Artificial Intelligence-Based Management Method and System for Displaying Cultural and Creative Works

Abstract

This application relates to the field of artificial intelligence technology, and provides a management method and system for the display of cultural and creative works based on artificial intelligence. The method includes: obtaining a cultural and creative work database, which contains multiple works and their recommended interest user portraits; establishing an exhibitor user portrait; identifying the common characteristics of the two, calculating the recommendation probability; extracting high-probability works to form a portfolio, and clustering and analyzing the similarity; eliminating duplicate works based on the similarity to optimize the portfolio; obtaining a display location map and counting the quantity; if the locations are insufficient, assigning values according to the recommendation probability and randomly selecting works; and managing the selected works through a display module within the limit of the display locations. This application solves the technical problems in the prior art that the display is affected by subjective factors and there is no precise recommendation, resulting in poor display effects, and realizes the technical effects of automatically screening, recommending, and displaying and managing cultural and creative works through artificial intelligence technology, improving the display effect and management efficiency.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and specifically relates to a cultural and creative works display management method and system based on artificial intelligence. Background Art

[0002] With the booming development of the cultural and creative industry and the rapid growth of the number of cultural and creative works, the traditional display management methods have been difficult to meet this large-scale and high-efficiency display demand. Those methods relying on manual screening and arrangement are not only inefficient, but also difficult to cover all works. Moreover, affected by subjective factors, many excellent cultural and creative works cannot be fully displayed and recognized. At the same time, as the aesthetic and interests of the audience towards cultural and creative works are constantly changing, their requirements for display management are also getting higher and higher. The audience hopes to see more works that meet their own preferences and needs, and requires that the display management can more accurately grasp the needs and preferences of the audience and provide a personalized display experience. Summary of the Invention

[0003] This application provides a cultural and creative works display management method and system based on artificial intelligence, aiming to solve the technical problems in the prior art that the display is affected by subjective factors and no accurate recommendation is made, resulting in poor display effects.

[0004] In view of the above problems, this application provides a cultural and creative works display management method and system based on artificial intelligence.

[0005] In a first aspect disclosed by this application, a cultural and creative works display management method based on artificial intelligence is provided. The method includes: obtaining a cultural and creative works database, where the cultural and creative works database includes multiple cultural and creative works, and establishing multiple recommended interest user portraits for the multiple cultural and creative works; establishing an exhibitor user portrait for the exhibitor users; performing common feature recognition on the exhibitor user portrait and the multiple recommended interest user portraits to obtain multiple recommended probabilities for the multiple cultural and creative works; extracting a first collection of works with recommended probabilities greater than a first preset probability, and performing work similarity clustering on the first collection of works to generate a similar work distribution index; based on the similar work distribution index, performing similar work elimination analysis, and deleting the eliminated works from the first collection of works to obtain a first optimized collection of works; obtaining a display position distribution map of the display module, and counting the number of display positions; determining whether the number of display positions is less than or equal to the number of the first works in the first optimized collection of works. If so, performing recommended probability assignment on the first optimized collection of works based on the multiple recommended probabilities to obtain a first display probability set; taking the number of display positions as a constraint, randomly extracting from the first optimized collection of works based on the first display probability set, and performing display management on the extracted works through the display module.

[0006] Another aspect disclosed in this application provides an AI-based cultural and creative works display management system, which includes: a database acquisition component for acquiring a cultural and creative works database, where the cultural and creative works database includes multiple cultural and creative works and establishes multiple recommended interest user portraits for the multiple cultural and creative works; a portrait establishment component for establishing an exhibition user portrait of the exhibition users; a feature recognition component for identifying common features between the exhibition user portrait and the multiple recommended interest user portraits to obtain multiple recommended probabilities for the multiple cultural and creative works; a similarity clustering component for extracting a first collection of works with recommended probabilities greater than a first preset probability and performing work similarity clustering on the first collection of works to generate a similar work distribution index; an elimination analysis component for performing similar work elimination analysis based on the similar work distribution index and deleting the eliminated works from the first collection of works to obtain a first optimized collection of works; a quantity statistics component for obtaining a display position distribution map of the display module and counting the number of display positions; a probability assignment component for determining whether the number of display positions is less than or equal to the number of the first works in the first optimized collection of works. If so, based on the multiple recommended probabilities, assign recommended probabilities to the first optimized collection of works to obtain a first display probability set; a random extraction component for randomly extracting the first optimized collection of works based on the first display probability set with the number of display positions as a constraint, and performing display management on the extracted works through the display module.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] The above AI-based cultural and creative works display management method can understand the audience characteristics of each multiple works by obtaining a cultural and creative works database and establishing recommended interest user portraits. Subsequently, according to the portraits of the exhibition users, identify the common features with the recommended interest user portraits, so as to calculate the recommended probabilities of each multiple works for the exhibition users. Then, screen out the works with higher recommended probabilities and perform similarity clustering analysis to avoid displaying overly similar works. Then, according to the available positions of the display module, intelligently extract the screened works to ensure that the most attractive works are displayed within the limited positions. This method not only improves the display efficiency, but also better meets the personalized needs of the audience and enhances the visiting experience.

