A method and system for evaluating the effect of intelligent communication of Chinese traditional culture

Abstract

The application relates to the technical field of intelligent communication evaluation, and discloses a Chinese traditional culture intelligent communication effect evaluation method and system. The method comprises the following steps: analyzing original culture communication content, extracting element features such as cultural connotation, emotional color and historical background; collecting direct interaction records and indirect behavior tracks of audiences; performing grid-by-grid matching and superposition of the element features and audience feedback in a preset space-time grid; quantitatively scoring the superposition data according to preset weights and rules; introducing initial scores into a simulator based on a culture communication law model to perform dynamic evolution calculation; performing multi-round iteration optimization on the evolution result by using incremental feedback data to generate an evaluation value; and mapping the evaluation value into a correction instruction for content form, communication channel and presentation rhythm. The method realizes fine evaluation and dynamic trend prediction of communication effect in the space-time dimension, and supports accurate optimization and adaptive adjustment of a communication strategy.

Description

Technical Field

[0001] This invention relates to the field of intelligent communication evaluation technology, specifically to a method and system for evaluating the effectiveness of intelligent communication of traditional Chinese culture. Background Technology

[0002] Current assessments of the effectiveness of the dissemination of traditional Chinese culture primarily rely on macro-level statistical data. Existing technologies typically treat the disseminated content as a whole or perform only coarse-grained classification, then correlate it with surface-level behavioral indicators of the audience. This approach provides information on the total volume and average of dissemination activities. However, the actual effects of cultural dissemination are spatially and temporally dependent; different regional cultural backgrounds and varying social focuses at different times profoundly influence audience acceptance and response patterns. Existing macro-level statistical methods cannot analyze the distributional differences of effects across specific time intervals and geographical areas, making it difficult to use assessment conclusions to guide the optimization of dissemination strategies for specific spatiotemporal scenarios.

[0003] Regarding the dynamics of assessment, existing technologies mostly employ regression models or trend extrapolation based on historical data. These methods are built upon the assumptions of static models, and their assessments are descriptions or linear extensions of past states. However, cultural transmission is a non-linear process involving social network diffusion and group psychological interaction, and its effects may accelerate, decay, or abruptly change. Existing static models cannot simulate such dynamic mechanisms; they cannot reflect real-time changes in the transmission path, nor can they predict the turning points in effects triggered by key nodes or events.

[0004] Current technology lacks an evaluation method that can accurately characterize the spatiotemporal distribution details of propagation effects and effectively simulate their dynamic evolution trends. This limits the accuracy and foresight of propagation strategy adjustments. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for evaluating the effectiveness of intelligent dissemination of traditional Chinese culture, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, this invention provides a method for evaluating the effectiveness of intelligent dissemination of traditional Chinese culture, the method comprising:

[0007] The original cultural transmission content is analyzed, and the elements containing cultural connotations, emotional coloring and historical background are extracted to construct a set of elements and features;

[0008] We obtain direct interaction records and indirect behavioral trajectories of the audience with cultural dissemination content from multiple independent channels to form an audience feedback dataset;

[0009] Within a pre-defined spatiotemporal grid, the feature set and the audience feedback dataset are matched and overlaid grid by grid to generate overlaid data.

[0010] The superimposed data is quantitatively scored based on a pre-defined cultural value weight library and behavioral value judgment rules, and the initial quantitative results are output.

[0011] The initial quantification results are imported into the propagation effect simulator. The propagation effect simulator performs dynamic evolution calculations on the initial quantification results based on the pre-stored cultural propagation law model to obtain the evolved data.

[0012] The evolved data is optimized and corrected through multiple rounds of iterations. The optimization and correction are based on the incremental feedback data acquired in real time, and a propagation effect evaluation value is generated.

[0013] The evaluation value of the dissemination effect is mapped to the correction instructions for the presentation form, dissemination channels and presentation pace of cultural dissemination content.

[0014] Preferably, the method for analyzing the original cultural dissemination content, extracting element features containing cultural connotations, emotional connotations, and historical background, and constructing an element feature set includes:

[0015] The content scanning unit is activated to deconstruct the text, images, audio, and video contained in the cultural dissemination content, separating unstructured content data and structured metadata.

[0016] Call the cultural knowledge graph interface to load concept nodes, attribute relationships, and event templates related to traditional culture from the cultural knowledge graph, forming subgraph data of the cultural knowledge graph;

[0017] For the deconstructed text content, semantic slicing and dependency analysis are used to identify specific cultural symbols, historical figures, and traditional craft terms mentioned in it; for the deconstructed image and video content, visual object recognition and scene classification are used to identify traditional artifacts, clothing patterns, and architectural style elements contained therein; for the deconstructed audio content, voiceprint feature extraction and melody analysis are used to identify traditional tunes, instrument timbres, and dialect speech fragments contained therein.

[0018] The identified text symbols, visual elements, and audio segments are matched with the corresponding nodes in the subgraph data of the cultural knowledge graph. The similarity matching is calculated by calculating the cosine distance of the feature vectors. When the cosine distance is less than a set threshold, it is considered a successful match. Each successfully matched item is labeled with its cultural connotation category.

[0019] Text symbols, visual elements, and audio clips labeled with cultural connotation categories are compiled and attached with sentiment analysis tags and historical period tags. The sentiment analysis tags are assigned based on the analysis of the context of the item corresponding to the current tag, and the historical period tags are assigned based on the query of the event time attribute in the cultural knowledge graph.

[0020] All data items with attached cultural connotation category tags, sentiment analysis tags, and historical period tags are arranged and combined according to their content type and time sequence to form an element feature set. The element feature set is stored in the form of a list. Each record in the list includes content identifier, element content, cultural connotation category, sentiment tag, historical period tag, timestamp, and spatial location information.

[0021] Preferably, the method for obtaining direct interaction records and indirect behavioral trajectories of the audience regarding cultural dissemination content from multiple independent channels to form an audience feedback dataset includes:

[0022] Define multiple audience interaction data collection points, including the comment section of cultural content publishing platforms, topic forwarding chains on social media, sensors in offline cultural experience venues, and feedback questionnaires actively submitted by users;

[0023] We collect the text of comments, the number of likes, and the number of shares posted by the audience from the comment section of cultural content publishing platforms, and conduct sentiment analysis on the comment text to generate platform interaction records.

[0024] By using topic tracking technology, we can collect audiences' forwarding comments, quoted content, and secondary creation links from the topic forwarding chain on social media, analyze the audience's dissemination path and content variation process, and generate social dissemination records. We can also collect audiences' dwell time, movement trajectory, and number of times interactive exhibits are triggered from sensors in offline cultural experience venues, and generate offline behavior records.

[0025] The system performs structured analysis on user-submitted feedback questionnaires, extracts scores and open-ended answers regarding cultural content comprehension, liking, and willingness to further disseminate the content, and generates questionnaire feedback records.

[0026] Establish a unified data timeline, arrange platform interaction records, social communication records, offline behavior records, and questionnaire feedback records in chronological order, and associate records from different sources within the same time window. The basis for association is the common content identifier or audience anonymous identifier contained in the records.

[0027] All records after time sorting and association are deduplicated and outlier filtered. Deduplication is performed to remove identical records based on their hash values, while outlier filtering is performed to remove records that are obviously illogical based on preset behavioral logic rules, thus forming a structured audience feedback dataset.

[0028] Preferably, the method for matching and overlaying the feature set and the audience feedback dataset grid by grid within a preset spatiotemporal grid to generate the overlaid data includes:

[0029] Based on the needs of the evaluation task, a virtual spatiotemporal coordinate system is defined, dividing the time dimension into continuous equal-length time periods and the spatial dimension into continuous equal-area regions. The spatiotemporal grid is the Cartesian product of the time dimension and the spatial dimension.

