Cultural product value evaluation and recommendation method and system based on multi-party collaboration

Abstract

The invention relates to a cultural product value evaluation and recommendation method and system based on multi-party cooperation, and is suitable for working scenes of cultural product value evaluation and analysis, cultural resource recommendation, cultural heritage protection and the like. The method comprises the following steps: constructing a data sharing network through an application programming interface or a data sharing protocol; data integration and privacy protection; in the three aspects of legal value, economic value and cultural value, the value of the cultural product is evaluated according to corresponding evaluation standards in combination with expert experience and a machine learning algorithm, and the weight of each dimension is dynamically adjusted; the method comprises the following steps: identifying interest points and preferences of a user through user behavior data, constructing a user interest map, converting the user interest map into low-dimensional vector representation, and generating a recommendation list by adopting a collaborative filtering algorithm. The invention provides an efficient, safe and explainable cultural product value evaluation and recommendation scheme for the digital cultural industry.

Description

Technical Field

[0001] The present invention relates to a method and system for evaluating and recommending the value of cultural products based on multi-party collaboration, which is applicable to application scenarios such as the evaluation and analysis of the value of cultural products, cultural resource recommendation, and cultural heritage protection. Background Art

[0002] At present, the methods for evaluating and recommending the value of cultural products have their own characteristics. The basic situation is as follows: In terms of value evaluation, books are often combined with sales volume, ratings, and the evaluations of professional book reviewers. Films and television works rely on the number of views, ratings, audience word-of-mouth, and the opinions of film critics. Artworks mostly refer to expert appraisals, auction prices, and market recognition. However, these methods generally have the problem of single dimension. Although data such as user comments and click-through rates are easy to obtain, they cannot reflect the deep value of cultural products. For example, for digital artworks, information such as artistic style, creation background, and text emotion are often ignored, resulting in the value judgment of cultural products only staying at the traffic level and unable to fully reflect their artistic value and ideological depth. At the recommendation level, it is usually based on the user's historical behavior data, and similar content is recommended through algorithms. However, this method is difficult to capture the change of user interests. For example, due to work needs, a user suddenly pays attention to another new field, and the recommendation algorithm often takes some time to adjust, resulting in the lag of recommended content, which makes the recommended content not match the user's current needs.

[0003] In addition, the data of cultural institutions such as publishing houses and museums are scattered and difficult to integrate. Under the traditional centralized storage method, user behavior data is easy to leak. Moreover, when the recommendation system processes cross-cultural content, it often ignores cultural differences, affecting the user experience. At the same time, existing algorithms are often like a kind of "black box" operation, and the recommendation results lack clear basis, which will not only reduce the user's trust, but also bring hidden dangers to the supervision of cultural products.

[0004] In the Chinese patent with the publication number CN114661982B and the name "Recommendation System Evaluation Method, Device, Electronic Device and Storage", a method for evaluating the diversity of a recommendation system is provided. The technical solution it provides focuses on coverage, personalization, novelty, and surprise, but the dynamic adjustment strategy is rigid, the update of diversity indicators depends on a fixed cycle, and it does not consider the user's real-time feedback. Only the historical detection cycle data is used to adjust the recommendation strategy, ignoring the user's immediate behavior. Summary of the Invention

[0005] The technical problem that the present invention attempts to solve is how to achieve an efficient, safe, real-time, and interpretable method and system for evaluating and recommending the value of cultural products through integrating multi-party data, intelligent analysis, and dynamic adjustment strategies.

[0006] To overcome the deficiencies in the prior art and solve the above-mentioned technical problems, the present invention provides a method and system for evaluating and recommending the value of cultural products based on multi-party collaboration, and adopts the following technical solutions.

[0007] In a first aspect, the present invention provides a method for evaluating and recommending the value of cultural products based on multi-party collaboration, the method comprising:

[0008] Step S100: Connect data sources of different cultural institutions through an application programming interface or a data sharing protocol;

[0009] Step S200: Encrypt user behavior data, product information, and spatio-temporal data, and transmit the encrypted feature data to a data sharing network;

[0010] Step S300: Evaluate the value of the cultural product in three aspects of legal value, economic value, and cultural value respectively according to corresponding evaluation criteria, combining expert experience and machine learning algorithms, and dynamically adjust the weights of each dimension;

[0011] Step S400: Identify the user's interest points through user behavior data, construct a user interest map, and convert the user interest map into a low-dimensional vector representation, and generate a recommendation list using a collaborative filtering algorithm.