[0009] The above description is only an overview of the technical solutions of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented in accordance with the content of the specification. And in order to make the above and other purposes, features and advantages of this application more obvious and understandable, the specific embodiments of this application are hereinafter specifically exemplified. Brief Description of the Drawings

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0011] Figure 1 It is a schematic flow chart of a method for displaying and managing cultural and creative works based on artificial intelligence in an embodiment;

[0012] Figure 2 It is an architecture diagram of a system for displaying and managing cultural and creative works based on artificial intelligence in an embodiment.

[0013] Description of the reference numerals: Database acquisition component 1, portrait establishment component 2, feature recognition component 3, similarity clustering component 4, elimination analysis component 5, quantity statistics component 6, probability assignment component 7, random extraction component 8. Detailed Embodiments

[0014] By providing a method and system for displaying and managing cultural and creative works based on artificial intelligence in the embodiments of the present application, the technical problems in the prior art that the display is affected by subjective factors and no accurate recommendation is made, resulting in poor display effects are solved.

[0015] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts belong to the scope of protection of the present application.

[0016] It should be noted that the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units does not necessarily have to be limited to those clearly listed steps or units, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products, or devices.

[0017] Embodiment 1

[0018] As Figure 1 shown, the present application provides a method for displaying and managing cultural and creative works based on artificial intelligence, and the method includes:

[0019] Obtain a cultural and creative works database, where the cultural and creative works database includes multiple cultural and creative works, and establish multiple recommended interest user portraits for the multiple cultural and creative works;

[0020] With the rapid development of the cultural and creative industry, the number of cultural and creative works has increased explosively. Traditional display management methods, such as manual screening and arrangement, are difficult to cope with this large-scale display demand. Therefore, introducing artificial intelligence technology can achieve automated and intelligent management of cultural and creative works, greatly improving the display efficiency and making accurate recommendations.

[0021] In the embodiment of the present application, during the process of cultural and creative works display management, the system terminal first obtains a cultural and creative works database, which contains information on numerous cultural and creative works. Subsequently, in order to recommend these works more accurately, the system terminal collects the historical display records of multiple cultural and creative works and conducts interest user identification. After that, user characteristics are extracted based on the identification results, and then similar characteristics are merged based on these characteristics to construct multiple recommended interest user portraits. These portraits can reflect which types of users may be interested in the works. Through this step, works that are more likely to attract the attention of the audience can be selected according to the recommended interest user portraits. In this way, the most attractive works can be selected for display from the limited display resources, thereby optimizing the viewing experience of the audience and laying a solid foundation for subsequent display management.

[0022] Furthermore, the present application provides a method for establishing multiple recommended interest user portraits for the multiple cultural and creative works, including:

[0023] Collect the historical display records of the multiple cultural and creative works for interest user identification to obtain multiple interest user sets;

[0024] Extract user characteristics from the multiple interest user sets to obtain multiple user characteristic sets;

[0025] Perform centralized similar feature merging on the multiple user characteristic sets to obtain the multiple recommended interest user portraits.

[0026] Preferably, when preparing the display of cultural and creative works, the system terminal first collects and analyzes the historical display records of multiple cultural and creative works. By analyzing these records, it is possible to identify the users who are interested in which works, and then obtain multiple sets of interested users. Subsequently, the characteristics of these interested users, such as age, gender, occupation, education level, etc., are further extracted to form multiple user characteristic sets. These characteristic sets provide the system terminal with detailed information about which user groups may like which works. After that, the system terminal performs similarity comparison between each pair to generate multiple similarities. Then, cluster analysis is performed on the similarities that meet the conditions to obtain multiple recommended interested user portraits. These portraits help the system terminal more accurately understand which users may be interested in which cultural and creative works, providing strong support for subsequent display management.

[0027] Furthermore, the present application provides a method for centrally merging similar features of the multiple user characteristic sets, and the method further includes:

[0028] Extract a first user characteristic set from the multiple user characteristic sets, and perform similarity comparison on any two user characteristics in the first user characteristic set to generate multiple first feature similarities;

[0029] Cluster the user characteristics with first feature similarities greater than a preset similarity to obtain multiple clustered characteristic sets;

[0030] Optionally, after obtaining multiple user characteristic sets, the system terminal randomly selects one user characteristic set as the first user characteristic set. Subsequently, similarity comparison is performed on any two user characteristics in the characteristic set. The system terminal selects a suitable similarity comparison method according to the type of the characteristics. For example, for numerical characteristics such as age, the Euclidean distance is used to calculate the similarity; for categorical characteristics such as gender or occupation, the chi-square test is used for similarity comparison. Then, all feature pairs in the first user characteristic set are traversed, and the similarity comparison operation is performed on each pair of user characteristics. This involves inputting the values of the two characteristics into the selected similarity comparison method to obtain the similarity value between them. For each pair of characteristics, a similarity value is generated according to the comparison result. This value reflects the similarity degree between the two characteristics, ranging from 0 to 1. Then, all calculated similarity values are recorded to generate multiple first feature similarities. These first feature similarities will serve as the data basis for subsequent analysis.