[0030] Each record in the feature set is mapped to the corresponding spatiotemporal grid based on its timestamp and spatial location information. If a feature spans multiple spatiotemporal grids, it is segmented and weighted according to its duration and coverage to complete the spatiotemporal grid mapping of the feature set.

[0031] Each record in the audience feedback dataset is mapped to a corresponding spatiotemporal grid based on its timestamp and the geographic location information implied by the source channel, thus completing the spatiotemporal grid mapping of the audience feedback dataset. For each spatiotemporal grid, it is checked whether there are mapped feature records and audience feedback records.

[0032] If both mapped feature records and audience feedback records exist simultaneously in the spatiotemporal grid, a data overlay operation is performed. The data overlay operation involves cross-combining the cultural connotation categories, emotional tags, and historical period tags of all feature records in the spatiotemporal grid with the emotional tendencies, communication behavior types, and ratings of all audience feedback records in the spatiotemporal grid to generate a series of "feature-feedback" paired data, and recording the frequency and time order of each paired data.

[0033] If the spatiotemporal grid contains only mapped feature records or only mapped audience feedback records, then the spatiotemporal grid is marked as an incomplete data grid and no data overlay operation is performed.

[0034] Collect all "feature-feedback" pairing data and their frequency and time sequence information generated by the spatiotemporal grids that have performed data overlay operations, and summarize them into overlay data. The structure of the overlay data is a multi-level nested dictionary, where the key of the dictionary is the spatiotemporal grid coordinate and the value is a list of all "feature-feedback" pairing data within the spatiotemporal grid.

[0035] Preferably, the method for quantifying and scoring the superimposed data, based on a preset cultural value weight library and behavioral value judgment rules, and outputting the initial quantification result includes:

[0036] Load the pre-built cultural value weight library. The cultural value weight library is a database that stores the value weight coefficients corresponding to different cultural connotations, different historical periods, and different emotional tones. The value weight coefficients are set based on expert evaluation and historical dissemination data analysis.

[0037] Load predefined behavioral value judgment rules. A behavioral value judgment rule is a set of logical judgment statements used to determine the value contribution score corresponding to the feedback behavior based on the behavior type, emotional tendency intensity, and dissemination scope in the audience feedback record.

[0038] Traverse the list of "feature-feedback" paired data within each spatiotemporal grid in the overlay data;

[0039] For each pair of "element-feedback" pairing data in the list, first, based on the cultural connotation category, historical period label, and emotional label of the element part in the pairing data, query the cultural value weight library to obtain the corresponding basic weight of the cultural element.

[0040] Secondly, based on the behavioral type, emotional tendency intensity, and dissemination range of the feedback portion in the paired data, the behavioral value judgment rules are applied to calculate the corresponding feedback behavioral value score.

[0041] The basic weights of the acquired cultural elements are multiplied by the value scores of the feedback behaviors to obtain a single quantitative score for the paired data of "elements-feedback".

[0042] The individual quantization scores of all “feature-feedback” paired data within a spatiotemporal grid are summed to obtain the total grid quantization score of the spatiotemporal grid.

[0043] The total grid quantization scores of all spatiotemporal grids are calculated and arranged in the order of spatiotemporal grid coordinates to form an initial quantization result matrix. The initial quantization result matrix is ​​a two-dimensional array, where the row index corresponds to the time dimension and the column index corresponds to the spatial dimension. The value of each array element is the total grid quantization score.

[0044] Preferably, the method for importing the initial quantification results into the dissemination effect simulator, and the dissemination effect simulator dynamically calculating the evolution of the initial quantification results based on a pre-stored cultural dissemination law model to obtain the evolved data, includes:

[0045] Initialize the propagation effect simulator. The propagation effect simulator is a discrete event simulation environment based on agent modeling. The simulation environment is divided into a spatiotemporal grid with the same dimensions as the initial quantization result matrix. Each grid initializes an agent, and the state of the agent is set to the grid quantization total score of the corresponding grid.

[0046] Load a pre-stored cultural transmission law model from the model library. The cultural transmission law model defines the rules for the change of agent state. The rules include: assimilation rule, the state values ​​of adjacent grid agents will affect each other, and agents with high state values ​​will spread their influence to agents with low state values; decay rule, the state value of each agent will decay naturally over time; mutation rule, when the state value of an agent exceeds a certain threshold, it may trigger a jump or a sudden drop in state.

[0047] Set the simulation clock and iteration step size, and start the propagation effect simulator;

[0048] Within each iteration step, the propagation effect simulator traverses all grid agents and calculates the state value of each agent in parallel at the next moment based on the assimilation rule, decay rule, and mutation rule in the cultural propagation law model.

[0049] During the calculation process, the simulator records the trajectory of each agent's state value change, as well as the grid coordinates where the state value change exceeds a set range;

[0050] The simulation stops when the simulation clock reaches the preset simulation termination time, or when the change in the state values ​​of all agents is less than the stability threshold for several consecutive iterations.

[0051] Extract the final state values ​​of all grid agents when the simulation stops to form the evolved data matrix. The structure of the evolved data matrix is ​​consistent with the initial quantization result matrix, but the element values ​​have been updated according to the cultural transmission law model.

[0052] Preferably, the method for performing multiple rounds of iterative optimization and correction on the evolved data, based on real-time acquired incremental feedback data, and generating a propagation effect evaluation value includes:

[0053] New audience feedback data generated since the last evaluation is acquired in real time from the data collection points and used as incremental feedback data.

[0054] The same data processing flow as that used to form the audience feedback dataset is performed on the incremental feedback data, including format deconstruction, sentiment analysis, and spatiotemporal mapping, to obtain the incremental spatiotemporal grid mapping result corresponding to the incremental feedback data. The incremental spatiotemporal grid mapping result is then compared with the evolved data matrix.

[0055] For grids in the evolved data matrix that have incremental spatiotemporal grid mapping results at the same spatiotemporal grid coordinates, optimization and correction calculations are performed. The optimization and correction calculations involve weighted fusion of the evolved values ​​of the spatiotemporal grids and the quantitative scores of the incremental feedback. The weighting coefficients are dynamically adjusted according to the time freshness and source credibility of the incremental feedback data.

[0056] For grids in the evolved data matrix that do not have incremental spatiotemporal grid mapping results at the same spatiotemporal grid coordinates, their original values ​​remain unchanged.

[0057] After completing one round of traversal and calculation of all grids, a corrected data matrix is ​​obtained;

[0058] The data matrix after one correction is used as the new evolved data matrix. The steps from obtaining incremental feedback data in real time to obtaining the data matrix after one correction are repeated for multiple rounds of iteration.

[0059] Set iteration termination conditions, including reaching the maximum number of iterations, or the difference norm between the corrected data matrices generated by two consecutive iterations being less than the set tolerance.

[0060] When the iteration termination condition is met, the iteration process is stopped. All grid values ​​in the corrected data matrix generated in the last iteration are globally normalized. The normalized value range is mapped to the integer range of zero to one hundred to generate the propagation effect evaluation value. The propagation effect evaluation value is an evaluation value matrix with the same spatiotemporal grid dimension. The value of each position in the matrix represents the propagation effect score of the corresponding spatiotemporal region at the evaluation time.

[0061] Preferably, the method for mapping the dissemination effect evaluation value to correction instructions for the presentation form, dissemination channels, and presentation rhythm of cultural dissemination content includes:

[0062] The propagation effect evaluation value matrix is ​​analyzed to identify inefficient grid areas with evaluation values ​​below a preset qualified threshold, and efficient grid areas with evaluation values ​​significantly higher than the surrounding areas.

[0063] For the identified inefficient grid areas, extract the cultural connotation category, emotional tag combination, and frequently occurring behavior type in the audience feedback dataset of the inefficient grid areas in the feature set corresponding to the inefficient grid areas;

[0064] The extracted features of inefficient regions are input into the policy mapping rule base for querying. The policy mapping rule base stores the mapping relationship from "feature combination" to "correction suggestion".