[0012] Preferably, in the step S100, it includes: allocating network traffic through load balancing and accessing data through caching technology.

[0013] Preferably, in the step S100, it includes: preventing malicious attacks and data leakage through a firewall and an intrusion detection system, and encrypting data transmission through a Secure Sockets Layer / Transport Layer Security (SSL / TLS) protocol.

[0014] Preferably, in the step S300, it includes: optimizing model parameters through cross-validation and model tuning methods, and explaining the decision-making process of the model through model analysis.

[0015] Preferably, in the step S300, it includes: adjusting the weights of each dimension in real time through a dynamic adjustment mechanism according to market changes and cultural trends.

[0016] Preferably, in the step S400, it includes: optimizing the recommendation algorithm through reinforcement learning and continuous user feedback, and introducing an anti-bias mechanism.

[0017] Preferably, in the step S400, it includes: generating real-time recommendations by processing user behavior data in real time through stream processing.

[0018] Preferably, in the step S400, it includes: improving the personalization and accuracy of recommendations through matrix factorization and deep learning.

[0019] In a second aspect, the present invention provides a cultural product value evaluation and recommendation system based on multi-party collaboration, and the system includes:

[0020] Module M100, configured to: connect data sources of different cultural institutions through an application programming interface or a data sharing protocol;

[0021] Module M200, configured to: encrypt user behavior data, product information, and spatio-temporal data, and transmit the encrypted feature data to a data sharing network;

[0022] Module M300, configured to: evaluate the value of the cultural product in three aspects of legal value, economic value, and cultural value respectively according to corresponding evaluation criteria, and dynamically adjust the weights of each dimension by combining expert experience and machine learning algorithms;

[0023] Module M400, configured to: identify the user's interest points through user behavior data, construct a user interest map, convert the user interest map into a low-dimensional vector representation, and generate a recommendation list by using a collaborative filtering algorithm.

[0024] The cultural product value evaluation and recommendation method and system based on multi-party collaboration of the present invention provide an efficient, secure, and interpretable cultural product value evaluation and recommendation solution for the digital cultural industry by constructing a data sharing network, realizing data integration and privacy protection, performing value evaluation, and intelligent recommendation. This solution can not only improve the user experience but also promote the innovation and development of the cultural industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 : The step block diagram of the method of the present invention;

[0026] Figure 2 : The module block diagram of the system of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] To more clearly illustrate the features of the technical solution of the present invention, the present invention will be further elaborated in detail below through specific embodiments and in conjunction with the accompanying drawings.

[0028] As a first embodiment, the present invention provides a cultural product value evaluation and recommendation method based on multi-party collaboration, and the method includes:

[0029] Step S100: Connect data sources of different cultural institutions through an application programming interface or a data sharing protocol;

[0030] Step S200: Encrypt user behavior data, product information, and spatio-temporal data, and transmit the encrypted feature data to a data sharing network;

[0031] Step S300: Evaluate the value of the cultural product in three aspects: legal value, economic value, and cultural value. According to the corresponding evaluation criteria, combined with expert experience and machine learning algorithms, and dynamically adjust the weights of each dimension.

[0032] Step S400: Identify the user's interest points through user behavior data, construct a user interest map, convert the user interest map into a low-dimensional vector representation, and generate a recommendation list using a collaborative filtering algorithm.

[0033] Figure 1 The block diagrams of the steps in the above method embodiments are given.