[0031] After obtaining multiple first feature similarities, the system terminal sets a preset similarity as a threshold. This threshold is used to screen out those user feature pairs with high enough similarities for subsequent clustering operations. Subsequently, all the calculated first feature similarities are traversed, and those user feature pairs with similarities greater than the preset similarity are screened out. These user feature pairs will be regarded as highly similar and suitable for clustering. After that, the screened user feature pairs are converted into numerical vectors suitable for processing by the K-means algorithm, and each vector represents a user feature. Then, according to the business requirements and the distribution of the data, a suitable value of K is determined. Then, K user feature vectors are randomly selected as the initial clustering centers. These clustering centers will serve as the starting point for iteration. Further, the iterative process of the K-means algorithm is started. In each iteration, each user feature vector is assigned to the nearest clustering center, and then the center position of each cluster is recalculated. This process is repeated continuously until the clustering centers no longer change significantly. After clustering, according to the cluster to which each user feature vector belongs, they are grouped into different cluster feature sets. Each cluster feature set represents a user group with similar user features. Through such processing, the system terminal can more clearly understand the features and differences between different user groups, providing more accurate data support for the establishment of subsequent recommended interest user portraits.

[0032] Calculate the clustering metrics of multiple cluster feature sets, merge the features of the cluster feature sets with clustering metrics greater than the preset metrics, obtain the first recommended interest user portrait and add it to the multiple recommended interest user portraits, where the clustering metrics include the clustering quantity metric and the clustering distance metric.

[0033] Optionally, after obtaining multiple clustering feature sets, the system terminal evaluates the quality and effectiveness of these clusters. This is achieved by calculating clustering metrics, including the clustering quantity metric and the clustering distance metric, which can reflect the number of clusters and the tightness of the internal features of the clusters. Specifically, the system terminal first calculates the number of features in each clustering feature set, and then adds these numbers to obtain the total number of features. Subsequently, the number of features in each clustering feature set is divided by the total number of features to obtain the clustering quantity metric for each clustering feature set. This metric reflects the proportion of each clustering feature set in the overall features. For each pair of clustering feature sets, the system terminal calculates the maximum distance between them as the clustering distance metric through the Euclidean distance. This maximum distance refers to the minimum similarity between the features in the two clustering feature sets. Then, according to the preset clustering metric threshold, the clustering feature sets with clustering metrics greater than the preset value are selected. These clustering feature sets have better clustering effects and representativeness, and can more accurately reflect the interests and preferences of users. For the selected clustering feature sets, a feature merging operation is performed. Feature merging is completed by taking the feature with the highest occurrence frequency in each clustering feature set. The merged feature set will form a comprehensive and more representative user profile. Then, based on the merged feature set, a first recommended interest user profile is generated. This profile will include the main interests, preferences, and behavioral characteristics of the user, and is used for subsequent recommendations and displays. Finally, the generated first recommended interest user profile is added to the existing multiple recommended interest user profiles. These profiles can more accurately reflect the interests and preferences of users, so they have a high recommendation value. Through the above process, the system terminal successfully uses clustering metrics to screen and merge clustering feature sets, and obtains recommended interest user profiles with higher recommendation value.

[0034] Further, the present application provides a method for establishing multiple recommended interest user profiles for the multiple cultural and creative works, and the method further includes:

[0035] Identifying the display frequency information of the historical display records of the multiple cultural and creative works;

[0036] Extracting the low-frequency cultural and creative works with display frequencies less than the preset frequency based on the display frequency information;

[0037] Optionally, when identifying the historical display records of multiple cultural and creative works, the system terminal pays attention to the display frequency information of each multiple works. These information reflect the number of times the works have been displayed in the past period, which is an important basis for evaluating the popularity and display effect of the works. Subsequently, based on these display frequency information, the low-frequency cultural and creative works with display frequency lower than the preset frequency are screened out. This preset frequency is set according to actual needs, aiming to identify those works that are relatively less displayed or receive less attention. Through this screening process, low-frequency cultural and creative works can be effectively identified, and these works may not have obtained sufficient display opportunities due to content, style or other reasons. Understanding the characteristics of these low-frequency works helps to better adjust the display strategy, improve their exposure and attention, and thus promote the diversity and richness of cultural and creative works. Generally speaking, identifying low-frequency cultural and creative works is a screening process based on the display frequency information in historical display records, aiming to discover those works that are less displayed and provide a basis for subsequent adjustment of the display strategy.

[0038] Mine the similar works of the low-frequency cultural and creative works to identify the user portraits of the similar works, and obtain the reference user portraits;

[0039] Compensate the recommended interest user portraits of the low-frequency cultural and creative works with the reference user portraits.

[0040] Optionally, in order to understand the audience characteristics of low-frequency cultural and creative works more deeply, the system terminal mines other works similar to these works. These similar works may have similar themes, styles or cultural connotations, so their audiences may also overlap with the potential users of low-frequency cultural and creative works. Subsequently, by identifying the user portraits of these similar works, a series of reference user portraits can be obtained. These reference user portraits describe the audience characteristics of the similar works, including their interests, preferences, behavior patterns, etc. After that, using these reference user portraits, the recommended interest user portraits of low-frequency cultural and creative works can be compensated. This means that the system terminal can combine the information of the reference user portraits to adjust or improve the recommended interest user portraits of low-frequency cultural and creative works to make them more accurate and comprehensive. In this way, when the system terminal recommends low-frequency cultural and creative works, it can more accurately find potential interested users and improve the accuracy and effectiveness of the recommendation. Generally speaking, by mining the similar works of low-frequency cultural and creative works and identifying the user portraits of these similar works, valuable reference information can be provided for the recommended interest user portraits of low-frequency cultural and creative works, thereby optimizing the recommendation strategy and enhancing the satisfaction and experience of users.

