Decentralization information promotion method based on big data

Abstract

The invention discloses a decentralized information promotion method based on big data, and relates to the technical field of information promotion, and the method comprises the steps: obtaining the historical behavior data of a user, and carrying out the analysis to generate a user information chain; for the information needing to be popularized, a popularization matching coefficient is calculated according to the user information chain, and the information needing to be popularized is popularized to users with the coefficients larger than a threshold value; a decentralized network is arranged in an information promotion system, any node in the network can execute an information promotion task, obtain information promotion data of the node in real time, calculate the promotion influence of the node, perform descending sorting, and perform node information synchronization on high-quality information in combination with the quality score and the promotion influence of the information in the node. Through accurate information matching and personalized recommendation, efficient resource utilization of a decentralized network, information quality guarantee and efficient propagation of high-quality content, the correlation, real-time performance and effect of information promotion are improved, propagation of high-quality information is ensured, and user experience is optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of information promotion, and in particular to a decentralized information promotion method based on big data. Background Art

[0002] With the rapid development of Internet technology, the way and carrier of information dissemination have also undergone fundamental changes. In particular, the rise of online platforms such as social media, e-commerce platforms, and search engines has made the flow speed and dissemination range of information reach an unprecedented scale. However, the problems of information overload and lack of accuracy have gradually emerged, and traditional information promotion methods are also facing more and more challenges.

[0003] Although the traditional information promotion methods based on keywords, clicks or simple recommendation algorithms can meet user needs to a certain extent, they have several significant shortcomings. Insufficient personalization: The recommendation algorithms used by most information promotion systems are still based on simple user behavior analysis, such as browsing history or purchase history. Although this method can capture some of the user's interests, it cannot deeply analyze the user's diverse needs and deep interests. Due to the lack of accurate personalized recommendations, users often receive a large amount of irrelevant information, resulting in inefficient promotion and information overload; Disadvantages of centralized systems: Most traditional information promotion platforms rely on centralized servers for data analysis and information push. Although this centralized approach can concentrate computing resources and optimize algorithms, it also brings problems such as single point failure, data leakage, system overload, and cannot efficiently utilize edge computing and distributed computing resources. With the surge in the number of users, traditional centralized promotion methods are prone to computing resource bottlenecks, affecting the real-time and effectiveness of promotion; Information quality is difficult to guarantee: The existing promotion system cannot effectively distinguish the quality of information, and many promotion platforms rely on advertisers' budgets and bidding systems to push information. This results in some low-quality information also being exposed in large quantities, while high-quality but low-budget content may be ignored. The lack of measurement standards for information quality often affects the quality of users' decisions and even causes a crisis of trust. Summary of the invention

[0004] Based on the above-mentioned shortcomings of the prior art, the purpose of the present invention is to provide a decentralized information promotion method based on big data to solve the above-mentioned technical problems.

[0005] To achieve the above purpose, the present invention provides the following technical solution: a decentralized information promotion method based on big data, comprising:

[0006] Obtaining the user's historical behavior data, analyzing and generating a user information chain, wherein the user information chain includes an information identifier, an information interest score, a chain information type, and a pointer, wherein the pointer points to the next accessed information according to the order in which the information is accessed;

[0007] For the information to be promoted, a promotion matching coefficient is calculated according to the user information chain, and the information to be promoted is promoted to users whose promotion matching coefficient is greater than a threshold;

[0008] In the information promotion system, a decentralized network is set up, and any node in the decentralized network can perform information promotion tasks, obtain the node's information promotion data in real time, calculate the node's promotion influence, sort all nodes in descending order according to the node's promotion influence, and combine the quality score of the information in the node to synchronize node information with high-quality information according to the promotion influence and quality score.

[0009] The present invention is further configured to obtain historical behavior data of users and analyze and generate user information chains, including:

[0010] Obtaining historical behavior data of users, including information identifiers, timestamps, information types, and interaction data of historically browsed information. The interaction data includes browsing time, information sources, and interaction data. The information sources include search acquisition and promotion acquisition. The interaction data includes likes, comments, favorites, and dislikes.