[0065] Based on the query results, targeted correction instructions are generated. If the inefficient area characteristic corresponds to "the presentation is too obscure", then the presentation form correction instruction is generated to "add visual interpretation or storytelling". If the inefficient area characteristic corresponds to "the audience of the communication channel does not match", then the communication channel correction instruction is generated to "adjust to a younger or more professional platform". If the inefficient area characteristic corresponds to "the presentation pace is too fast", then the presentation pace correction instruction is generated to "insert a phased review and interactive session".

[0066] For the identified high-efficiency grid regions, their feature combinations are also extracted, the policy mapping rule base is queried, and correction instructions to "maintain or enhance the current features" are generated.

[0067] The correction instructions generated for all inefficient and efficient grid areas are compiled, classified and dereduplicated according to instruction type, and the same or similar instructions are merged to form the final set of instructions for correcting the presentation of cultural dissemination content, the set of instructions for correcting dissemination channels, and the set of instructions for correcting presentation rhythm.

[0068] Preferably, the method for constructing and updating the policy mapping rule base includes:

[0069] Collect successful and unsuccessful cases of cultural dissemination in history, conduct feature analysis on each case, and extract the characteristics of the cultural content's form of expression, the characteristics of the dissemination channels used, the characteristics of the content presentation rhythm, and the grid evaluation value pattern corresponding to the case in the final evaluation.

[0070] Cultural communication experts were invited to review the cases, and suggestions for improvement were marked for each case's combination of features. These suggestions included specific optimization recommendations for the form of expression, dissemination channels, and presentation rhythm.

[0071] The feature combination of the case is paired with the correction suggestions of the expert annotation to form an initial mapping rule, and all the initial mapping rules are stored in the initial strategy mapping rule library.

[0072] In actual use, an effect tracking cycle is started whenever the system generates a correction command and executes it;

[0073] After the effect tracking period ends, the propagation effect evaluation value of the relevant area is re-evaluated. If the evaluation value improves, the pairing of "feature combination - correction instruction" is used as positive feedback to enhance the weight of the corresponding rule in the strategy mapping rule base. If the evaluation value decreases or remains unchanged, it is used as negative feedback to reduce the weight of the corresponding rule in the strategy mapping rule base, or the corresponding rule is deleted under specific conditions.

[0074] New successful models and suggestions are regularly extracted from publicly available cultural communication research materials and industry reports. After manual or automatic review, they are added to the strategy mapping rule base in the form of new rules to achieve continuous updates to the strategy mapping rule base.

[0075] Preferably, the present invention also includes a system for evaluating the effectiveness of intelligent dissemination of traditional Chinese culture. The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the method for evaluating the effectiveness of intelligent dissemination of traditional Chinese culture as described above.

[0076] Compared with the prior art, the beneficial effects of the present invention are:

[0077] By constructing a pre-defined spatiotemporal grid, the analyzed cultural elements are matched and overlaid with audience feedback data from multiple channels, grid by grid. This allows for a precise correspondence between the specific cultural connotations and emotional elements in the content and the audience behavior within a specific time and geographical region. This clearly depicts the differentiated feedback patterns elicited by different cultural elements in different regions and time periods, revealing the spatiotemporal heterogeneity and clustering of communication effects. Optimization of communication strategies is no longer based on an overall overview, but rather on pinpointing weak links or advantageous features within specific spatiotemporal grids, thereby achieving precise configuration and scheduling of content delivery and channel selection in the spatiotemporal dimension.

[0078] The quantitative scoring results are imported into a simulator with a built-in model of cultural dissemination patterns for dynamic evolution calculations. This model encodes patterns such as the social network effects of cultural diffusion, audience psychological acceptance thresholds, and topic evolution paths. This approach transforms the evaluation process from a static snapshot to a dynamic extrapolation, capable of simulating the nonlinear growth, decay, or abrupt changes in dissemination effects over time. This generates predictions of dissemination trends over a future period, as well as the identification of key influencing nodes and potential risks. Iterative corrections based on the prediction results and real-time incremental data give the evaluation system adaptive evolutionary characteristics, enabling it to continuously approximate the complex real-world dissemination landscape and providing a simulation-based preliminary decision-making basis for dynamically adjusting the dissemination pace and intervention timing. Attached Figure Description

[0079] Figure 1 This is a schematic diagram illustrating the working principle of the intelligent dissemination effect evaluation method for traditional Chinese culture described in this invention.

[0080] Figure 2 A flowchart for generating the overlay data;

[0081] Figure 3 A flowchart for obtaining the evolved data;

[0082] Figure 4 Heatmap of the evaluation values ​​of the dissemination effect of traditional Chinese culture under different spatiotemporal grids;

[0083] Figure 5 A heat map for evaluating the spatiotemporal grid effect of short videos on the art of landscape design in Jiangnan gardens. Detailed Implementation

[0084] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0085] Please see Figure 1 This invention provides a method for evaluating the intelligent dissemination effect of traditional Chinese culture. The method includes: analyzing the original cultural dissemination content, extracting element features containing cultural connotations, emotional connotations, and historical background, and constructing an element feature set; obtaining direct interaction records and indirect behavioral trajectories of the audience regarding the cultural dissemination content from multiple independent channels to form an audience feedback dataset; matching and overlaying the element feature set and the audience feedback dataset grid by grid in a preset spatiotemporal grid to generate overlaid data; quantifying and scoring the overlaid data based on a preset cultural value weight library and behavioral value judgment rules, and outputting an initial quantification result; importing the initial quantification result into a dissemination effect simulator, which dynamically evolves the initial quantification result based on a pre-stored cultural dissemination law model to obtain evolved data; performing multiple rounds of iterative optimization and correction on the evolved data based on real-time incremental feedback data to generate a dissemination effect evaluation value; and mapping the dissemination effect evaluation value to correction instructions for the expression form, dissemination channels, and presentation rhythm of the cultural dissemination content.

[0086] In one embodiment of the present invention, see [reference] Figure 2 Taking a digital tweet about the architectural culture of the Forbidden City as an example, the original cultural dissemination content includes text introducing the roof ridge beasts of the Hall of Supreme Harmony, a front view photo of the Hall of Supreme Harmony, and an audio clip with background music from the chime bells. The content scanning unit initiates and deconstructs the digital tweet, separating structured metadata such as the posting time and the poster's account from the tweet's metadata. Simultaneously, the tweet's text, the image file of the Hall of Supreme Harmony, and the background music audio file are output as unstructured content data. The cultural knowledge graph interface is invoked, loading concept nodes, attribute relationships, and event templates related to "Forbidden City," "Hall of Supreme Harmony," "roof ridge beasts," and "chime bells" from a remote cultural knowledge graph server, forming a subgraph of cultural knowledge graph data containing these nodes and their relationships.

[0087] In the specific implementation, for the deconstructed text content "Ten ridge beasts are arranged on the roof ridge of the Hall of Supreme Harmony, led by an immortal riding a phoenix," semantic slicing and dependency analysis identified "Hall of Supreme Harmony," "ridge beasts," and "immortal riding a phoenix" as text symbols. For the deconstructed image file of the Hall of Supreme Harmony, visual object recognition and scene classification identified "hipped roof," "glazed tiles," and "ridge beast sculpture" as visual elements. For the deconstructed background music audio file, voiceprint feature extraction and melody analysis identified an audio fragment with typical harmonic characteristics of bronze musical instruments as a traditional melody fragment. These identified text symbols, visual elements, and audio fragments were then matched with the subgraph data of the cultural knowledge graph for similarity, and the cosine distance between their respective feature vectors and the node vectors of "Hall of Supreme Harmony," "ridge beasts," "immortal riding a chicken," and "chime bell music" in the subgraph data was calculated. It can be understood that the feature vectors were extracted by a trained deep learning model.

