Personalized content recommendation method based on user behavior and preferences

By storing data in a decentralized manner on the blockchain and using zero-knowledge proof technology to build user profiles, combined with smart contracts and incentive mechanisms, the problem of user privacy protection in personalized recommendation systems is solved, achieving efficient and accurate personalized content recommendations and improving user trust and satisfaction.

CN118568359BActive Publication Date: 2024-11-19ZHUANGMEIHUI (SHANGHAI) INFORMATION TECHNOLOGY CO LTD
View PDF 9 Cites 0 Cited by

Patent Information

Application Number
CN202411055747.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-02
Publication Date
2024-11-19
Estimated Expiration
2044-08-02

AI Technical Summary

Technical Problem

While providing personalized services, existing personalized recommendation systems suffer from user privacy protection issues. The security and privacy of user data cannot be fully guaranteed, resulting in low user acceptance and trust in these systems.

Method used

Decentralized data is stored on the blockchain, user profiles are built using zero-knowledge proof cryptography, and recommendation logic is implemented through smart contracts to ensure the immutability and transparency of the data. At the same time, an incentive mechanism is designed to encourage users to contribute data and protect their privacy rights.

Benefits of technology

It improves the security and anonymity of user data, enhances user trust in the system, optimizes the accuracy and personalization of recommendation algorithms, increases user satisfaction and engagement, ensures the fairness and transparency of the incentive process, and protects user privacy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118568359B_ABST
    Figure CN118568359B_ABST
Patent Text Reader

Abstract

The present invention discloses a personalized content recommendation method based on user behavior and preference, and relates to the technical field of content recommendation. The present invention comprises the following steps: decentralized data storage: storing user data on a blockchain to ensure the data is tamperproof and transparent; firstly, performing a preliminary screening on the data to be stored; the present invention, by adopting a decentralized data storage method, stores user data on a blockchain, utilizes the tamperproof and transparency of the blockchain, significantly improves the security of the data, combines with privacy protection technologies such as zero-knowledge proof, ensures the anonymity of the user on the blockchain, and enhances the user's trust in the system. This trust is based on the encryption processing and secure storage of the data, as well as the fairness and transparency of the smart contract, and provides users with a safe and reliable personalized recommendation environment.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of content recommendation, and in particular to a personalized content recommendation method based on user behavior and preference. Background Art

[0002] With the rapid development of information technology, personalized recommendation systems have become a core component of online platforms such as e-commerce, social media, and content providers. These systems analyze users' historical behaviors and preferences to provide customized content or product recommendations to enhance user experience and satisfaction. However, existing personalized recommendation systems often need to collect and process a large amount of user data while providing personalized services, which raises the issue of user privacy protection. How to provide personalized recommendation services without infringing user privacy is an urgent problem to be solved;

[0003] Existing recommendation systems mostly use centralized data storage and processing methods, and the security and privacy of user data are difficult to fully guarantee. Users often lack sufficient understanding and control over how their data is collected, stored, and analyzed. In addition, the risk of data leakage and abuse also increases with the amount of data. Users' concerns about personal privacy limit their acceptance and trust in recommendation systems, which poses a major obstacle to the development and application of recommendation systems.

[0004] In order to solve the above defects, a technical solution is now provided. Summary of the invention

[0005] The purpose of the present invention is to solve the problem of how to provide efficient and accurate personalized content recommendation without infringing user privacy, and to propose a personalized content recommendation method based on user behavior and preference.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] The personalized content recommendation method based on user behavior and preference includes the following steps:

[0008] S1: Decentralized data storage: Store user data on the blockchain to ensure data immutability and transparency;

[0009] S2: Privacy-preserving user profile construction: Use zero-knowledge proof cryptography technology to construct a user profile that does not disclose the user's specific information;

[0010] S3: Smart contract recommendation logic: Use smart contracts to implement recommendation logic to ensure the fairness and transparency of the recommendation process;

[0011] S4: Encrypted analysis of user behavior: Encrypt user behavior data and analyze it only when the user authorizes it;

[0012] S5: Incentive mechanism design: Encourage users to contribute data while protecting their privacy rights through incentive mechanisms.

[0013] Furthermore, the specific steps of S1 are as follows:

[0014] Perform preliminary screening of user data to select data that meets preset standards;

[0015] Encrypt the filtered data to ensure the security of the data during transmission;

[0016] Split the encrypted data into multiple fragments to adapt to the storage structure of the blockchain;

[0017] Generate a hash value for each data fragment as a unique identifier of the data to ensure data integrity and traceability;

[0018] Write smart contracts to define the rules and conditions for data storage, and users guarantee data access rights and storage locations;

[0019] Upload data fragments and data fingerprints to the blockchain network through smart contracts to achieve decentralized storage of data;

[0020] Use multiple nodes in the blockchain network to verify the uploaded data to ensure data consistency and immutability;

[0021] Build an index for data stored on the blockchain;

[0022] Set data access permissions to ensure that only authorized users or systems can access data fragments;

[0023] Realize multi-node backup of data in the blockchain network;

[0024] Implement data update and maintenance rules through smart contracts;

[0025] Adopt zero-knowledge proof privacy protection technology to ensure the privacy of user data and maintain anonymity even on the blockchain.