[0011] Within a preset time window, the information identifiers are sorted in order according to the order of timestamps to generate a user information chain;

[0012] Calculate the information interest score according to browsing time, information source and interaction data, and set the information interest score as an attribute of the information identifier;

[0013] The information type and the information interest score are obtained, the interest type score of the user information chain is calculated, and the information types with the top N interest type scores are set as chain information types.

[0014] The present invention is further configured to calculate the information interest score according to the browsing time, information source and interaction data, and the calculation logic of the information interest score is: Among them, Sii(IList i ) is the information interest score of the information corresponding to the i-th information identifier in the user information chain IList, ISo i is the information source of the information corresponding to the i-th information identifier. When the information source is search acquisition, the value is 1. When the information source is promotion acquisition, the value is 0.8. i and DOWN i is the likes and dislikes of the information corresponding to the i-th information identifier, with a value of 0 or 1. It is 1 when there is a like or dislike, and 0 when there is no like or dislike. i is the browsing time of the information corresponding to the i-th information identifier, IW iis the number of words of the information corresponding to the i-th information identifier, CW i is the number of words in the comments of the information corresponding to the i-th information identifier, β1 and β2 are weight coefficients, both greater than 0;

[0015] The information type and the information interest score are obtained, and the interest type score of the user information chain is calculated. The calculation logic of the interest type score is as follows: all information types of all information in the user information chain are obtained, and the information type of the information corresponding to the i-th information identifier is assigned the information interest score. The information interest scores of all the same information types in the user information chain are added together to obtain the interest type score of each information type, wherein each information includes at least one information type.

[0016] The present invention is further configured to calculate the promotion matching coefficient according to the user information chain, including:

[0017] When the information to be promoted has not been accessed by any user, a first promotion matching coefficient is calculated based on the matching degree between the information type of the information to be promoted and the link information type of the user information chain;

[0018] When the information to be promoted is accessed by any user, the user who has accessed the information to be promoted is marked as the accessing user, and the remaining users are marked as the users to be promoted. The user information chain of the accessing user is obtained, and the type matching coefficient is calculated based on the matching degree between the information type of the information to be promoted and the chain information type of the user information chain of the user to be promoted;

[0019] According to all the chain information types of users whose information interest scores of the visiting users for the information to be promoted are greater than the information interest score threshold, an access chain information type set is set, and a user matching coefficient of the chain information type of the user information chain of the user to be promoted and the access chain information type set is calculated;

[0020] A second promotion matching coefficient is calculated according to the type matching coefficient and the user matching coefficient.

[0021] The present invention is further configured to use a Word2Vec model to map words in each information type to a vector space of a fixed dimension, generate an embedding vector corresponding to each information type, and when vectorizing the chain information type of the user information chain and the information type of the information to be promoted, add the embedding vectors of all words in the information type to obtain the vector representation of the chain information type of the user information chain and the information type of the information to be promoted, wherein, since the words in each information type are mapped to a vector space of a fixed dimension, the dimension of the vector representation is the number of all information types;

[0022] When performing the first promotion matching coefficient, the type matching coefficient and the user matching coefficient, it is achieved by calculating the cosine similarity between vectors; the second promotion matching coefficient is calculated by weighted summing the type matching coefficient and the user matching coefficient, the weights are all greater than 0, and the sum is 1.

[0023] The present invention is further configured to obtain information promotion data of nodes in real time, including: number of information visitors, forwarding rate, average sequence number of information in user information chain, like rate, comment rate and collection rate;

[0024] The promotion influence of computing nodes includes:

[0025] Calculate the promotion breadth based on the number of people who visit the information and the forwarding rate;

[0026] Calculate the promotion depth based on the information's sequence number, like rate, comment rate, and collection rate in the user's information chain;

[0027] The promotion influence is calculated according to the promotion breadth and the promotion depth.