[0088] In practice, all data items with attached cultural connotation category tags, sentiment analysis tags, and historical period tags are arranged and combined according to their text, image, and audio content types and the order in which they appear in the tweet. A record about the visual element of "ridge beast sculpture" contains the following in the feature set list: content identifier "Tweet_001_Image_1", element content "ridge beast sculpture image feature vector", cultural connotation category "architectural decoration", sentiment tag "solemn", historical period tag "Ming and Qing Dynasties", timestamp "2023-10-01 10:00:00", and spatial location information "Beijing" as resolved from the IP address of the tweet publisher. A record about the audio clip of "traditional melody" contains the following: content identifier "Tweet_001_Audio_1", element content "Chime bell music audio feature vector", cultural connotation category "traditional musical instrument", sentiment tag "distant", historical period tag "pre-Qin", timestamp "2023-10-01 10:00:00", and spatial location information "Beijing". Optionally, for multiple elements from the same tweet, their timestamps and spatial location information can be obtained from the tweet's metadata. The final feature set is a list, with each record in the list having the above structure. In some embodiments, the feature set is stored in JSON format or a database table for easy access by subsequent processing programs. The similarity matching formula uses cosine similarity calculation, and its formula is... Represented as:

[0089] ;

[0090] Where: symbol Represents a feature vector extracted from text symbols, visual elements, or audio clips; symbols This represents the feature vector extracted from the corresponding concept node in the cultural knowledge graph subgraph data, with the symbol... Representative vector with vector The angle between them, symbol The dot product operation represents vectors, with the sign... and They represent vectors respectively sum vector The modulus is calculated. The result is a similarity value, and a threshold is set to determine whether a match is successful. Optionally, the feature vector needs to be normalized before calculation.

[0091] In one embodiment of the present invention, see [reference] Figure 3 In practice, audience feedback is collected from multiple independent channels to construct datasets, and feature sets are matched and overlaid with audience feedback datasets in a spatiotemporal grid. Multiple audience interaction data collection points are defined, including comment sections of cultural content publishing platforms, topic reposting chains on social media, sensors at offline cultural experience venues, and user-submitted feedback questionnaires. Taking an online live-streamed lecture on "Song Dynasty Tea-Making Techniques" and its related offline experiences as an example, the comment text, number of likes, and number of shares posted by the audience are collected from the comment section of the live-streaming platform. For instance, if a comment text "The demonstration of whisking the tea is very clear" is collected, sentiment analysis determines that the text is "positive," generating a platform interaction record containing this comment and its corresponding number of likes and shares. By analyzing the reposting chain of the topic "#Song Dynasty Tea Ceremony#" on social media, we used topic tracking technology to collect audience reposts, quoted content, and links to secondary creations. For example, we collected a reposted Weibo link with the comment "I want to try this technique." We analyzed the audience's dissemination path and content variation process to generate social dissemination records. We also collected data from sensors in museums offering tea ceremony experiences, including audience dwell time, movement patterns, and the number of times interactive exhibits were triggered. For example, we collected data on a visitor who stayed for 15 minutes and triggered the tea set touchscreen explanation three times, generating offline behavior records. We also conducted structured analysis on the feedback questionnaires distributed online regarding the live lecture, extracting scores and open-ended answers regarding cultural content comprehension, liking, and willingness to further disseminate the content.

[0092] A unified data timeline is established, arranging platform interaction records, social media dissemination records, offline behavior records, and questionnaire feedback records in chronological order. For example, comments 5 minutes after the live stream started, Weibo reposts at 10 minutes, museum experiences the next day, and questionnaire submissions on the third day are sorted by timestamps, and records from different sources within the same time window are associated. In some embodiments, the association is based on a common content identifier contained in the records. For example, if both platform comments and social media reposts contain the video ID "VID_20231001_SongTea" from the live lecture, these two types of records are associated within their corresponding time windows. Optionally, the association can also be based on audience anonymity identifiers. For example, the MAC address of an anonymous device collected by a museum Wi-Fi probe is matched with the same device identifier voluntarily submitted in the questionnaire system, thereby associating the offline behavior records and questionnaire feedback records of the same anonymous user. All records after time sorting and association are deduplicated and outlier filtered. Deduplication is performed to remove identical records based on their hash values. For example, only one of two identical forwarding records is kept. Outlier filtering is performed to remove records that are obviously illogical based on preset behavioral logic rules. For example, a platform interaction record that shows "a long comment was posted 1 hour before the live broadcast" is removed, forming a structured audience feedback dataset.

[0093] In practical implementation, a virtual spatiotemporal coordinate system is defined according to the needs of the evaluation task. The time dimension is divided into continuous equal-length time periods, and the spatial dimension is divided into continuous equal-area regions, such as a city's administrative division as a region. The spatiotemporal grid is the Cartesian product of the time and spatial dimensions. Each record in the feature set is mapped to the corresponding spatiotemporal grid based on its timestamp and spatial location information. For example, a visual feature record about "tea whisk" with a timestamp of "2023-10-01 14:00:00" (the start time of the live stream) and spatial location information of "Hangzhou City" (the registered location of the live stream account) is mapped to a specific spatiotemporal grid with a time index of "2023-10-01 14:00-15:00" and a spatial index of "Hangzhou City". If an element spans multiple spatiotemporal grids, it is segmented and weighted according to its duration and coverage, completing the spatiotemporal grid mapping of the element's feature set. For example, an audio element lasting 90 minutes, describing the general situation of tea culture in China, has a coverage spatial information of "nationwide." This element will be segmented and mapped to two consecutive 1-hour time slots, and a weight value calculated based on the city's population weight will be assigned to the spatial grid corresponding to each relevant city. It can be understood that the weight allocation calculation can be determined based on the theoretical influence coverage ratio of the element's content in each grid.

[0094] Each record in the audience feedback dataset is mapped to a corresponding spatiotemporal grid based on its timestamp and the implicit geographic location information of its source channel, thus completing the spatiotemporal grid mapping of the audience feedback dataset. For example, a comment posted from a Shanghai IP address 25 minutes after the start of the live stream is mapped to a spatiotemporal grid with a time index of "2023-10-01 14:00-15:00" and a spatial index of "Shanghai". A record of offline behavior from a museum in Beijing, with a timestamp of "2023-10-02 10:30", is mapped to a spatiotemporal grid with a time index of "2023-10-02 10:00-11:00" and a spatial index of "Beijing". For each spatiotemporal grid, it is checked whether there are mapped feature records and audience feedback records. If both mapped feature records and audience feedback records exist simultaneously in the spatiotemporal grid, a data overlay operation is performed. This operation cross-combines the cultural connotation category, emotional label, and historical period label of all feature records within the spatiotemporal grid with the emotional tendency, communication behavior type, and rating of all audience feedback records. For example, in the spatiotemporal grid for "2023-10-01 14:00-15:00, Hangzhou," there is a feature record for "tea ceremony" with a cultural connotation category of "traditional crafts," an emotional label of "elegance," and a historical period label of "Song Dynasty," and simultaneously, there is an audience feedback record with an emotional tendency of "positive" and a communication behavior type of "commentary." This generates a "feature-feedback" pairing. The frequency and chronological order of each pairing are recorded; for example, this pairing occurs 150 times in this grid. If the spatiotemporal grid contains only mapped feature records or only mapped audience feedback records, the spatiotemporal grid is marked as an incomplete data grid, and the data overlay operation is not performed. Collect all feature-feedback pairing data and their frequency and temporal order information generated by spatiotemporal grids that have undergone data overlay operations, and summarize them into overlay data. In some embodiments, the structure of the overlay data is a multi-level nested dictionary, where the key is the spatiotemporal grid coordinate and the value is a list of all feature-feedback pairing data within the spatiotemporal grid. Feature cross-grid segmentation and weight allocation can be calculated based on their spatiotemporal coverage and grid intersection ratio. An optional weight calculation formula is expressed as:

[0095] ;

[0096] Where: symbol Representative elements In the spacetime grid The weights assigned to the middle, the symbols Representative elements Duration interval, symbol Represents a spacetime grid The time interval represented, symbol The duration of the intersection between the duration of the representative element and the grid time interval, with the symbol... Representative elements The set of potential spatial coverage areas, symbols Represents a spacetime grid The spatial region represented, symbol This represents the intersection of the feature space coverage and the grid space region. It's understandable that the result of the intersection operation on the numerator needs to be measured to calculate the proportion.