[0026] Furthermore, the specific operation steps of performing preliminary screening of user data in S1 are as follows:

[0027] Determine the standards and rules for data screening based on application scenarios and business needs, and collect user data from various sources;

[0028] Clean the collected data, remove invalid, erroneous or incomplete data records, verify the accuracy and consistency of the data, ensure that the data meets the preset standards, and remove duplicate data records;

[0029] Classify data according to data type and purpose, and use data compression technology to reduce data storage space requirements.

[0030] Furthermore, the specific operation steps of S2 are as follows:

[0031] Determine the attributes required for user profiles, excluding sensitive information that can identify individuals;

[0032] Collect and clean user behavior data, and perform pre-processing to remove invalid or irrelevant information;

[0033] Anonymize user data to ensure that it does not contain information that can be traced back to personal identities;

[0034] Use zero-knowledge proof technology to prove user attribute groups and avoid leaking specific personal information;

[0035] Encrypt user portrait attributes;

[0036] Integrate encryption attributes to form encrypted user profiles for recommendation algorithms;

[0037] Use differential privacy to protect data analysis technology, analyze user portraits, and discover user group trends and preferences;

[0038] Verify and test user portraits to ensure accuracy and effectiveness of privacy protection;

[0039] Design regular update mechanisms to reflect the latest changes in user behavior and preferences while protecting privacy;

[0040] Implement access control mechanisms to ensure that only authorized systems and personnel have access to user profile data;

[0041] Conduct regular data access audits;

[0042] Obtain users’ explicit consent to collect and process data, and provide options to opt out or delete the data.

[0043] Furthermore, the user portrait is verified and tested in S2 to ensure accuracy and effectiveness of privacy protection. The specific process is as follows:

[0044] The metrics for measuring the accuracy of user portraits and the effectiveness of privacy protection include:

[0045] Consistency: the degree to which the user profile matches the actual user behavior;

[0046] Coverage: the breadth of user characteristics and behaviors covered by the user profile;

[0047] Predictability: The effectiveness of the user profile in predicting the user's future behavior;

[0048] Privacy protection: the performance of user profiles in protecting user privacy;

[0049] Select samples from the user dataset for testing;

[0050] Prepare test sets, including actual user behavior data and corresponding user profile data;

[0051] Analyze the determined measurement indicators one by one, specifically:

[0052] Consistency check: compare the features in the user portrait with the user's actual behavior data to evaluate consistency;

[0053] Coverage analysis: Analyze the breadth of user characteristics and behaviors covered by user profiles;

[0054] Predictability testing: Use user portraits to predict future user behavior and compare it with actual results to evaluate the accuracy of the prediction;

[0055] Privacy protection evaluation: The privacy protection effect of user portraits is evaluated through obfuscation testing to ensure that even if the data is illegally accessed, the user's specific information cannot be accurately inferred, and the performance of user portraits in protecting user privacy is evaluated;

[0056] Through A / B testing, compare the effects of different user portrait construction methods and select the best method;

[0057] Collect user feedback on recommended content and evaluate the accuracy of user portraits and user satisfaction;

[0058] Use statistical methods to analyze user portrait data and evaluate the reliability and validity of user portraits;

[0059] Simulate potential privacy attacks and test the user portrait’s ability to resist attacks.

[0060] Furthermore, the specific operation steps of S3 are as follows:

[0061] Determine recommendation goals and requirements, including recommended content types, user profile attributes, and recommendation logic rules;

[0062] Design smart contract logic to implement user profile input, recommendation algorithm execution, and recommendation result output;

[0063] Write smart contract code using blockchain programming languages;

[0064] Embed privacy protection measures in smart contracts;

[0065] Conduct unit testing, integration testing, and security testing to ensure the functionality and security of smart contracts;

[0066] Deploy smart contracts to blockchain platforms to achieve decentralized execution;

[0067] Put user data on the chain to ensure data immutability and transparency;

[0068] Call the smart contract, input user profile and parameters, and automatically execute the recommendation logic;

[0069] Output the recommendation results and verify whether the recommendation results meet the expected recommendation rules and standards;

[0070] Collect user feedback and optimize recommendation algorithms;

[0071] Generate transparency reports regularly to show the number of times the recommendation logic is called and the distribution of recommendation results;

[0072] Smart contracts are audited regularly to ensure logical correctness and security.