[0028] The present invention is further configured such that the calculation logic of the promotion breadth is: Among them, Rad is the promotion breadth, m is the information in the node, Gacc i is the number of people who visited the i-th information, Gsh i is the forwarding rate of the i-th information;

[0029] The calculation logic of the promotion depth is: Among them, Dadv is the promotion depth, Num i is the average sequence number of the i-th information in the user information chain, L i is the like rate of the i-th information, C i is the comment rate of the i-th information, F i is the collection rate of the i-th information;

[0030] The promotion breadth and the promotion depth are normalized, and weighted sum is performed to obtain the promotion influence, where the weights are all greater than 0 and the sum is 1.

[0031] The present invention is further configured to synchronize node information of high-quality information according to promotion influence and quality score in combination with the quality score of the information in the node, including:

[0032] The quality score of information is calculated based on the number of visitors, forwarding rate, like rate, comment rate and collection rate of the information. The calculation logic is: Qua i =Gacc i ·(1+Gsh i +L i+C i +F i ), where Qua i Score the quality of the i-th information;

[0033] The comprehensive score of the information is calculated based on the promotion influence and quality score. The calculation logic is: Sco i =Qua i *Inf i , among which Sco i is the comprehensive score of the i-th information, Inf i is the promotion influence of the i-th information;

[0034] Information with a comprehensive score greater than the comprehensive score threshold is set as high-quality information.

[0035] The present invention is further configured to synchronize node information of high-quality information, including: obtaining a node embedding vector of a node where the high-quality information to be synchronized is located, and the logic for obtaining the node embedding vector is: obtaining embedding vectors of all users in the node, adding and averaging the embedding vectors of all users to obtain the node embedding vector;

[0036] Calculate the matching degree of the node embedding vectors of the users of the remaining nodes and the nodes where the high-quality information to be synchronized is located, which is recorded as the node user similarity. The calculation method is to calculate the cosine similarity between the vectors.

[0037] Obtain the user overlap rate of the remaining nodes and the node where the high-quality information to be synchronized is located, which is recorded as the node user overlap rate;

[0038] When the node user similarity between the remaining nodes and the node where the high-quality information to be synchronized is located is greater than the similarity threshold and the node user overlap rate is less than the overlap threshold, information synchronization is performed.

[0039] The present invention is further configured to include: before promoting information, matching the user's historical behavior data is performed, and when the information has been visited historically, no further promotion is performed.

[0040] The present invention provides a decentralized information promotion method based on big data, wherein the method obtains historical behavior data of users, analyzes and generates a user information chain, wherein the user information chain includes an information identifier, an information interest score, a chain information type and a pointer, and the pointer points to the next accessed information according to the access order of the information; for information to be promoted, a promotion matching coefficient is calculated according to the user information chain, and the information to be promoted is promoted to users whose promotion matching coefficient is greater than a threshold; in an information promotion system, a decentralized network is set, and any node in the decentralized network can perform information promotion tasks, obtain information promotion data of the node in real time, calculate the promotion influence of the node, sort all nodes in descending order according to the promotion influence of the node, and synchronize node information of high-quality information according to the promotion influence and quality score in combination with the quality score of the information in the node, and the beneficial effects produced include:

[0041] 1. Accurate information matching and personalized recommendation: By building a user information chain, we can accurately analyze user interests, understand users' potential needs more comprehensively, and provide personalized promotional content. In particular, through the comprehensive calculation of information interest scores and chain information types, we can accurately push information to different user groups, greatly improving the relevance of information promotion and user satisfaction;

[0042] 2. Efficient promotion and resource utilization of decentralized networks: The decentralized network architecture enables information promotion tasks to be distributed to various nodes in the network, making full use of distributed computing and storage resources, effectively avoiding the computing bottlenecks and single point failure problems of traditional centralized systems, and improving information processing capabilities and real-time promotion. Each node can independently perform information promotion tasks and optimize according to its own computing resources and user needs, thereby improving the efficiency and coverage of information promotion;