[0034] Regarding the construction of the data sharing network in step S100, the following explanations are made:

[0035] In the rapid development of the digital cultural industry, data sharing has become one of the key links to improve service quality and optimize the user experience. In order to break data silos, the present invention first endeavors to construct an efficient and secure data sharing network. This network realizes seamless connection of data sources of different cultural institutions through application programming interfaces (APIs, Application Programming Interfaces) or data sharing agreements (DSAs, Data Sharing Agreements). These institutions include, but are not limited to, libraries, museums, art galleries, publishing houses, and film production companies, etc. By constructing a unified data sharing network, scattered user behavior (such as browsing, collecting, etc.) data, product information (such as texts, pictures, etc.), spatio-temporal data (such as access locations, times, etc.), etc. are integrated together, and only encrypted features are transmitted after local processing of user data, without revealing the original information.

[0036] In terms of technical implementation, a microservices architecture and containerization technology are adopted, such as the open-source platform Docker based on containerization technology TM and the open-source container orchestration platform Kubernetes TM , the scalability and elasticity of the system enable it to adapt to changes in the operating status, data scale, etc. of each cultural institution in real time and flexibly. Through service meshes, such as Istio TM or Linkerd TM etc., the communication between services can be managed and monitored to ensure the reliability and security of data transmission, enabling seamless connection of data sources of different cultural institutions and breaking data silos. In addition, in order to further improve the stability and fault tolerance of the system, a distributed storage technology is used to store and manage massive data. The purpose of using this technical means is to improve the stability and fault tolerance of the system, enabling the system to process massive data to continuously meet the needs of the industry.

[0037] Regarding the data integration and privacy protection in step S200, the following explanations are made:

[0038] Based on data sharing, in order to further achieve data integration and privacy protection, before transmitting data to the shared network, user behavior data, product information, spatio-temporal data, etc. are encrypted. For example, the encryption can be achieved by using the Advanced Encryption Standard (AES) and the Elliptic Curve Cryptography (ECC) algorithm. In this way, only the encrypted feature data is transmitted to the data sharing network, thus greatly reducing the risk of privacy leakage and protecting the security of user data.

[0039] Among the above: The Advanced Encryption Standard (AES), as a widely used symmetric encryption standard, has replaced the previous Data Encryption Standard (DES) because the security of the latter is no longer sufficient to meet the requirements of modern technology. The Elliptic Curve Cryptography (ECC) is a public-key encryption technology based on elliptic curve mathematics, which has the characteristics of short key length, high security, and high efficiency, and is widely used in modern encrypted communications.

[0040] At the same time, the security and integrity of the transmitted data are ensured in a blockchain manner. The immutability and decentralization characteristics of the blockchain make the data more secure during transmission and storage, and also guarantee the integrity of the data. Through smart contracts, fine-grained management of data access permissions can be realized to ensure that only authorized users and institutions can access the data.

[0041] To further improve the efficiency and quality of data integration, through data warehouse technologies such as Amazon Redshift TM and Google BigQuery TM , the centralized storage and management of data are realized. The purpose is to improve the efficiency and quality of data integration, provide data support for value evaluation and intelligent recommendation, and at the same time, combined with data mining and machine learning, such as decision trees and clustering algorithms, to discover patterns and associations in the data and provide data support for value evaluation and intelligent recommendation.

[0042] Regarding the value evaluation in step S300, the following explanations are made:

[0043] When evaluating the value of cultural products, in the three aspects of legal value, economic value, and cultural value, according to the corresponding evaluation criteria, combined with expert experience and machine learning algorithms, a comprehensive, objective, and accurate evaluation method for product value is provided, which will better reflect the value of cultural products. In addition, the evaluation dynamically adjusts the weights of each dimension to adapt to market changes and cultural trends.

[0044] In the evaluation criteria, the legal value tends to examine whether the copyright ownership is clear, such as whether the identity of the copyright owner and the publication time are clear; the economic value tends to predict the market potential, such as the sales volume of similar products; the cultural value tends to analyze the inheritance significance, such as whether it reflects intangible cultural heritage skills. Combining expert experience with machine learning, the weights of each dimension are dynamically adjusted. For example, when evaluating the value of cultural relics, more emphasis is placed on their cultural value.