[0073] Furthermore, the specific operation steps of verifying whether the recommendation result meets the expected recommendation rules and standards in S3 are as follows:

[0074] First, a comprehensive analysis is conducted through the recommended factors, which include:

[0075] Accuracy: Analyze the ratio of the number of correct recommendations to the total number of recommendations in the recommendation results;

[0076] Contextual relevance: Evaluate the relevance of the recommendation results to the user’s current context and quantify it through a relevance percentage, recorded as the relevance ratio;

[0077] Personalization strength: measures the degree of personalization of the recommendation results, that is, the degree of match between the recommended content and the user's unique attributes, and assigns a matching score of 1-100, recorded as the matching score;

[0078] Personalized innovation of recommendation results: the proportion of novel elements or innovative content contained in the recommendation results, recorded as the personalized innovation ratio;

[0079] Normalize the obtained accuracy, correlation ratio, matching score and individual creation ratio and substitute them into the following formula: To get the recommendation value TPZ, where zq, xb, pp and gc are accuracy, correlation ratio, matching score and individual creation ratio respectively, and the recommendation value TPZ is used as the standard for measuring the verification of recommendation results;

[0080] Then collect the user's feedback parameters on the recommendation results to modify the calculated recommendation value to obtain the comprehensive score ZFZ;

[0081] The obtained comprehensive score ZFZ is compared with the preset comprehensive score threshold. When the comprehensive score ZFZ is greater than the comprehensive score threshold, the recommendation result is judged to meet the preset standard. Otherwise, the recommendation result is judged to not meet the preset standard. When it is determined that the recommendation result does not meet the preset standard, the recommendation algorithm is optimized according to the user's specific feedback on the recommendation result.

[0082] Furthermore, the process of collecting the user's feedback parameters on the recommendation results and using them to correct the calculated recommendation value is as follows:

[0083] The feedback parameters include:

[0084] User parameter degree: the depth and frequency of user interaction with recommended content, which is quantified by dwell time and number of interactions. The dwell time and number of interactions are normalized, multiplied by the corresponding weight factors and summed to obtain the parameter value;

[0085] User decision time: the time it takes for a user to make a decision after receiving a recommendation, which reflects the relevance and attractiveness of the recommendation;

[0086] The frequency and timeliness of dynamic updates of user portraits reflect the response speed of the recommendation system to changes in user behavior;

[0087] After normalizing the obtained usage parameter value, user decision time and user portrait update frequency, the product of the usage parameter value and the user portrait update frequency is divided by the user decision time to obtain the usage rating value, which is used as the evaluation standard for measuring the user's feedback on the recommendation results;

[0088] The obtained recommendation value TPZ and the evaluation value are normalized and then put into the following formula: To get the comprehensive score ZFZ, where YPZ is the evaluation value, the comprehensive score ZFZ is used as the final evaluation standard for the recommendation result.

[0089] Furthermore, the specific steps of S4 are as follows:

[0090] Identify and categorize the types of user behavior data that need to be encrypted;

[0091] Use strong encryption algorithms to encrypt user behavior data;

[0092] Manage encryption keys and ensure key security and lifecycle;

[0093] Design a user authorization mechanism to obtain users’ consent for data collection and analysis;

[0094] Verify user authorization before analysis to ensure data usage is in line with user wishes;

[0095] Implement secure data access control to limit unauthorized access;

[0096] Decrypt data with user authorization and perform privacy-preserving data analysis;

[0097] Use differential privacy or homomorphic encryption technology to protect the privacy of decrypted data;

[0098] Generate and encrypt analysis results;

[0099] Logging and auditing data usage;

[0100] Provide user feedback mechanism;

[0101] Develop a data breach response plan to respond to data security incidents.

[0102] Furthermore, the specific steps of S5 are as follows:

[0103] Determine the goals of the incentive mechanism, including improving user engagement and data quality;

[0104] Investigate user needs and preferences to determine incentive types and user acceptance;

[0105] Establish data contribution standards and clarify the type and quality of data that can be incentivized;

[0106] Design incentive rules, including conditions, calculation methods, and issuance time;

[0107] Use smart contracts to automatically enforce incentive rules to ensure fairness and transparency;

[0108] Conduct simulation tests of incentive mechanisms to assess the impact on user behavior and system performance;

[0109] After implementing the incentive mechanism, collect user feedback and evaluate the effect and satisfaction;

[0110] Adjust and optimize incentive mechanisms based on feedback and system performance;

[0111] Disclose the details of the incentive mechanism to users;

[0112] Record and audit incentive payments;

[0113] Balance incentive effects and privacy protection to avoid user privacy leakage;

[0114] Provide users with the option to exit the incentive mechanism and respect the user's right to make independent choices.