[0097] In one embodiment of the present invention, the superimposed data is quantitatively scored and imported into a dissemination effect simulator for dynamic evolution calculation. Taking a digital performance of "Dunhuang Flying Apsaras Dance" as an example, the superimposed data is a multi-layered nested dictionary, with the key being the spatiotemporal grid coordinates and the value being a list of "element-feedback" paired data within the grid. For example, for the spatiotemporal grid coordinates "2023-11-2020:00-21:00, Xi'an City", its paired data list contains 20 records, one of which shows that the cultural connotation category of the element part is "traditional dance", the historical period label is "Tang Dynasty", the emotional label is "ethereal", the behavior type of the feedback part is "in-depth commentary", the emotional tendency intensity is 0.9, and the dissemination scope is "provincial level". Load a pre-built cultural value weight library, which is a table stored in a relational database. This library stores the value weight coefficients corresponding to different cultural connotations, historical periods, and emotional tones. For example, querying the table reveals a base weight of 1.2 for the "Traditional Dance" category, a period weight multiplier of 1.5 for the "Tang Dynasty" historical period, and an emotional weight multiplier of 1.1 for the "Elegant" emotional tone. It can be understood that the value weight coefficients are set based on expert evaluation and historical dissemination data analysis, such as by soliciting expert opinions through the Delphi method and combining them with regression analysis of historical dissemination data of similar content. Load pre-defined behavioral value judgment rules, which are a set of logical judgment statements stored in a configuration file. For example, a rule might be defined as: "IF Behavior type is 'In-depth commentary' AND Emotional tendency intensity is greater than 0.8 AND Dissemination scope is 'Provincial' THEN Feedback behavioral value score = 8."

[0098] The process iterates through the list of "element-feedback" pairings within each spatiotemporal grid of the overlaid data. For each pair, the cultural value weight library is first queried based on the cultural connotation category, historical period label, and emotional label of the element in the pairing data to obtain the corresponding basic weight of the cultural element. For example, for the element combination "traditional dance - Tang Dynasty - graceful" mentioned above, the basic weight obtained from the cultural value weight library might be calculated using a multiplicative model.

[0099] ;

[0100] Where: symbol Represents the basic weight of cultural elements, symbols The symbol represents the baseline weight (value 1.2) for the "Traditional Dance" category retrieved from the cultural value weight database. The period weight multiplier (value 1.5) representing the "Tang Dynasty" historical period, symbol [symbol missing]. The emotional weight multiplier (with a value of 1.1), representing the "ethereal" emotional tone, is used to calculate the basic weight of cultural elements. Secondly, based on the behavioral type, emotional intensity, and dissemination range of the feedback portion in the paired data, behavioral value judgment rules are applied to calculate the corresponding feedback behavioral value score. For example, the feedback behavioral value score is obtained by applying the aforementioned rules. The acquired cultural elements will have a basic weight. Feedback behavior value score Multiplying the results yields a single quantitative score for the paired "factor-feedback" data. The individual quantization scores of all "feature-feedback" paired data within a spatiotemporal grid are summed to obtain the total grid quantization score of the spatiotemporal grid.

[0101] In one embodiment of the present invention, the evolved data is iteratively optimized and corrected multiple times to generate a dissemination effect evaluation value. Continuing with the aforementioned example of the "Dunhuang Flying Apsaras Dance" digital performance, the dissemination effect simulator outputs the evolved data matrix. New audience feedback data generated since the last evaluation is acquired in real time from data collection points as incremental feedback data. For example, within 24 hours after the first evaluation, 50 reposts and 120 comments about the performance are newly collected from social media platforms, and 200 visitor dwell records are newly collected from sensors in offline museums. This new data constitutes incremental feedback data. The incremental feedback data undergoes the same data processing flow as that used to form the audience feedback dataset, including format deconstruction, sentiment analysis, and spatiotemporal mapping, to obtain the incremental spatiotemporal grid mapping result corresponding to the incremental feedback data. For example, after processing, new comment data is mapped to spatiotemporal grids such as "2023-11-21 10:00-11:00, Shanghai", and new offline behavior data is mapped to spatiotemporal grids such as "2023-11-21 14:00-15:00, Lanzhou". The incremental spatiotemporal grid mapping results are compared with the evolved data matrix to identify which spatiotemporal grid coordinates show new feedback mappings.

[0102] For grids in the evolved data matrix that have incremental spatiotemporal grid mapping results at the same spatiotemporal grid coordinates, optimization and correction calculations are performed. This optimization and correction calculation involves a weighted fusion of the evolved values ​​of the spatiotemporal grids and the quantitative scores of the incremental feedback. The weighting coefficient is dynamically adjusted based on the time freshness and source reliability of the incremental feedback data. Time freshness refers to the time difference between the data generation time and the current evaluation time, while source reliability is based on the authority of the data collection channel and a preset level of historical data quality. For example, the evolved value of the grid "2023-11-21 10:00-11:00, Shanghai" is 180.5. The incremental feedback corresponding to this grid, after quantitative scoring, yields a score of 25.3. If the calculated weighting coefficient is 0.2, the corrected value will be... For grids in the evolved data matrix that do not have incremental spatiotemporal grid mapping results at the same spatiotemporal grid coordinates, their original values ​​are kept unchanged.

[0103] The corrected data matrix is ​​used as the new evolved data matrix, and the steps from acquiring incremental feedback data in real time to obtaining the corrected data matrix are repeated for multiple iterations. For example, in the next 6 hours, the system automatically triggers a new incremental data acquisition and correction calculation process every 2 hours, for a total of 3 iterations. An iteration termination condition is set, including reaching the maximum number of iterations or the difference norm between the corrected data matrices generated by two consecutive iterations being less than a set tolerance. The difference norm can be understood as the square root of the sum of the squares of the differences between corresponding elements of two matrices. When the iteration termination condition is met, the iteration process stops. For example, if the system sets the maximum number of iterations to 5, after the 4th iteration, the difference norm between two consecutive matrices is less than the tolerance of 0.5, and the iteration stops. All grid values ​​in the corrected data matrix generated in the last iteration are globally normalized, mapping the normalized value range to the integer range of 0 to 100, generating a propagation effect evaluation value. In some embodiments, the global normalization uses a min-max normalization method, linearly transforming all values ​​in the matrix to the [0, 100] interval. The propagation effect evaluation value is an evaluation value matrix with the same dimensions as the spatiotemporal grid. The value at each position in the matrix represents the propagation effect score of the corresponding spatiotemporal region at the evaluation time. Refer to Table 1 for the evolved data, incremental feedback data, and correction calculation process of part of the spatiotemporal grid.