[0045] The value evaluation process is as follows:

[0046] First, collect a large amount of cultural product data, including but not limited to copyright information, market performance, user evaluations, expert scores, etc.;

[0047] Then, use natural language processing (NLP) and text mining (Text Mining) techniques to extract valuable information from unstructured data. This structured processing process will improve the efficiency and accuracy of data processing, enabling the evaluation model to better capture the value information of products;

[0048] Finally, adopt machine learning algorithms, such as support vector machine (SVM) and random forest (Random Forest), to train the extracted features and build an evaluation model, aiming to provide an intelligent and automated evaluation method that can better adapt to market changes and cultural trends.

[0049] Regarding the intelligent recommendation in step S400, the following explanations are made:

[0050] After completing the value evaluation, the present invention needs to implement an intelligent recommendation function, that is, construct a user interest map, which identifies the user's interest points and preferences by analyzing the user's behavior data, such as browsing records, purchase history, and evaluations, enabling the user to better discover and obtain cultural products of interest.

[0051] Next, use graph embedding technology to transform the user interest map into a low-dimensional vector representation. The purpose of this move is to more conveniently calculate the similarity between the user's interests and product features, which will improve the efficiency and accuracy of recommendations and enable the system to better process massive user and product data.

[0052] On this basis, collaborative filtering algorithms such as user-based collaborative filtering (UserCF) and item-based collaborative filtering (ItemCF) are used to generate recommendation lists, which will improve the diversity and novelty of recommendations, enabling users to access a wider range of and more interesting cultural products.

[0053] As a preferred embodiment, in the step S100, it includes: distributing network traffic through load balancing and accessing data through caching technology.

[0054] In this preferred embodiment, in order to achieve the efficient operation of the data sharing network, load balancing technologies such as Nginx TM and HAProxy TM are used to distribute network traffic to ensure the stability and availability of the system. They are both load balancing tools focused on high performance, especially good at handling complex load balancing strategies and high-throughput scenarios. At the same time, caching technologies such as Redis or Memcached are used to improve data access speed and relieve the pressure on the database. They are two widely used in-memory data storage systems, both known for their high performance.

[0055] As a preferred embodiment, in the step S100, it includes: preventing malicious attacks and data leakage through firewalls and intrusion detection systems, and encrypting data transmission through the Secure Sockets Layer / Transport Layer Security (SSL / TLS) protocol.

[0056] In this preferred embodiment, in order to ensure the security of the data sharing network, firewalls and intrusion detection systems such as Snort and Suricata are used to prevent malicious attacks and data leakage. The Secure Sockets Layer (SSL) and Transport Layer Security (TLS) protocols are used to encrypt data transmission to ensure data security.

[0057] As a preferred embodiment, in the step S300, it includes: optimizing model parameters through cross-validation and model tuning methods, and explaining the decision-making process of the model through model analysis.

[0058] In this preferred embodiment, in order to improve the accuracy and reliability of the evaluation model, cross-validation and model tuning such as grid search and random search are used to optimize model parameters. At the same time, through model analysis such as feature importance analysis and model visualization, the decision-making process of the model is explained to improve the interpretability of the evaluation results.

[0059] As a preferred embodiment, in the step S300, it includes: adjusting each dimension weight in real time through a dynamic adjustment mechanism according to market changes and cultural trends.

[0060] In this preferred embodiment, by introducing a dynamic adjustment mechanism, the weights of each dimension are adjusted in real time according to market changes and cultural trends. In this way, the evaluation model can not only reflect the current value of the product but also predict its future development trend.

[0061] As a preferred embodiment, in the step S400, it includes: optimizing the recommendation algorithm through reinforcement learning and continuous user feedback, and introducing an anti-bias mechanism.

[0062] In this preferred embodiment, the recommendation algorithm is optimized through reinforcement learning and continuous user feedback to improve the quality and accuracy of recommendations. At the same time, an anti-bias mechanism is introduced to avoid over-recommending popular content to users.

[0063] As a preferred embodiment, in the step S400, it includes: through stream processing, real-time processing of user behavior data to generate real-time recommendations.

[0064] In this preferred embodiment, in order to achieve the real-time nature of intelligent recommendations, user behavior data is processed in real time through stream processing technologies such as Apache Kafka and Apache Flink to generate real-time recommendations.

[0065] As a preferred embodiment, in the step S400, it includes: through matrix factorization and deep learning, improving the personalization and accuracy of recommendations.