[0115] Compared with the prior art, the present invention has the following beneficial effects:

[0116] The present invention adopts a decentralized data storage method to store user data on the blockchain, and utilizes the immutability and transparency of the blockchain to significantly improve the security of the data. In combination with privacy protection technologies such as zero-knowledge proof, the anonymity of users on the blockchain is ensured, and the user's trust in the system is enhanced. This trust is based on the encryption processing and secure storage of data, as well as the fairness and transparency of smart contracts, providing users with a safe and reliable personalized recommendation environment.

[0117] The present invention obtains the recommendation value by comprehensively analyzing the accuracy, context relevance, personalization strength and personalization innovation of the recommendation results, using normalization processing and a specific calculation formula, and then corrects it in combination with user feedback parameters to obtain a comprehensive score. This process not only optimizes the recommendation algorithm and ensures the relevance and attractiveness of the recommended content, but also effectively improves user satisfaction and participation by regularly collecting user feedback and dynamically adjusting the recommendation logic.

[0118] The present invention encourages users to contribute high-quality data through incentive mechanism design, and ensures the automatic execution of incentive rules through smart contracts, thereby ensuring the fairness and transparency of the incentive process. By balancing the incentive effect and privacy protection, the present invention avoids the potential leakage of user privacy during the incentive process, provides users with the option to exit the incentive mechanism at any time, respects the user's right of autonomous choice, and further strengthens the user's control over their own data and the transparency of the system. These measures jointly promote a recommendation system environment that both encourages user participation and strictly protects user privacy. BRIEF DESCRIPTION OF THE DRAWINGS

[0119] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings;

[0120] Figure 1 The present invention is a flow chart of the method. DETAILED DESCRIPTION

[0121] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0122] It should be understood that the terms "include" and "comprising" used in the specification and claims of the present disclosure indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0123] It should also be understood that the terms used in this disclosure are only for the purpose of describing specific embodiments and are not intended to limit the disclosure. As used in this disclosure and claims, the singular forms of "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise. It should also be further understood that the term "and / or" used in this disclosure and claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0124] like Figure 1 As shown, the personalized content recommendation method based on user behavior and preference includes the following steps:

[0125] Decentralized data storage: Store user data on the blockchain to ensure data immutability and transparency;

[0126] First, the data to be stored is preliminarily screened. The specific process is as follows:

[0127] According to the application scenarios and business needs, determine the standards and rules for data screening, and collect user data from various sources, including log files, user input, sensor data, etc.; clean the collected data, remove invalid, erroneous or incomplete data records, verify the accuracy and consistency of the data, ensure that the data meets the preset standards, and remove duplicate data records; classify data according to data type and purpose, and use data compression technology to reduce data storage space requirements and improve storage efficiency.

[0128] After the initial screening, encryption is performed to ensure the security of the data during transmission. The encrypted data is divided into multiple fragments to adapt to the storage structure of the blockchain and improve the efficiency of data storage. A hash value is generated for the encrypted data fragment as the unique identifier of the data to ensure the integrity and traceability of the data.

[0129] Write smart contracts to define the rules and conditions for data storage, including data access rights, storage location, etc.; upload data fragments and data fingerprints to the blockchain network through smart contracts to achieve decentralized data storage, and use multiple nodes in the blockchain network to verify the uploaded data to ensure data consistency and immutability; build indexes for data stored on the blockchain to facilitate subsequent data retrieval and analysis, and set data access rights to ensure that only authorized users or systems can access specific data fragments;

[0130] Implement multi-node backup of data in the blockchain network to improve data reliability and fault tolerance, implement data update and maintenance rules through smart contracts to ensure data real-time and accuracy, and use zero-knowledge proof privacy protection technology to ensure the privacy of user data and maintain anonymity even on the blockchain.

[0131] Privacy-preserving user profile construction: Through cryptographic technologies such as zero-knowledge proof, a user profile is constructed that does not disclose the user's specific information;

[0132] Determine the attributes that the user profile needs to include, such as age range, interest preferences, consumption habits, etc., but do not include any sensitive information that can identify an individual; collect user behavior data, such as browsing history, purchase history, search queries, etc.; clean and pre-process the collected data to remove invalid or irrelevant information; anonymize user data to ensure that no information that can be traced back to an individual is included in the process of building the user profile;

[0133] Using zero-knowledge proof technology, it is possible to prove that a user belongs to a specific attribute group without revealing specific personal information. The attributes in the user portrait are encrypted to ensure that even if the data is illegally accessed, the original information cannot be interpreted. The encrypted attributes are integrated into an encrypted user portrait, which can be used in the recommendation algorithm without exposing the user's specific information. Privacy-preserving data analysis techniques, such as differential privacy, are used to analyze the user portrait to discover the general trends and preferences of the user group. The constructed user portrait is verified and tested to ensure its accuracy. The specific process is as follows:

[0134] Determine the metrics for measuring the accuracy of user portraits and the effectiveness of privacy protection, including:

[0135] Consistency: the degree of match between the user profile and actual user behavior; Coverage: the breadth of user characteristics and behaviors covered by the user profile; Predictability: the effectiveness of the user profile in predicting the user's future behavior;

[0136] Privacy protection: the performance of user portraits in protecting user privacy; select representative samples from user data sets for testing to ensure the diversity and breadth of samples; prepare test sets, including users' actual behavior data and corresponding user portrait data;

[0137] Analyze the determined measurement indicators one by one, specifically:

[0138] Consistency test: compare the features in the user portrait with the user's actual behavior data to evaluate consistency. For example, check whether the interest preferences in the user portrait are consistent with the user's purchase history; Coverage analysis: analyze the breadth of user features and behaviors covered by the user portrait to ensure that the user portrait can cover the behavior patterns of most users; Predictability test: use the user portrait to predict the user's future behavior, such as purchase intention, content preference, etc., and compare with the actual results to evaluate the accuracy of the prediction; Privacy protection evaluation: evaluate the privacy protection effect of the user portrait through obfuscation testing to ensure that even if the data is illegally accessed, the user's specific information cannot be accurately inferred, and evaluate the performance of the user portrait in protecting user privacy;

[0139] Compare the effects of different user portrait construction methods and select the best method; collect user feedback on recommended content and evaluate the accuracy of user portraits and user satisfaction; use statistical methods to analyze user portrait data and evaluate the reliability and effectiveness of user portraits; simulate potential privacy attacks and test the user portrait's ability to resist attacks.

[0140] Design mechanisms to regularly update user profiles to reflect the latest changes in user behavior and preferences, while ensuring privacy protection during the update process; implement access control mechanisms to ensure that only authorized systems and personnel can access user profile data; conduct regular audits of data access and use to ensure compliance with privacy protection policies; obtain users' explicit consent to collect and process their data, and provide users with the option to exit or delete their data at any time.

[0141] Smart contract recommendation logic: Use smart contracts to implement recommendation logic to ensure the fairness and transparency of the recommendation process.

[0142] Determine the goals and requirements of the recommendation, including the type of recommended content, the attributes of the user portrait, the rules of the recommendation logic, etc., and design the logic of the smart contract, including the input of the user portrait, the execution of the recommendation algorithm, the output of the recommendation results, etc.; use Solidity or other blockchain programming languages ​​to write smart contract code to implement the above-designed recommendation logic;

[0143] Embed privacy protection measures in smart contracts, such as data encryption and zero-knowledge proof, to ensure the security of user data. Before deploying smart contracts, conduct sufficient testing, including unit testing, integration testing and security testing, to ensure the functionality and security of the contract. Deploy the written smart contracts to the blockchain platform so that they can be executed in the blockchain network, and upload user data to the blockchain according to the steps of decentralized data storage to ensure the data is tamper-proof and transparent. When recommendations are needed, call the smart contract, input user portraits and related parameters, and the smart contract automatically executes the recommendation logic. The smart contract processes user data according to the preset recommendation logic, outputs recommendation results, and ensures the fairness and transparency of the results. Verify the recommendation results output by the smart contract to determine whether the recommendation results meet the expected recommendation rules and standards. Specifically, conduct a comprehensive analysis through recommendation factors, which include:

[0144] Accuracy: Analyze the ratio of the number of correct recommendations to the total number of recommendations in the recommendation results; Contextual relevance: Evaluate the relevance of the recommendation results to the user's current context and quantify it through the relevant percentage, recorded as the relevance ratio; Personalization strength: Measure the degree of personalization of the recommendation results, that is, the match between the recommended content and the user's unique attributes, and assign a matching score of 1-100, recorded as the match score; Personalized innovation of recommendation results: The proportion of novel elements or innovative content contained in the recommendation results, recorded as the personal innovation ratio;

[0145] Normalize the obtained accuracy, correlation ratio, matching score and individual creation ratio and substitute them into the following formula: To get the recommendation value TPZ, where zq, xb, pp and gc are accuracy, correlation ratio, matching score and individual creation ratio respectively, and the recommendation value TPZ is used as the standard for measuring the verification of recommendation results;

[0146] Then collect the user's feedback parameters on the recommendation results to modify the calculated recommendation value, where the feedback parameters include:

[0147] User parameter degree: the depth and frequency of user interaction with recommended content, which is quantified by dwell time and number of interactions. The dwell time and number of interactions are normalized and multiplied by the corresponding weight factors and summed to obtain the parameter value; User decision time: the time required for users to make a decision after receiving a recommendation, which is used to reflect the relevance and attractiveness of the recommendation; The dynamic update frequency of user portraits, the frequency and timeliness of user portrait updates, reflects the response speed of the recommendation system to changes in user behavior;