[0104] Table 1: Incremental Feedback and Correction Calculation Table

[0105]

[0106] In practical implementation, weighting coefficients The dynamic adjustment can be calculated based on a predefined function, and an optional calculation formula is expressed as:

[0107] ;

[0108] Where: symbol Represents the coefficients ultimately used for weighted fusion, with the symbol... This represents the time weighting factor calculated based on the time freshness of incremental feedback data, with the symbol [symbol missing]. The symbol represents the preset weight of time freshness in the total weighted coefficient. This represents the credibility weighting factor calculated based on the credibility of the source of the incremental feedback data, with the symbol [symbol missing]. This represents the pre-defined weight of source credibility in the total weighted coefficient, and usually has... This is understandable; it's a time-weighted factor. The value is positively correlated with data freshness; for example, the more recently the data was generated, the fresher it is. The higher the value, the greater the credibility weight factor. The value is positively correlated with the preset credibility level of the data source. The formula calculates... The values ​​will be constrained between 0 and 1 for subsequent weighted fusion calculations. Optionally, for grid G1 in Table 1, its incremental feedback data has high freshness, originating from authoritative social media platforms, and the calculated values ​​are... ,set up ,but However, in practical applications, 0.2 may be ultimately adopted based on the requirements of smoothing.

[0109] See Figure 4The study presents the distribution of propagation effect evaluation values ​​corresponding to different spatiotemporal grids. The time dimension is divided into four time periods: T1 (10:00-11:00), T2 (14:00-15:00), T3 (16:00-17:00), and T4 (19:00-20:00). The spatial dimension covers five regions: S1 (Shanghai), S2 (Beijing), S3 (Lanzhou), S4 (Dunhuang), and S5 (Guangzhou). The evaluation value range is mapped to a quantitative score of 0-100, and the color gradient (from dark red to dark green) corresponds to the change of evaluation value from low to high. In terms of specific distribution characteristics, high evaluation values ​​were concentrated in the S2 (Beijing)-T1 period (95.6), S4 (Dunhuang)-T2 period (97.3), and S4 (Dunhuang)-T4 period (100.0), which is related to the high degree of matching between the dissemination of Dunhuang cultural content in the geographically related areas. In contrast, the evaluation values ​​of the S3 (Lanzhou) region were generally low in all periods, and the evaluation value of S3 (Lanzhou) in the T4 period was even 0, reflecting the insufficient matching between the cultural dissemination elements and audience feedback in this region during the corresponding period. In addition, the evaluation values ​​of the same spatial region fluctuated significantly in different periods (e.g., the evaluation values ​​of S4 (Dunhuang) in the T1-T4 periods were 87.2, 97.3, 75.4, and 100.0), reflecting the dynamic evolution of the dissemination effect in the time dimension. The difference in evaluation values ​​of different spatial regions within the same period (e.g., the difference between S2 (95.6) and S3 (14.0) in the T1 period) reflects the difference in the regional adaptability of cultural dissemination.

[0110] In one embodiment of the present invention, in a specific implementation, the dissemination effect evaluation value is mapped to the correction instructions for the expression form, dissemination channel and presentation rhythm of cultural dissemination content, as well as the construction and updating of the strategy mapping rule base. A dissemination effect evaluation value matrix for a series of short videos on "Jiangnan Garden Landscape Art" is analyzed. The rows of the matrix correspond to the time dimension and the columns correspond to the spatial dimension. Inefficient grid areas with evaluation values ​​lower than a preset qualified threshold and efficient grid areas with evaluation values ​​significantly higher than the surrounding areas are identified. For example, the grid with an evaluation value of 45 in the "Third Release Cycle, North China" is identified as an inefficient grid area, while the grid with an evaluation value of 85 in the "First Release Cycle, East China" is identified as an efficient grid area, which is significantly higher than its adjacent grids.

[0111] For the identified inefficient grid areas, the cultural connotation category, emotional tag combination, and frequently occurring behavior type in the audience feedback dataset of the inefficient grid areas are extracted from the feature set of the inefficient grid areas. For example, for the inefficient grid area of ​​"third release cycle, North China", the extracted feature combination is the cultural connotation category "garden couplets", the emotional tag "subtle", the historical period tag "Ming and Qing", and the frequently occurring "video exit" in the audience feedback behavior type. The extracted features of inefficient areas are input into the strategy mapping rule base for querying. The strategy mapping rule base is a knowledge base stored in a graph database or relational database, which stores the mapping relationship from "feature combinations" to "correction suggestions". Based on the query results, targeted correction instructions are generated. If the rules stored in the strategy mapping rule base indicate that the mapping between the feature combination "garden couplets + subtlety + Ming and Qing Dynasties" and the behavior "video exit midway" is related to "the presentation is too obscure", then the presentation correction instruction "add visual interpretation or storytelling" is generated. If the query shows that the feature combination mapping is related to "mismatch between the dissemination channel and the audience", then the dissemination channel correction instruction "adjust to a younger or more professional platform" is generated. If the query shows that the feature combination mapping is related to "the presentation pace is too fast", then the presentation pace correction instruction "insert a phased review and interactive segment" is generated.

[0112] For the identified high-efficiency grid areas, feature combinations are also extracted. For example, the feature combination extracted from the high-efficiency grid area "East China in the first release cycle" is the cultural connotation category "Stacked Mountains and Waters," the emotional tag "leisurely," the historical period tag "Song, Ming, and Qing Dynasties," and the high-frequency audience feedback behavior "watched the entire video and forwarded it." The strategy mapping rule base is queried to generate correction instructions to "maintain or strengthen the current features." Specific instructions might be "maintain the existing narrative rhythm centered on the visual presentation of 'Stacked Mountains and Waters'" and "continue to release similar style content in East China." The correction instructions generated for all low-efficiency and high-efficiency grid areas are summarized, classified and deredundant according to instruction type, and the same or similar instructions are merged. For example, instructions from different low-efficiency grid areas but all pointing to "increase visual interpretation" are merged into one, forming the final set of instructions for correcting the cultural dissemination content presentation, the dissemination channel, and the presentation rhythm.

[0113] In practical use, each time the system generates and executes a correction instruction, an effect tracking cycle is initiated. For example, the system generated a correction instruction to "add visual interpretation" for the "Jiangnan Gardens" case, which was adopted by the content production team, and animated annotations of couplets were added to the subsequently released video. After the effect tracking cycle ends, the dissemination effect evaluation value of the relevant area is re-evaluated. If the evaluation value improves, the pairing of "feature combination - correction instruction" is used as positive feedback to enhance the weight of the corresponding rule in the policy mapping rule base; if the evaluation value decreases or remains unchanged, it is used as negative feedback to reduce the weight of the corresponding rule in the policy mapping rule base, or the corresponding rule is deleted under specific conditions. It can be understood that the weight adjustment can be based on a reinforcement learning framework, and an optional weight update formula is expressed as:

[0114] ;

[0115] Where: symbol Represents the updated weight of a rule in the policy mapping rule base, with the symbol... This represents the weight of the rule before it was updated, symbol [symbol]. The learning rate is a small positive constant, with the sign... This represents the effect value of the target area reassessment after the execution of the correction command, symbol [symbol missing]. This represents the benchmark of the average effect value of similar operations in history. The result of the formula calculation makes the current effect... Outperform historical benchmarks The weight of a rule increases when it is active, and decreases when it is passive.

[0116] See Figure 5In evaluating the dissemination effect of short videos showcasing the art of Jiangnan garden landscaping, a heatmap using a spatiotemporal grid visually presents the dissemination effect scores under different release cycles (time dimension) and regions (spatial dimension). Specifically, the time dimension is divided into four release cycles, and the spatial dimension covers East China, South China, North China, Northwest China, and Southwest China. The color and value of each grid correspond to the dissemination effect score (value range 40-90). Among them, high-efficiency regions (marked with blue boxes), such as "First Release Cycle - East China (85 points)" and "Second Release Cycle - South China (82 points)," have scores significantly higher than surrounding grids; low-efficiency regions (marked with red boxes), such as "Third Release Cycle - North China (45 points)" and "Fourth Release Cycle - North China (55 points)," have scores below the passing threshold. The numerical logic of the heatmap is based on the communication effect evaluation method: first, the initial results are generated by matching the spatiotemporal grid of element characteristics and audience feedback and quantifying the scores. After the evolution of the communication law model and incremental feedback optimization, the normalized evaluation values ​​are mapped to the spatiotemporal grid scores. Finally, the spatiotemporal distribution characteristics of efficient / inefficient areas are visualized in the form of a heatmap, providing an intuitive basis for subsequent correction of communication strategies.