[0041] Establish the exhibitor user portraits of the exhibitors;

[0042] In one embodiment, the system terminal collects relevant data of the exhibitors, including their basic information, exhibition history, behavior records at the exhibition, such as the exhibition areas visited, stay time, interaction, etc., as well as feedback and opinions that may be collected through questionnaires or interviews. Subsequently, these data are sorted and analyzed to identify the common characteristics and differences of the exhibitors. For example, it is found that some users prefer certain types of cultural and creative products, or their activity patterns at the exhibition have certain regularities. Afterwards, based on these analysis results, user portraits of the exhibitors are constructed. These portraits will describe the characteristics of the users in detail, such as their interests, purchasing habits, and expectations for the exhibition. At the same time, the portraits will also reveal the differences between different user groups, helping the system terminal to understand the needs of each group more accurately. In summary, establishing the exhibitor user portrait of the exhibitor is a process of understanding and describing the characteristics and behaviors of the exhibitors by collecting and analyzing data. These portraits not only help to understand the users more deeply, but also provide strong support for the optimization and improvement of the exhibition.

[0043] Identify common features of the exhibitor user portrait and the multiple recommended interest user portraits to obtain multiple recommendation probabilities of the multiple cultural and creative works;

[0044] In one embodiment, in order to more accurately recommend cultural and creative works to exhibitors, the system terminal identifies common features of the exhibitor user portraits and multiple existing recommended interest user portraits. Common features refer to features that appear in different user portraits, which can reflect the similarities and common needs of user groups in certain aspects. By comparing and analyzing the common features of these user portraits, the potential connection between the exhibitor users and different recommended interest user groups can be understood. For example, it is found that some exhibitors and user portraits that like a certain type of cultural and creative works have a large overlap in age, occupation or interest, which suggests that these exhibitors may also be interested in this type of cultural and creative works. Based on the identification of these common features, the system terminal further calculates the recommendation probability of multiple cultural and creative works for the exhibitors. The recommendation probability is calculated based on the matching degree of the user portraits, which reflects the degree of fit between the works and the potential interests of the exhibitors. By comprehensively considering the similarity of the common features, a corresponding recommendation probability can be calculated for each cultural and creative work. Finally, multiple recommendation probabilities for multiple cultural and creative works are obtained, and these probabilities can help the system terminal more accurately determine which works are more likely to be favored by the exhibitors.

[0045] Furthermore, the present application provides for identifying common features of the exhibitor user portrait and the plurality of recommended interest user portraits to obtain a plurality of recommendation probabilities of the plurality of cultural and creative works, and the method further includes:

[0046] The exhibitor user portrait includes multiple user features of multiple users;

[0047] Compare the multiple user features with the multiple recommended interest user portraits, identify the proportion of the multiple user features falling into the multiple recommended interest user portraits, and generate the multiple recommendation probabilities.

[0048] Preferably, in order to provide more accurate cultural and creative work recommendations for the participating users, the system terminal compares these user features with multiple existing recommended interest user portraits. During the comparison process, each user feature in the participating user portrait is checked one by one to analyze whether a matching or similar feature can be found in the recommended interest user portrait. By counting the proportion of user features in the participating user portrait that fall into each recommended interest user portrait, it is possible to initially understand which recommended interest user groups the participating users are closer to. Subsequently, based on this proportion information, multiple recommendation probabilities are generated. The recommendation probability is calculated according to the degree of user feature matching, which reflects the degree of fit of the participating user to a specific recommended interest user portrait. If multiple user features of the participating user have a high degree of matching in a certain recommended interest user portrait, then the recommendation probability of this cultural and creative work for this participating user will be relatively high. After that, by comparing the participating user portrait with the recommended interest user portrait and calculating the corresponding recommendation probabilities, more personalized and accurate cultural and creative work recommendations can be provided for the participating users. These recommendation probabilities not only consider the degree of user feature matching but also combine the characteristics of the recommended interest user portraits, thus improving the accuracy and effectiveness of the recommendations. In summary, by comparing the user features in the participating user portrait and the recommended interest user portrait and calculating the recommendation probabilities, a more accurate and personalized cultural and creative work recommendation service can be provided for the participating users.

[0049] Extract a first portfolio with a recommendation probability greater than a first preset probability, and perform work similarity clustering on the first portfolio to generate a similar work distribution index;

[0050] In one embodiment, after generating the recommendation probabilities of multiple cultural and creative works, the system terminal sets a first preset probability, which is set according to the actual recommendation needs and is used to screen out the works that are most likely to be of interest to the exhibitors. By comparing the recommendation probability of each of the multiple works with the first preset probability, a collection of works with a recommendation probability greater than the preset probability can be extracted, namely, the first collection of works. Subsequently, in order to gain a deeper understanding of the relationship and distribution between the works in the first collection of works, the system terminal performs work similarity clustering. The system terminal calculates the similarity between each pair of works in the first collection of works through a work similarity analysis network. Afterwards, based on these similarity values, the works are clustered so that the works in the same cluster are relatively close in similarity, while the works in different clusters have relatively large differences. Through the work similarity clustering, similar work distribution indicators are generated. These indicators describe the distribution and clustering structure of the works in the first collection of works, which helps to understand which works are relatively close in style, theme, etc., so as to provide exhibitors with more diversified and differentiated recommendations.