[0148] After normalizing the obtained usage parameter value, user decision time and user portrait update frequency, the product of the usage parameter value and the user portrait update frequency is divided by the user decision time to obtain the usage rating value, and this usage rating value is used as the evaluation standard for measuring user feedback on the recommendation results; the obtained recommendation rating value TPZ and the usage rating value are normalized and then entered into the following formula: To obtain the comprehensive score ZFZ, where YPZ is the evaluation value, and the comprehensive score ZFZ is used as the final evaluation standard for the recommendation result; the comprehensive score ZFZ is compared with the preset comprehensive score threshold. When the comprehensive score ZFZ is greater than the comprehensive score threshold, it is judged that the recommendation result reaches the preset standard, otherwise it is judged that the recommendation result does not reach the preset standard. When it is determined that the recommendation result does not reach the preset standard, the recommendation algorithm is optimized according to the user's specific feedback on the recommendation result.

[0149] Based on user feedback and system performance monitoring results, smart contracts are upgraded and optimized to improve the accuracy of the recommendation system and user experience. Transparency reports on smart contract execution are regularly generated, including the number of calls to the recommendation logic, the distribution of recommendation results, etc., to enhance the transparency of the system. Smart contracts are regularly audited to ensure the correctness and security of their logic and prevent potential vulnerabilities and attacks.

[0150] Encrypted analysis of user behavior: User behavior data is encrypted and analyzed only with user authorization.

[0151] Identify the types of user behavior data that need to be encrypted, such as browsing history, click behavior, search queries, etc., and classify them; encrypt the identified user behavior data and use strong encryption algorithms (such as AES) to ensure the confidentiality of the data; manage the keys used in the encryption process to ensure the security and lifecycle management of the keys; design a user authorization mechanism to allow users to explicitly consent to the collection and analysis of their data;

[0152] Before analyzing encrypted data, verify whether the user is authorized to ensure that the use of data is in line with the user's wishes; implement secure data access control to ensure that only authorized systems or personnel can access encrypted data; with user authorization, use the corresponding key to decrypt the encrypted data for analysis; use privacy-preserving data analysis technologies, such as differential privacy or homomorphic encryption, to protect the privacy of user data.

[0153] Based on the decrypted data, generate analysis results, such as user behavior patterns, preference trends, etc., and encrypt the analysis results again to ensure security during storage and transmission; record and audit data usage to ensure that data usage complies with privacy policies and laws and regulations; provide a user feedback mechanism to allow users to understand how their data is used, and provide options to modify or revoke authorization; formulate a data breach response plan to deal with possible data security incidents.

[0154] Incentive mechanism design: Design an incentive mechanism to encourage users to contribute data while protecting their privacy rights.

[0155] Determine the goals of the incentive mechanism, including increasing user engagement, improving data diversity and quality, and encouraging long-term user engagement; investigate user needs and preferences to understand their expectations and acceptance of the incentive mechanism; determine the type of incentive, such as material rewards (tokens, discount coupons, gifts, etc.), spiritual rewards (rankings, badges, certifications, etc.) or service rewards (advanced features, personalized services, etc.). When designing the incentive mechanism, ensure that the use and processing of user data follows the privacy protection principle and does not infringe on user privacy;

[0156] Establish standards and requirements for user-contributed data, clarify what type and quality of data can be incentivized, and design specific incentive rules, including incentive conditions, incentive calculation methods, incentive issuance time, etc.; use smart contracts to automatically execute incentive rules to ensure the fairness and transparency of incentives; conduct simulation tests on incentive mechanisms before implementation to evaluate their impact on user behavior and system performance; after the incentive mechanism is implemented, collect user feedback to understand the actual effect of the incentive mechanism and user satisfaction;

[0157] Adjust and optimize the incentive mechanism based on user feedback and system performance, and disclose detailed information about the incentive mechanism to users, including the incentive calculation method and issuance process, to improve transparency; record and audit the issuance of incentives to ensure that the implementation of the incentive mechanism meets expectations; balance the incentive effect and privacy protection needs when designing the incentive mechanism to avoid user privacy leakage due to excessive incentives; provide users with the option to exit the incentive mechanism at any time, and respect the user's right to choose independently.