[0117] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0118] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for evaluating the effectiveness of intelligent dissemination of traditional Chinese culture, characterized in that, The method includes: The original cultural transmission content is analyzed, and the elements containing cultural connotations, emotional coloring and historical background are extracted to construct a set of elements and features; We obtain direct interaction records and indirect behavioral trajectories of the audience regarding cultural dissemination content from multiple independent channels to form an audience feedback dataset; Within a pre-defined spatiotemporal grid, the feature set and the audience feedback dataset are matched and overlaid grid by grid to generate overlaid data. The superimposed data is quantitatively scored based on a pre-defined cultural value weight library and behavioral value judgment rules, and the initial quantitative results are output. The initial quantification results are imported into the propagation effect simulator. The simulator dynamically calculates the evolution of the initial quantification results based on a pre-stored cultural propagation law model, resulting in evolved data, including: Initialize the propagation effect simulator. The propagation effect simulator is a discrete event simulation environment based on agent modeling. The simulation environment is divided into a spatiotemporal grid with the same dimensions as the initial quantization result matrix. Each grid initializes an agent, and the state of the agent is set to the grid quantization total score of the corresponding grid. Load a pre-stored cultural transmission law model from the model library. The cultural transmission law model defines the rules for the change of agent state. The rules include: assimilation rule, the state values ​​of adjacent grid agents will affect each other, and agents with high state values ​​will spread their influence to agents with low state values; decay rule, the state value of each agent will decay naturally over time; mutation rule, when the state value of an agent exceeds a certain threshold, it will trigger a jump or a sudden drop in state. Set the simulation clock and iteration step size, and start the propagation effect simulator; Within each iteration step, the propagation effect simulator traverses all grid agents and calculates the state value of each agent in parallel at the next moment based on the assimilation rule, decay rule, and mutation rule in the cultural propagation law model. During the calculation process, the simulator records the trajectory of each agent's state value change, as well as the grid coordinates where the state value change exceeds a set range; The simulation stops when the simulation clock reaches the preset simulation termination time, or when the change in the state values ​​of all agents is less than the stability threshold for several consecutive iterations. Extract the final state values ​​of all grid agents when the simulation stops to form the evolved data matrix. The structure of the evolved data matrix is ​​consistent with the initial quantization result matrix, but the element values ​​have been updated according to the cultural transmission law model. The evolved data is optimized and corrected through multiple rounds of iterations. The optimization and correction are based on the incremental feedback data acquired in real time, and a propagation effect evaluation value is generated. The evaluation value of the dissemination effect is mapped to the correction instructions for the presentation form, dissemination channel and presentation pace of cultural dissemination content.

2. The method for evaluating the effectiveness of intelligent dissemination of traditional Chinese culture according to claim 1, characterized in that, The method for analyzing the original cultural transmission content, extracting element features containing cultural connotations, emotional connotations, and historical background, and constructing an element feature set includes: The content scanning unit is activated to deconstruct the text, images, audio, and video contained in the cultural dissemination content, separating unstructured content data and structured metadata. Call the cultural knowledge graph interface to load concept nodes, attribute relationships, and event templates related to traditional culture from the cultural knowledge graph, forming subgraph data of the cultural knowledge graph; For the deconstructed text content, semantic slicing and dependency analysis are used to identify specific cultural symbols, historical figures, and traditional craft terms mentioned in it; for the deconstructed image and video content, visual object recognition and scene classification are used to identify traditional artifacts, clothing patterns, and architectural style elements contained therein; for the deconstructed audio content, voiceprint feature extraction and melody analysis are used to identify traditional tunes, instrument timbres, and dialect speech fragments contained therein. The identified text symbols, visual elements, and audio segments are matched with the corresponding nodes in the subgraph data of the cultural knowledge graph. The similarity matching is calculated by calculating the cosine distance of the feature vectors. When the cosine distance is less than a set threshold, it is considered a successful match. Each successfully matched item is labeled with its cultural connotation category. Text symbols, visual elements, and audio clips labeled with cultural connotation categories are compiled and attached with sentiment analysis tags and historical period tags. The sentiment analysis tags are assigned based on the analysis of the context of the item corresponding to the current tag, and the historical period tags are assigned based on the query of the event time attribute in the cultural knowledge graph. All data items with attached cultural connotation category tags, sentiment analysis tags, and historical period tags are arranged and combined according to their content type and time sequence to form an element feature set. The element feature set is stored in the form of a list. Each record in the list includes content identifier, element content, cultural connotation category, sentiment tag, historical period tag, timestamp, and spatial location information.

3. The method for evaluating the effectiveness of intelligent dissemination of traditional Chinese culture according to claim 2, characterized in that, The method for obtaining direct interaction records and indirect behavioral trajectories of the audience regarding cultural dissemination content from multiple independent channels to form an audience feedback dataset includes: Define multiple audience interaction data collection points, including the comment section of cultural content publishing platforms, topic forwarding chains on social media, sensors in offline cultural experience venues, and feedback questionnaires actively submitted by users; We collect the text of comments, the number of likes, and the number of shares posted by the audience from the comment section of cultural content publishing platforms, and conduct sentiment analysis on the comment text to generate platform interaction records. By using topic tracking technology, we can collect audiences' forwarding comments, quoted content, and secondary creation links from the topic forwarding chain on social media, analyze the audience's dissemination path and content variation process, and generate social dissemination records. We can also collect audiences' dwell time, movement trajectory, and number of times interactive exhibits are triggered from sensors in offline cultural experience venues, and generate offline behavior records. The system performs structured analysis on user-submitted feedback questionnaires, extracts scores and open-ended answers regarding cultural content comprehension, liking, and willingness to further disseminate the content, and generates questionnaire feedback records. Establish a unified data timeline, arrange platform interaction records, social communication records, offline behavior records, and questionnaire feedback records in chronological order, and associate records from different sources within the same time window. The basis for association is the common content identifier or audience anonymous identifier contained in the records. All records after time sorting and association are deduplicated and outlier filtered. Deduplication is performed to remove identical records based on their hash values, while outlier filtering is performed to remove records that are obviously illogical based on preset behavioral logic rules, thus forming a structured audience feedback dataset.

4. The method for evaluating the effectiveness of intelligent dissemination of traditional Chinese culture according to claim 3, characterized in that, The method for matching and overlaying the feature set and audience feedback dataset grid by grid within a preset spatiotemporal grid to generate overlaid data includes: Based on the needs of the evaluation task, a virtual spatiotemporal coordinate system is defined, dividing the time dimension into continuous equal-length time periods and the spatial dimension into continuous equal-area regions. The spatiotemporal grid is the Cartesian product of the time dimension and the spatial dimension. Each record in the feature set is mapped to the corresponding spatiotemporal grid based on its timestamp and spatial location information. If a feature spans multiple spatiotemporal grids, it is segmented and weighted according to its duration and coverage to complete the spatiotemporal grid mapping of the feature set. Each record in the audience feedback dataset is mapped to a corresponding spatiotemporal grid based on its timestamp and the geographic location information implied by the source channel, thus completing the spatiotemporal grid mapping of the audience feedback dataset. For each spatiotemporal grid, it is checked whether there are mapped feature records and audience feedback records. If both mapped feature records and audience feedback records exist simultaneously in the spatiotemporal grid, a data overlay operation is performed. The data overlay operation involves cross-combining the cultural connotation categories, emotional tags, and historical period tags of all feature records in the spatiotemporal grid with the emotional tendencies, communication behavior types, and ratings of all audience feedback records in the spatiotemporal grid to generate a series of "feature-feedback" paired data, and recording the frequency and time order of each paired data. If the spatiotemporal grid contains only mapped feature records or only mapped audience feedback records, then the spatiotemporal grid is marked as an incomplete data grid and no data overlay operation is performed. Collect all "feature-feedback" pairing data and their frequency and time sequence information generated by the spatiotemporal grids that have performed data overlay operations, and summarize them into overlay data. The structure of the overlay data is a multi-level nested dictionary, where the key of the dictionary is the spatiotemporal grid coordinate and the value is a list of all "feature-feedback" pairing data within the spatiotemporal grid.