[0051] Furthermore, the present application provides clustering the similarity of the first collection of works to generate a similar work distribution index, and the method further includes:

[0052] Enumerate and combine the works in the first collection of works in pairs to obtain multiple combinations of works;

[0053] Building a work similarity analysis network based on the twin network, performing similarity identification on the plurality of work combinations, and obtaining the similarities of the plurality of works;

[0054] Preferably, the system terminal enumerates and combines each work in the first set of works in pairs to form multiple work combinations. The purpose of doing this is to find all possible work pairings, so as to more comprehensively evaluate the similarity between works. Subsequently, in order to perform similarity analysis on these work combinations, the system terminal collects a large number of work pairs with similarity labels as training data and divides a part of them as the validation set. After that, two sub-networks with the same structure and shared parameters are designed to process the input work pairs respectively. Then, determine the structure of the sub-network, including convolutional layers (for image works), fully connected layers, recurrent neural networks (for text works), etc., to effectively extract the features of the works. Then, train the sub-network to learn how to extract meaningful features, which are crucial for distinguishing different works and judging the similarity between works. Further, design a module to calculate the similarity between the feature vectors extracted from the sub-network. This module has the Euclidean distance built in to calculate the similarity score between work pairs. Subsequently, use the training data with similarity labels to train the entire siamese network. And define a loss function to measure the difference between the similarity score output by the network and the true label. Adjust the parameters of the network through gradient descent to minimize the loss function, so that the network can better predict the similarity between works. After training is completed, the system terminal outputs these multiple work similarity analysis networks. Then, the system terminal inputs these work combinations into the constructed work similarity analysis network for similarity identification. The network will extract and compare the features of each pair of works and output the similarity value between them. These similarity values reflect the similarity degree of the works in terms of content, style, theme, etc., and provide important information about the relationship between works for the system terminal. Then, the system terminal obtains multiple work similarity values, which can help to further understand the similarity relationship between the works in the first set of works. By analyzing and processing these similarity values, more personalized and accurate work recommendations can be provided for the participating users, and basic data support for further recommendations and optimizations can be provided.

[0055] Cluster the works with work similarity greater than the work similarity threshold, and use the average value of the work similarity in the clustering result for work similarity identification;

[0056] Generate the similar work distribution index based on the clustering result and the work similarity identification.

[0057] Preferably, after analyzing the similarity of works, the system terminal obtains the similarity scores between each pair of works. To further understand and organize these works, clustering operations are performed using the method of obtaining multiple identical clustering feature sets described above, clustering works with higher similarity together to form different work clusters. During the clustering process, the system terminal sets multiple work similarity thresholds. This threshold determines which works should be classified into the same cluster. When the similarity scores of two or more works are higher than this threshold, they are regarded as similar works and assigned to the same cluster. After clustering, each cluster contains a group of works with higher similarity. To more intuitively represent the work similarity level of each cluster, the system terminal calculates the average value of the work similarity within each cluster and uses this average value as the work similarity identifier for that cluster. This identifier helps to quickly understand the overall similarity of the works in each cluster. Finally, based on these clustering results and work similarity identifiers, the system terminal generates similar work distribution metrics. These metrics reflect the distribution of different similarity levels in the works, such as how many clusters, the size of each cluster, and the work similarity identifier of each cluster. These metrics provide the system terminal with comprehensive and in-depth information about work similarity, helping to better understand the relationships and distributions between works and providing strong support for subsequent recommendation and display strategies.

[0058] Based on the described similar work distribution metrics, perform similar work elimination analysis, delete the eliminated works from the first work set to obtain a first optimized work set;

[0059] In one embodiment, based on the obtained similar work distribution metrics, the system terminal performs similar work elimination analysis. The purpose of this analysis is to ensure that the types of works displayed are rich enough while maintaining user interest and avoiding over-concentration on a certain type of work. During the elimination analysis, the system terminal examines the scale and work similarity identifier of each work cluster. For those clusters with too large a scale and too high work similarity, resulting in over-concentration of types, appropriate work elimination is carried out. Specifically, the system terminal formulates a set of elimination strategies based on the average value of work similarity in the cluster and the cluster size, and preferentially eliminates those works with high similarity but large duplication with the works in other clusters. Through this elimination analysis, some works are successfully deleted from the first work set, thus obtaining a more optimized first optimized work set with a richer type. This optimized work set not only retains the works that users are interested in but also largely avoids the simplification of work types, providing users with more diverse and interesting choices.

[0060] Obtain the display position distribution map of the display module and count the number of display positions;

[0061] In one embodiment, the system terminal collects specific information about the display module at various different positions. By organizing this data, a display position distribution map is obtained. This distribution map presents the distribution of the display module at each position. Through this map, it can be intuitively seen which positions have a larger number of display modules, which positions have a smaller number, and their relative relationships. At the same time, the number of display positions is also counted. This statistical result can help understand the overall scale of the display module and the specific number at each position. By comparing the display quantities at different positions, it is possible to further analyze whether the layout of the display module is reasonable and whether optimization and adjustment are needed. In summary, obtaining the display position distribution map of the display module and counting the number of display positions. These tasks not only help the system terminal understand the actual layout of the display module but also provide an important basis for subsequent optimization and adjustment. By reasonably utilizing this information, the display effect of the display module can be further improved, and the user experience can be enhanced.