[0158] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. Personalized content recommendation method based on user behavior and preferences, It is characterized in that The following steps are involved: S1: Decentralized data storage: Store user data on the blockchain to ensure data immutability and transparency; S2: Privacy-preserving user portrait construction: Use zero-knowledge proof cryptography technology to construct a user portrait that does not disclose the user's specific information. The specific steps are as follows: Determine the attributes required for user profiles, excluding sensitive information that can identify individuals; Collect and clean user behavior data, and perform pre-processing to remove invalid or irrelevant information; Anonymize user data to ensure that it does not contain information that can be traced back to personal identities; Use zero-knowledge proof technology to prove user attribute groups and avoid leaking specific personal information; Encrypt user portrait attributes; Integrate encryption attributes to form encrypted user profiles for recommendation algorithms; Use differential privacy to protect data analysis technology, analyze user portraits, and discover user group trends and preferences; Verify and test user portraits to ensure accuracy and effectiveness of privacy protection; Design regular update mechanisms to reflect the latest changes in user behavior and preferences while protecting privacy; Implement access control mechanisms to ensure that only authorized systems and personnel have access to user profile data; Conduct regular data access audits; Obtain explicit consent from users to collect and process data, and provide options to opt out or delete data; S3: Smart contract recommendation logic: Use smart contracts to implement recommendation logic to ensure the fairness and transparency of the recommendation process. The specific steps are as follows: Determine recommendation goals and requirements, including recommended content types, user profile attributes, and recommendation logic rules; Design smart contract logic to implement user portrait input, recommendation algorithm execution and recommendation result output; Write smart contract code using blockchain programming languages; Embed privacy protection measures in smart contracts; Conduct unit testing, integration testing, and security testing to ensure the functionality and security of smart contracts; Deploy smart contracts to blockchain platforms to achieve decentralized execution; Put user data on the chain to ensure data immutability and transparency; Call the smart contract, input user profile and parameters, and automatically execute the recommendation logic; Output the recommendation results and verify whether the recommendation results meet the expected recommendation rules and standards; Collect user feedback and optimize recommendation algorithms; Generate transparency reports regularly to show the number of times the recommendation logic is called and the distribution of recommendation results; Regularly audit smart contracts to ensure logical correctness and security; The specific process of verifying whether the recommendation results meet the expected recommendation rules and standards is as follows: First, a comprehensive analysis is conducted through the recommended factors, which include: Accuracy: Analyze the ratio of the number of correct recommendations to the total number of recommendations in the recommendation results; Contextual relevance: Evaluate the relevance of the recommendation results to the user’s current context and quantify it through a relevance percentage, recorded as the relevance ratio; Personalization strength: measures the personalization degree of the recommendation results, that is, the matching degree between the recommended content and the user's unique attributes, and assigns a matching score of 1-100, which is recorded as the matching score; Personalized innovation of recommendation results: the proportion of novel elements or innovative content contained in the recommendation results, recorded as the personalized innovation ratio; The obtained accuracy, correlation ratio, matching score and individual creation ratio are normalized and substituted into the following formula: TPZ = 0.95 (zq + xb + pp) + (gc + 1.25) 1.21 To get the recommendation value TPZ, where zq, xb, pp and gc are accuracy, correlation ratio, matching score and individual creation ratio respectively, and the recommendation value TPZ is used as the standard for measuring the verification of recommendation results; Then collect the user's feedback parameters on the recommendation results to modify the calculated recommendation value to obtain the comprehensive score ZFZ; The obtained comprehensive score ZFZ is compared with the preset comprehensive score threshold. When the comprehensive score ZFZ is greater than the comprehensive score threshold, it is judged that the recommendation result meets the preset standard. Otherwise, it is judged that the recommendation result does not meet the preset standard. When it is determined that the recommendation result does not meet the preset standard, the recommendation algorithm is optimized according to the specific feedback of the user on the recommendation result. The process of collecting the user's feedback parameters on the recommendation results and correcting the calculated recommendation value is as follows: The feedback parameters include: User parameter degree: the depth and frequency of user interaction with recommended content, which is quantified by dwell time and number of interactions. The dwell time and number of interactions are normalized, multiplied by the corresponding weight factors and summed to obtain the parameter value; User decision time: the time it takes for a user to make a decision after receiving a recommendation, which reflects the relevance and attractiveness of the recommendation; The frequency and timeliness of dynamic updates of user portraits reflect the response speed of the recommendation system to changes in user behavior; After normalizing the obtained usage parameter value, user decision time and user portrait update frequency, the product of the usage parameter value and the user portrait update frequency is divided by the user decision time to obtain the usage rating value, which is used as the evaluation standard for measuring the user's feedback on the recommendation results; The obtained recommendation value TPZ and the evaluation value are normalized and then put into the following formula: ZFZ = (TPZ + 1.22) * (1-YPZ -1 )-0.26 to get the comprehensive score ZFZ, where YPZ is the evaluation value, and the obtained comprehensive score ZFZ is used as the final evaluation standard of the recommendation result; S4: Encrypted analysis of user behavior: Encrypt user behavior data and analyze it only when the user authorizes it; S5: Incentive mechanism design: Encourage users to contribute data while protecting their privacy rights through incentive mechanisms.