5. The method for evaluating the effectiveness of intelligent dissemination of traditional Chinese culture according to claim 4, characterized in that, The method for quantifying and scoring the superimposed data, based on a preset cultural value weight library and behavioral value judgment rules, and outputting the initial quantification result includes: Load the pre-built cultural value weight library. The cultural value weight library is a database that stores the value weight coefficients corresponding to different cultural connotations, different historical periods, and different emotional tones. The value weight coefficients are set based on expert evaluation and historical dissemination data analysis. Load predefined behavioral value judgment rules. A behavioral value judgment rule is a set of logical judgment statements used to determine the value contribution score corresponding to the feedback behavior based on the behavior type, emotional tendency intensity, and dissemination scope in the audience feedback record. Traverse the list of "feature-feedback" paired data within each spatiotemporal grid in the overlay data; For each pair of "element-feedback" pairings in the list, first, based on the cultural connotation category, historical period label, and emotional label of the element part in the pairing data, query the cultural value weight library to obtain the corresponding basic weight of the cultural element. Secondly, based on the behavioral type, emotional tendency intensity, and dissemination range of the feedback portion in the paired data, the behavioral value judgment rules are applied to calculate the corresponding feedback behavioral value score. The basic weights of the acquired cultural elements are multiplied by the value scores of the feedback behaviors to obtain a single quantitative score for the paired data of "elements-feedback". The individual quantization scores of all "feature-feedback" paired data within a spatiotemporal grid are summed to obtain the total grid quantization score of the spatiotemporal grid. The total grid quantization scores of all spatiotemporal grids are calculated and arranged in the order of spatiotemporal grid coordinates to form an initial quantization result matrix. The initial quantization result matrix is ​​a two-dimensional array, where the row index corresponds to the time dimension and the column index corresponds to the spatial dimension. The value of each array element is the total grid quantization score.

6. The method for evaluating the effectiveness of intelligent dissemination of traditional Chinese culture according to claim 5, characterized in that, The method for performing multiple rounds of iterative optimization and correction on the evolved data, based on real-time acquired incremental feedback data, and generating a propagation effect evaluation value includes: New audience feedback data generated since the last evaluation is acquired in real time from the data collection points and used as incremental feedback data. The same data processing flow as that used to form the audience feedback dataset is performed on the incremental feedback data, including format deconstruction, sentiment analysis, and spatiotemporal mapping, to obtain the incremental spatiotemporal grid mapping result corresponding to the incremental feedback data. The incremental spatiotemporal grid mapping result is then compared with the evolved data matrix. For grids in the evolved data matrix that have incremental spatiotemporal grid mapping results at the same spatiotemporal grid coordinates, optimization and correction calculations are performed. The optimization and correction calculations involve weighted fusion of the evolved values ​​of the spatiotemporal grids and the quantitative scores of the incremental feedback. The weighting coefficients are dynamically adjusted according to the time freshness and source credibility of the incremental feedback data. For grids in the evolved data matrix that do not have incremental spatiotemporal grid mapping results at the same spatiotemporal grid coordinates, their original values ​​remain unchanged. After completing one round of traversal and calculation of all grids, a corrected data matrix is ​​obtained; The data matrix after one correction is used as the new evolved data matrix. The steps from obtaining incremental feedback data in real time to obtaining the data matrix after one correction are repeated for multiple rounds of iteration. Set iteration termination conditions, including reaching the maximum number of iterations, or the difference norm between the corrected data matrices generated by two consecutive iterations being less than the set tolerance. When the iteration termination condition is met, the iteration process is stopped. All grid values ​​in the corrected data matrix generated in the last iteration are globally normalized. The normalized value range is mapped to the integer range of zero to one hundred to generate the propagation effect evaluation value. The propagation effect evaluation value is an evaluation value matrix with the same spatiotemporal grid dimension. The value of each position in the matrix represents the propagation effect score of the corresponding spatiotemporal region at the evaluation time.

7. The method for evaluating the effectiveness of intelligent dissemination of traditional Chinese culture according to claim 6, characterized in that, The method for mapping dissemination effect evaluation values ​​into correction instructions for the presentation format, dissemination channels, and presentation pace of cultural dissemination content includes: The propagation effect evaluation value matrix is ​​analyzed to identify inefficient grid areas with evaluation values ​​below a preset qualified threshold, and efficient grid areas with evaluation values ​​significantly higher than the surrounding areas. For the identified inefficient grid areas, extract the cultural connotation category, emotional tag combination, and frequently occurring behavior type in the audience feedback dataset of the inefficient grid areas in the feature set corresponding to the inefficient grid areas; The extracted features of inefficient regions are input into the policy mapping rule base for querying. The policy mapping rule base stores the mapping relationship from "feature combination" to "correction suggestion". Based on the query results, generate targeted correction instructions. If the inefficient area characteristic corresponds to "the presentation is too obscure", then generate a presentation correction instruction to "add visual interpretation or storytelling". If the inefficient area characteristic corresponds to "the audience of the communication channel does not match", then generate a communication channel correction instruction to "adjust to a younger or more professional platform". If the inefficient area characteristic corresponds to "the presentation pace is too fast", then generate a presentation pace correction instruction to "insert a phased review and interactive segment". For the identified high-efficiency grid regions, their feature combinations are also extracted, the policy mapping rule base is queried, and correction instructions to "maintain or enhance the current features" are generated. The correction instructions generated for all inefficient and efficient grid areas are compiled, classified and de-redundanted according to instruction type, and the same or similar instructions are merged to form the final set of instructions for correcting the presentation form of cultural dissemination content, the set of instructions for correcting dissemination channels, and the set of instructions for correcting the presentation rhythm.

8. The method for evaluating the effectiveness of intelligent dissemination of traditional Chinese culture according to claim 7, characterized in that, The method for constructing and updating the policy mapping rule base includes: Collect historical successful and unsuccessful cultural communication cases, conduct feature analysis on each case, and extract the characteristics of the cultural content's presentation form, the characteristics of the communication channels used, the characteristics of the content presentation rhythm, and the grid evaluation value pattern corresponding to the case in the final evaluation. Cultural communication experts were invited to review the cases, and suggestions for improvement were marked for each case's combination of features. These suggestions included specific optimization recommendations for the form of expression, dissemination channels, and presentation rhythm. The feature combination of the case is paired with the correction suggestions of the expert annotation to form an initial mapping rule, and all the initial mapping rules are stored in the initial strategy mapping rule library. In actual use, an effect tracking cycle is started whenever the system generates a correction command and executes it; After the effect tracking period ends, the propagation effect evaluation value of the relevant area is re-evaluated. If the evaluation value improves, the pairing of "feature combination - correction instruction" is used as positive feedback to enhance the weight of the corresponding rule in the strategy mapping rule base. If the evaluation value decreases or remains unchanged, it is used as negative feedback to reduce the weight of the corresponding rule in the strategy mapping rule base, or the corresponding rule is deleted under specific conditions. New successful models and suggestions are regularly extracted from publicly available cultural communication research materials and industry reports. After manual or automatic review, they are added to the strategy mapping rule base in the form of new rules to achieve continuous updates to the strategy mapping rule base.

9. A system for evaluating the effectiveness of intelligent dissemination of traditional Chinese culture, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for evaluating the intelligent dissemination effect of traditional Chinese culture as described in any one of claims 1 to 8.

Images