[0062] Determine whether the number of display positions is less than or equal to the number of the first works in the first optimized work collection. If so, assign recommendation probabilities to the first optimized work collection based on the multiple recommendation probabilities to obtain a first display probability set;

[0063] In one embodiment, the number of display positions is compared with the number of works in the first optimized work collection. The purpose of this step is to determine whether the display positions are sufficient to accommodate all the optimized works. If the number of display positions is less than or equal to the number of works in the first optimized work collection, it means that each display position can correspond to one or more works for display. At this time, the system terminal extracts a second work collection with a probability greater than a second preset probability from the cultural and creative work database and repeats the similar work elimination analysis to increase the number of works until the number of display positions is less than the number of works in the optimized work collection. Subsequently, the system terminal assigns recommendation probabilities to each multiple works in the first optimized work collection based on the multiple recommendation probabilities calculated previously. This assignment process is determined according to the recommendation probabilities of each multiple works. The higher the recommendation probability of a work, the greater the chance it has when being displayed. Through this step, the system terminal obtains a first display probability set. Each element in this set corresponds to the display probability of one or more works in the first optimized work collection, which reflects the priority and possibility of this work when being displayed. In summary, when the number of display positions is sufficient or less than the number of works in the first optimized work collection, recommendation probabilities are assigned to them according to the recommendation probabilities of the works, obtaining a display probability set, which provides an important basis for subsequent work display.

[0064] Furthermore, the present application provides a method for determining whether the number of display positions is less than the number of the first works in the first optimized work collection, and the method further includes:

[0065] If the number of the display positions is greater than the number of the first works in the first optimized work collection, extract a second work collection with a probability greater than a second preset probability from the cultural and creative work database, where the second preset probability is less than the first preset probability;

[0066] Perform a similar work elimination analysis on the second work collection to obtain a second optimized work collection;

[0067] Determine whether the number of the display positions is less than the number of the second works in the second optimized work collection. If not, update the second preset probability according to a preset decreasing step length, and repeat the similar work elimination analysis until the number of the display positions is less than the number of the works in the optimized work collection.

[0068] Preferably, when the number of the display positions exceeds the number of the works in the first optimized work collection, the system terminal extracts the works with a recommendation probability greater than the second preset probability from the cultural and creative work database to form a second work collection. Here, the second preset probability is lower than the first preset probability, indicating that the system terminal relaxes the screening conditions to obtain more candidate works. Subsequently, a similar work elimination analysis is performed on the second work collection. The purpose of this step is to remove those works that are too similar or repetitive to the first optimized work collection, ensuring that the finally displayed works are both diverse and in line with the user's interests. Through the elimination analysis, a second optimized work collection is obtained. Then, it is determined again whether the remaining number of the display positions is still greater than the number of the works in the second optimized work collection. If there are still sufficient display positions, the screening conditions need to be further adjusted to expand the work scope. Specifically, the system terminal updates the second preset probability according to a preset decreasing step length and repeats the similar work elimination analysis. This process will continue until the number of the display positions is less than the number of the works in the optimized work collection. Generally speaking, when there are too many display positions, the work scope is gradually narrowed through steps such as extracting the second work collection, performing a similar work elimination analysis, and adjusting the screening conditions until the requirements of the display positions are met.

[0069] With the number of the display positions as a constraint, randomly extract from the first optimized work collection based on the first display probability set, and perform display management on the extracted works through a display module.

[0070] In one embodiment, based on the known number of display positions, the system terminal performs targeted work extraction and display management. First, according to the limitation of the display positions, that is, the number of display positions cannot be exceeded for work display. Therefore, the system terminal randomly extracts from the first optimized work collection according to the first display probability set. In this process, the chance of each multiple works being extracted is proportional to its probability value in the first display probability set, which means that works with a higher recommendation probability are more likely to be selected for display. After the extraction is completed, the display module is used to perform display management on the extracted works. This includes determining the positions, arrangement methods, and interaction functions of each multiple works on the display interface, ensuring that the works can be presented in a preset manner and that users can browse and interact conveniently.

[0071] In summary, the embodiments of the present application at least have the following technical effects:

[0072] The embodiments of the present application obtain a cultural and creative work database and establish recommended interest user portraits for multiple cultural and creative works. These portraits are based on the identification of interested users and the extraction of user characteristics from historical display records. For low-frequency cultural and creative works, the user portraits of similar works are mined for compensation. Subsequently, user portraits of exhibition participants are established and common feature identification is performed with the recommended interest user portraits to obtain the recommendation probability of each cultural and creative work. Then, works with a recommendation probability greater than the first preset probability are extracted to form a first work collection, and similarity clustering analysis is performed on it. A work similarity analysis network is constructed through a siamese network to compare works pairwise, identify similarities, and cluster works with high similarities to generate a similar work distribution index. Based on the similar work distribution index, elimination analysis is performed to optimize the first work collection. Then, the number of display positions of the display module is obtained. If it is less than or equal to the number of works in the optimized work collection, the work collection is assigned values based on the recommendation probability to form a display probability set. Constrained by the number of display positions, random extraction is performed on the work collection based on the display probability set, and work display management is performed through the display module. If the number of display positions is greater than the number of works in the optimized work collection, a second work collection with a recommendation probability greater than the second preset probability is extracted from the database, and similar work elimination analysis is performed to obtain a second optimized work collection. If there are still too many display positions, the second preset probability is reduced and this process is repeated until the display position requirements are met. These technical effects together solve the technical problems in the prior art that the display is affected by subjective factors and no precise recommendation is made, resulting in poor display effects, and achieve the technical effects of automatically screening, recommending, and displaying and managing cultural and creative works through artificial intelligence technology, improving the display effect and management efficiency.