2. The personalized content recommendation method based on user behavior and preference according to claim 1, It is characterized in that The specific steps of S1 are as follows: Perform preliminary screening of user data to select data that meets preset standards; Encrypt the filtered data to ensure the security of the data during transmission; Split the encrypted data into multiple fragments to adapt to the storage structure of the blockchain; Generate a hash value for each data fragment as a unique identifier of the data to ensure data integrity and traceability; Write smart contracts to define the rules and conditions for data storage, and users guarantee data access rights and storage locations; Upload data fragments and data fingerprints to the blockchain network through smart contracts to achieve decentralized storage of data; Use multiple nodes in the blockchain network to verify the uploaded data to ensure data consistency and immutability; Build an index for data stored on the blockchain; Set data access permissions to ensure that only authorized users or systems can access data fragments; Realize multi-node backup of data in the blockchain network; Implement data update and maintenance rules through smart contracts; Adopt zero-knowledge proof privacy protection technology to ensure the privacy of user data and maintain anonymity even on the blockchain.

3. The personalized content recommendation method based on user behavior and preference according to claim 2, It is characterized in that The specific operation steps for preliminary screening of user data in S1 are as follows: Determine the standards and rules for data screening based on application scenarios and business needs, and collect user data from various sources; Clean the collected data, remove invalid, erroneous or incomplete data records, verify the accuracy and consistency of the data, ensure that the data meets the preset standards, and remove duplicate data records; Classify data according to data type and purpose, and use data compression technology to reduce data storage space requirements.

4. The personalized content recommendation method based on user behavior and preference according to claim 1, It is characterized in that The user portrait is verified and tested in S2 to ensure accuracy and effectiveness of privacy protection. The specific process is as follows: The metrics for measuring the accuracy of user portraits and the effectiveness of privacy protection include: Consistency: the degree to which the user profile matches the actual user behavior; Coverage: the breadth of user characteristics and behaviors covered by the user profile; Predictability: The effectiveness of the user profile in predicting the user's future behavior; Privacy protection: the performance of user profiles in protecting user privacy; Select samples from the user dataset for testing; Prepare test sets, including actual user behavior data and corresponding user profile data; Analyze the determined measurement indicators one by one, specifically: Consistency check: compare the features in the user portrait with the user's actual behavior data to evaluate consistency; Coverage analysis: Analyze the breadth of user characteristics and behaviors covered by user profiles; Predictability testing: Use user portraits to predict future user behavior and compare it with actual results to evaluate the accuracy of the prediction; Privacy protection evaluation: The privacy protection effect of user portraits is evaluated through obfuscation testing to ensure that even if the data is illegally accessed, the user's specific information cannot be accurately inferred, and the performance of user portraits in protecting user privacy is evaluated; Through A / B testing, compare the effects of different user portrait construction methods and select the best method; Collect user feedback on recommended content and evaluate the accuracy of user portraits and user satisfaction; Use statistical methods to analyze user portrait data and evaluate the reliability and validity of user portraits; Simulate potential privacy attacks and test the user portrait’s ability to resist attacks.

5. The personalized content recommendation method based on user behavior and preference according to claim 1, It is characterized in that The specific steps of S4 are as follows: Identify and categorize the types of user behavior data that need to be encrypted; Use strong encryption algorithms to encrypt user behavior data; Manage encryption keys and ensure key security and lifecycle; Design a user authorization mechanism to obtain users’ consent for data collection and analysis; Verify user authorization before analysis to ensure data usage is in line with user wishes; Implement secure data access control to limit unauthorized access; Decrypt data with user authorization and perform privacy-preserving data analysis; Use differential privacy or homomorphic encryption technology to protect the privacy of decrypted data; Generate and encrypt analysis results; Logging and auditing data usage; Provide user feedback mechanism; Develop a data breach response plan to respond to data security incidents.

6. The personalized content recommendation method based on user behavior and preference according to claim 1, It is characterized in that The specific steps of S5 are as follows: Determine the goals of the incentive mechanism, including improving user engagement and data quality; Investigate user needs and preferences to determine incentive types and user acceptance; Establish data contribution standards and clarify the type and quality of data that can be incentivized; Design incentive rules, including conditions, calculation methods, and issuance time; Use smart contracts to automatically enforce incentive rules to ensure fairness and transparency; Conduct simulation tests of incentive mechanisms to assess the impact on user behavior and system performance; After implementing the incentive mechanism, collect user feedback and evaluate the effect and satisfaction; Adjust and optimize incentive mechanisms based on feedback and system performance; Disclose the details of the incentive mechanism to users; Record and audit incentive payments; Balance incentive effects and privacy protection to avoid user privacy leakage; Provide users with the option to exit the incentive mechanism and respect the user's right to make independent choices.

Citation Information

Patent Citations

  • Product recommendation method and device, equipment and storage medium

    CN116128607A

  • Social platform media content display method and system based on block chain technology

    CN117609642A

  • Search engine accurate recommendation method based on user behavior analysis

    CN117851689A

  • Federal generalized matrix decomposition recommendation method based on incentive mechanism

    CN118132845A

  • Digital Profile System of Personal Attributes, Tendencies, Recommended Actions, and Historical Events with Privacy Preserving Controls

    US20110238482A1