blinking frequency, to represent the trust level.

4. The method for spatiotemporal narrative and immersive interactive display of the evolution of the Sui and Tang Dynasty military system as described in claim 3, is characterized in that... The performing adaptive attenuation and confidence visualization marking on the credibility weights of historical materials corresponding to the key turning events of the Fubing System based on the attention distribution pattern includes: Generate an event importance index E according to the sliding direction and speed change of the user on the timeline, defined as E = V×dT / dt, where dT / dt is the rate of speed change; Combine the attention distribution vector A and the event importance index E, and define the confidence threshold K = A_avg + λ×E_avg, where A_avg is the mean value of A and E_avg is the mean value of E; Compare the historical credibility weight C_i with the threshold K. If C_i < K, then enable the confidence attenuation mechanism; Display the key historical material content with confidence decline in the form of dynamic highlighting and pop-up prompts through the graphical interface.

5. The method for spatiotemporal narrative and immersive interactive display of the evolution of the Sui and Tang Dynasty military system according to claim 4, characterized in that, The performing adaptive attenuation and confidence visualization marking on the credibility weights of historical materials corresponding to the key turning events of the Fubing System based on the attention distribution pattern further includes: Adopt the dynamic window sliding method to analyze the change trend of user attention, and define the attention slope ΔA = (A_current - A_prev) / Δt, where Δt is the time window length; If ΔA is less than the threshold η, it is judged as the attention stable stage, and the confidence weight is allowed to remain at the original level; If ΔA is greater than η, the decay mechanism is activated, and the weight decay rate is defined as ρ=(ΔA / η)^2; The decay mechanism is applied to the confidence level of each historical material segment, and the weight changes are presented in real-time charts in the visualization.

6. The spatiotemporal narrative and immersive interactive display method for the evolution of the Sui and Tang Dynasty military system as described in claim 5, is characterized in that... The step of dynamically presenting the temporal authority tension relationship of multi-source historical materials in the spatiotemporal narrative interface based on the confidence visualization mark results further includes: Establish a timeline coordinate system, with the horizontal axis representing the time range and the vertical axis representing the authorization level; The historical credibility weight C_i and attention weight W_i are normalized to the [0,1] interval and used as the time point and access position of each historical material; A visual distance index is constructed using the spatial mapping formula D=w1*C_i+w2*W_i, where w1 and w2 are weighting coefficients used to distinguish the relative influence of time and attention. Historical materials are displayed in the form of nodes or broken lines. The greater the distance, the greater the difference in time and authority. Details can be expanded through user interaction.

7. The spatiotemporal narrative and immersive interactive display method for the evolution of the Sui and Tang Dynasty military system as described in claim 6, is characterized in that... The step of dynamically presenting the temporal authority tension relationship of multi-source historical materials in the spatiotemporal narrative interface based on the confidence visualization mark results further includes: Establish independent time anchors for key events and set up dynamic triggering mechanisms; When a user stays near the anchor point, the system automatically reads the historical weights and corresponding confidence scores of the surrounding historical materials and constructs a local tension matrix M=[C_i;W_i]; Introduce the time tension index T_E=∑|C_i-C_j| and the authority tension index A_E=∑|W_i-W_j|, where j represents adjacent historical data items; The tension matrix is ​​dynamically refreshed on the interface and presented in the form of heatmaps, path lines, etc., to help users perceive the conflicts or connections between historical materials.

8. The method for spatiotemporal narrative and immersive interactive display of the evolution of the Sui and Tang Dynasty military system according to claim 7, characterized in that, The step of dynamically presenting the temporal authority tension relationship of multi-source historical materials in the spatiotemporal narrative interface based on the confidence visualization mark results further includes: Set up event-sensitive areas and determine whether to trigger the in-depth exploration mode based on the user's dwell time; In the deep exploration mode, the system extracts text content from multiple historical sources and matches it with the current confidence score to output a credibility ranking; Construct the text consistency ratio R = C_total / (C_total + E_total), where C_total is the total confidence score and E_total is the total error contribution. The texts are arranged in a clustered manner, with those having high consistency in the core layer and those having lower consistency scattered on the periphery, and clickable to view detailed comparative analysis.

9. The method for spatiotemporal narrative and immersive interactive display of the evolution of the Sui and Tang Dynasty military system according to claim 8, characterized in that, The step of dynamically presenting the temporal authority tension relationship of multi-source historical materials in the spatiotemporal narrative interface based on the confidence visualization mark results further includes: Define the event influence function F=log(1+|ΔW_i|), where ΔW_i is the difference in user trust in the event before and after its history; The historical credibility weight C_i is moved along the time axis to form a dynamic weight cloud map, which reflects the long-term evolution trend; The personalized presentation parameter P is generated using the formula P=(C_i×α+ΔW_i×β) / (C_i+ΔW_i), where α is the weight bias term. Historical data is dynamically sorted according to the P parameter, prioritizing the display of information that is of interest to users and has a high level of trust.

10. A spatiotemporal narrative and immersive interactive display system for the evolution of the Sui and Tang dynasty military system, characterized by: include: The data acquisition module obtains the user's speed and dwell time during the timeline scrolling process and generates user behavior trajectory data; The determination module, based on the user behavior trajectory data, determines the user's attention distribution pattern on key turning points in the Fubing system; The visualization module, based on the attention distribution pattern, adaptively decays and marks the confidence of historical materials corresponding to key turning points in the Fubing system with visualization. The presentation module dynamically presents the time-authority tension relationship of multi-source historical materials in the spatiotemporal narrative interface based on the confidence visualization mark results.

Images