[0073] Embodiment 2

[0074] Based on the same inventive concept as the method for displaying and managing cultural and creative works based on artificial intelligence in the foregoing embodiment, asFigure 2 As shown in the figure, the present application provides a cultural and creative work display management system based on artificial intelligence. The system includes:

[0075] Database acquisition component 1: The database acquisition component 1 is used to acquire a cultural and creative work database. Among them, the cultural and creative work database includes a plurality of cultural and creative works, and a plurality of recommended interest user portraits of the plurality of cultural and creative works are established;

[0076] Portrait establishment component 2: The portrait establishment component 2 is used to establish an exhibition user portrait of the exhibition users;

[0077] Feature recognition component 3: The feature recognition component 3 is used to perform common feature recognition on the exhibition user portrait and the plurality of recommended interest user portraits to obtain a plurality of recommended probabilities of the plurality of cultural and creative works;

[0078] Similarity clustering component 4: The similarity clustering component 4 is used to extract a first collection of works with a recommended probability greater than a first preset probability, and perform work similarity clustering on the first collection of works to generate a similar work distribution index;

[0079] Elimination analysis component 5: The elimination analysis component 5 is used to perform similar work elimination analysis based on the similar work distribution index, delete the eliminated works from the first collection of works to obtain a first optimized collection of works;

[0080] Quantity statistics component 6: The quantity statistics component 6 is used to obtain a display position distribution map of the display module and count the number of display positions;

[0081] Probability assignment component 7: The probability assignment component 7 is used to determine whether the number of display positions is less than or equal to the number of the first works in the first optimized collection of works. If so, based on the plurality of recommended probabilities, perform recommended probability assignment on the first optimized collection of works to obtain a first display probability set;

[0082] Random extraction component 8: The random extraction component 8 is used to perform random extraction on the first optimized collection of works based on the first display probability set with the number of display positions as a constraint, and perform display management on the extracted works through the display module.

[0083] Furthermore, the database acquisition component 1 is also used to execute the following methods:

[0084] Collect the historical display records of the plurality of cultural and creative works for interest user identification to obtain a plurality of interest user sets;

[0085] Extract user features from the plurality of interest user sets to obtain a plurality of user feature sets;

[0086] Perform centralized merging of similar features for the multiple user feature sets to obtain the multiple recommended interest user portraits.

[0087] Furthermore, the database acquisition component 1 is also used to execute the following method:

[0088] Extract the first user feature set from the multiple user feature sets, and compare the similarity of any two user features in the first user feature set to generate multiple first feature similarities;

[0089] Cluster the user features with first feature similarities greater than the preset similarity to obtain multiple clustered feature sets;

[0090] Calculate the clustering metrics of the multiple clustered feature sets, and merge the feature sets with clustering metrics greater than the preset metrics to obtain the first recommended interest user portrait and add it to the multiple recommended interest user portraits, where the clustering metrics include clustering quantity metrics and clustering distance metrics.

[0091] Furthermore, the database acquisition component 1 is also used to execute the following method:

[0092] Identify the display frequency information of the historical display records of the multiple cultural and creative works;

[0093] Extract the low-frequency cultural and creative works with display frequencies less than the preset frequency based on the display frequency information;

[0094] Mine the similar works of the low-frequency cultural and creative works to identify the user portraits of the similar works and obtain the reference user portraits;

[0095] Compensate the recommended interest user portraits of the low-frequency cultural and creative works with the reference user portraits.

[0096] Furthermore, the feature recognition component 3 is also used to execute the following method:

[0097] The participating user portrait contains multiple user features of multiple users;

[0098] Compare the multiple user features with the multiple recommended interest user portraits, and identify the proportion of the multiple user features falling into the multiple recommended interest user portraits to generate the multiple recommended probabilities.

[0099] Furthermore, the similarity clustering component 4 is also used to execute the following method:

[0100] Enumerate and combine the works in the first work set pairwise to obtain multiple work combinations;

[0101] Construct a work similarity analysis network based on the siamese network, and identify the similarities of the multiple work combinations to obtain multiple work similarities;

[0102] Cluster the works with a work similarity greater than the work similarity threshold, and use the average work similarity in the clustering result to identify the work similarity;

[0103] Generate the similar work distribution index based on the clustering result and the work similarity identification.

[0104] Furthermore, the probability assignment component 7 is also used to execute the following method:

[0105] If the number of display positions is greater than the number of first works in the first optimized work collection, extract a second work collection with a probability greater than a second preset probability from the cultural and creative work database, where the second preset probability is less than the first preset probability;

[0106] Conduct a similar work elimination analysis on the second work collection to obtain a second optimized work collection;

[0107] Judge whether the number of display positions is less than the number of second works in the second optimized work collection. If not, update the second preset probability according to a preset reduction step, and repeat the similar work elimination analysis until the number of display positions is less than the number of works in the optimized work collection.

[0108] It should be noted that the above order of the embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above specific embodiments of the present specification have been described. The processes depicted in the drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0109] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included within the protection scope of the present application.