[0066] In this preferred embodiment, through personalized recommendations such as matrix factorization and deep learning, the personalization and accuracy of recommendations are improved.

[0067] As a second embodiment, the present invention provides a cultural product value evaluation and recommendation system based on multi-party collaboration. The system includes:

[0068] Module M100, used for: connecting data sources of different cultural institutions through application programming interfaces or data sharing protocols;

[0069] Module M200, used for: encrypting user behavior data, product information, and spatio-temporal data, and transmitting the encrypted feature data to the data sharing network;

[0070] Module M300, used for: evaluating the value of the cultural product in three aspects of legal value, economic value, and cultural value respectively according to the corresponding evaluation criteria, combining expert experience and machine learning algorithms, and dynamically adjusting the weights of each dimension;

[0071] Module M400, used for: identifying the user's interest points through user behavior data, constructing a user interest map, and converting the user interest map into a low-dimensional vector representation, and generating a recommendation list using a collaborative filtering algorithm.

[0072] Figure 2 The block diagrams of the respective modules in the above system embodiments are given.

[0073] Finally, it should be noted that although the present invention has been exemplarily described through specific embodiments, it does not constitute a limitation to the scope of patent protection of the present invention. Those skilled in the art should understand that various equivalent substitutions and optimization improvements can still be made to the specific embodiments of the present invention, and any substitution and improvement without departing from the spirit of the present invention should be covered within the scope of patent protection of the present invention.

Claims

1. A method for evaluating and recommending the value of cultural products based on multi-party collaboration, characterized in that, The method includes: Step S100: Connect data sources of different cultural institutions through an application programming interface or a data sharing protocol; Step S200: Encrypt user behavior data, product information, and spatio-temporal data, and transmit the encrypted feature data to a data sharing network; Step S300: Evaluate the value of the cultural product in three aspects of legal value, economic value, and cultural value respectively according to the corresponding evaluation criteria, combine expert experience and machine learning algorithms, and dynamically adjust the weights of each dimension; Step S400: Identify the user's interest points through user behavior data, construct a user interest graph, transform the user interest graph into a low-dimensional vector representation, and generate a recommendation list using a collaborative filtering algorithm.

2. The method according to claim 1, wherein In the step S100, it includes: Allocate network traffic through load balancing and access data through caching technology.

3. The method according to claim 1, characterized in that In the step S100, it includes: Prevent malicious attacks and data leakage through a firewall and an intrusion detection system, and encrypt data transmission through a Secure Sockets Layer / Transport Layer Security (SSL / TLS) protocol.

4. The method according to claim 1, wherein In the step S300, it includes: Optimize model parameters through cross-validation and model tuning methods, and explain the decision-making process of the model through model analysis.

5. The method according to claim 1, wherein In the step S300, it includes: Dynamically adjust the weights of each dimension in real time through a dynamic adjustment mechanism according to market changes and cultural trends.

6. The method according to claim 1, characterized in that In the step S400, it includes: Optimize the recommendation algorithm through reinforcement learning and continuous user feedback, and introduce an anti-bias mechanism.

7. The method according to claim 1, wherein In the step S400, it includes: Process user behavior data in real time through stream processing to generate real-time recommendations.

8. The method according to claim 1, wherein In the step S400, it includes: Improve the personalization and accuracy of recommendations through matrix factorization and deep learning.

9. A cultural product value evaluation and recommendation system based on multi-party collaboration, characterized in that, The system includes: Module M100, which is used to: Connect data sources of different cultural institutions through an application programming interface or a data sharing protocol; Module M200, which is used to: Encrypt user behavior data, product information, and spatio-temporal data, and transmit the encrypted feature data to a data sharing network; Module M300, which is used to: Evaluate the value of the cultural product in three aspects of legal value, economic value, and cultural value respectively according to the corresponding evaluation criteria, combine expert experience and machine learning algorithms, and dynamically adjust the weights of each dimension; Module M400, which is used to: Identify the user's interest points through user behavior data, construct a user interest graph, transform the user interest graph into a low-dimensional vector representation, and generate a recommendation list using a collaborative filtering algorithm.

Images