[0110] This specification and the drawings are only exemplary descriptions of the present application and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. An artificial intelligence-based management method for displaying cultural and creative works, characterized in that, The method includes: Obtain a cultural and creative works database, where the cultural and creative works database includes multiple cultural and creative works, and establish multiple recommended interest user portraits for the multiple cultural and creative works; Establish an exhibitor user portrait for the exhibitor users; Identify common features between the exhibitor user portrait and the multiple recommended interest user portraits to obtain multiple recommended probabilities for the multiple cultural and creative works; Extract a first collection of works with a recommended probability greater than a first preset probability, and perform work similarity clustering on the first collection of works to generate a similar work distribution index; Based on the similar work distribution index, conduct similar work elimination analysis, delete the eliminated works from the first collection of works, and obtain a first optimized collection of works; Obtain a display position distribution map of the display module and count the number of display positions; Determine whether the number of display positions is less than or equal to the number of first works in the first optimized collection of works. If so, assign recommended probabilities to the first optimized collection of works based on the multiple recommended probabilities to obtain a first display probability set; With the number of display positions as a constraint, randomly select from the first optimized collection of works based on the first display probability set, and perform display management on the selected works through the display module; Among them, establishing multiple recommended interest user portraits for the multiple cultural and creative works includes: Collect historical display records of the multiple cultural and creative works for interest user identification to obtain multiple interest user sets; Extract user features from the multiple interest user sets to obtain multiple user feature sets; Perform centralized similar feature merging on the multiple user feature sets to obtain the multiple recommended interest user portraits; Among them, performing centralized similar feature merging on the multiple user feature sets includes: Extract a first user feature set from the multiple user feature sets, and compare the similarity of any two user features in the first user feature set to generate multiple first feature similarities; Cluster user features with a first feature similarity greater than a preset similarity to obtain multiple clustered feature sets; Calculate the clustering indicators of the multiple clustered feature sets, merge the clustered feature sets with a clustering indicator greater than a preset indicator, obtain a first recommended interest user portrait and add it to the multiple recommended interest user portraits, where the clustering indicator includes a clustering quantity indicator and a clustering distance indicator; Among them, establishing multiple recommended interest user portraits for the multiple cultural and creative works further includes: Identify the display frequency information of the historical display records of the multiple cultural and creative works; Extract low-frequency cultural and creative works with a display frequency less than a preset frequency based on the display frequency information; Mine similar works of the low-frequency cultural and creative works for user portrait identification of the similar works to obtain a reference user portrait; Compensate the recommended interest user portraits of the low-frequency cultural and creative works with the reference user portrait.

2. The method according to claim 1, wherein Determining whether the number of display positions is less than the number of first works in the first optimized collection of works further includes: If the number of display positions is greater than the number of first works in the first optimized collection of works, extract a second collection of works with a probability greater than a second preset probability from the cultural and creative works database, where the second preset probability is less than the first preset probability; Perform a similar work elimination analysis on the second work collection to obtain a second optimized work collection; Determine whether the number of display positions is less than the number of second works in the second optimized work collection. If not, update the second preset probability according to a preset reduction step, and repeat the similar work elimination analysis until the number of display positions is less than the number of works in the optimized work collection.

3. The method according to claim 1, wherein Perform common feature recognition on the participating user portrait and the multiple recommended interest user portraits to obtain multiple recommended probabilities for the multiple cultural and creative works, including: The participating user portrait includes multiple user features of multiple users; Compare the multiple user features with the multiple recommended interest user portraits, identify the proportion of the multiple user features falling into the multiple recommended interest user portraits, and generate the multiple recommended probabilities.

4. The method according to claim 1, characterized in that Perform work similarity clustering on the first work collection to generate a similar work distribution index, including: Enumerate and combine the works in the first work collection pairwise to obtain multiple work combinations; Construct a work similarity analysis network based on a siamese network, perform similarity recognition on the multiple work combinations, and obtain multiple work similarities; Cluster the works with work similarity greater than the work similarity threshold, and use the average value of the work similarity in the clustering result for work similarity identification; Generate the similar work distribution index based on the clustering result and the work similarity identification.

5. An AI-based cultural and creative work display management system, characterized in that, For implementing the steps of the artificial intelligence-based cultural and creative work display management method according to any one of claims 1 to 4, the system includes: Database acquisition component: Acquire a cultural and creative work database, where the cultural and creative work database includes multiple cultural and creative works, and establish multiple recommended interest user portraits for the multiple cultural and creative works; Portrait establishment component: Establish a participating user portrait of the participating user; Feature recognition component: Perform common feature recognition on the participating user portrait and the multiple recommended interest user portraits to obtain multiple recommended probabilities for the multiple cultural and creative works; Similarity clustering component: Extract a first work collection with a recommended probability greater than a first preset probability, and perform work similarity clustering on the first work collection to generate a similar work distribution index; Elimination analysis component: Based on the similar work distribution index, perform a similar work elimination analysis, delete the eliminated works from the first work collection to obtain a first optimized work collection; Quantity statistics component: Obtain the display position distribution map of the display module and count the number of display positions; Probability assignment component: Determine whether the number of display positions is less than or equal to the number of first works in the first optimized work collection. If so, assign recommended probabilities to the first optimized work collection based on the multiple recommended probabilities to obtain a first display probability set; Random extraction component: With the number of display positions as a constraint, randomly extract the first optimized work collection based on the first display probability set, and perform display management on the extracted works through the display module.

Images