[0043] 3. Information quality assurance and efficient dissemination of high-quality content: Through the comprehensive evaluation mechanism of promotion influence and quality score, through the dynamic calculation of the promotion influence of the node, the number of visitors to the information, the forwarding rate, the like rate and other indicators, it can ensure that high-quality information is promoted first. Especially in the process of information synchronization between nodes, by combining promotion influence and quality score, high-quality information can be effectively screened out and synchronized and disseminated in the network. It improves the dissemination effect of information, prevents the proliferation of low-quality information, and optimizes the user experience.

[0044] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0046] Figure 1 The present invention is a flowchart of a decentralized information promotion method based on big data, showing an exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0047] The following will describe the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention, not for limiting the scope of protection of the present invention.

[0048] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and thus the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0049] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it is obvious to those skilled in the art that the embodiments of the present invention can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present invention difficult to understand.

[0050] Decentralized information promotion methods based on big data, such as Figure 1 As shown, including:

[0051] Obtaining the user's historical behavior data, analyzing and generating a user information chain, wherein the user information chain includes an information identifier, an information interest score, a chain information type, and a pointer, wherein the pointer points to the next accessed information according to the order in which the information is accessed;

[0052] For the information to be promoted, a promotion matching coefficient is calculated according to the user information chain, and the information to be promoted is promoted to users whose promotion matching coefficient is greater than a threshold;

[0053] In the information promotion system, a decentralized network is set up, and any node in the decentralized network can perform information promotion tasks, obtain the node's information promotion data in real time, calculate the node's promotion influence, sort all nodes in descending order according to the node's promotion influence, and combine the quality score of the information in the node to synchronize node information with high-quality information according to the promotion influence and quality score.

[0054] The present invention is further configured to obtain historical behavior data of users and analyze and generate user information chains, including:

[0055] Obtain the user's historical behavior data, including the information identifier, timestamp, information type and interaction data of the historically browsed information. The interaction data includes browsing time, information source and interaction data. The information source includes search acquisition and promotion acquisition. The interaction data includes likes, comments, favorites and dislikes. Specifically, the information identifier represents the unique identifier of each information, which is used to distinguish different information; the timestamp represents the time when the user browses a certain information, which is convenient for subsequent sorting by time; the information type is used to identify the category to which the information belongs;

[0056] Within a preset time window, the information identifiers are sorted in order according to the order of timestamps to generate a user information chain; the preset time window can be set to one day, one week, and one month, and no specific restrictions are made here;

[0057] The information interest score is calculated according to the browsing time, information source and interaction data, and the information interest score is set as the attribute of the information identifier; the present invention is further configured to calculate the information interest score according to the browsing time, information source and interaction data, and the calculation logic of the information interest score is:

[0058]

[0059] Among them, Sii(IList i ) is the information interest score of the information corresponding to the i-th information identifier in the user information chain IList, ISo i is the information source of the information corresponding to the i-th information identifier. When the information source is search acquisition, the value is 1. When the information source is promotion acquisition, the value is 0.8. i and DOWN i is the likes and dislikes of the information corresponding to the i-th information identifier, with a value of 0 or 1. It is 1 when there is a like or dislike, and 0 when there is no like or dislike. i is the browsing time of the information corresponding to the i-th information identifier, IW i is the number of words of the information corresponding to the i-th information identifier, CW iis the number of words in the comments of the information corresponding to the i-th information identifier, β1 and β2 are weight coefficients, both greater than 0; specifically, the purpose of the above calculation logic is to comprehensively evaluate the attractiveness of a certain information to the user through the user's historical behavior data, so as to calculate an information interest score, which is used to determine whether the information is suitable for continued promotion to the user or priority recommendation. Through this scoring mechanism, the accuracy of information promotion can be improved, ensuring that the information seen by the user is more in line with their interest preferences, thereby enhancing the user experience; the information that users actively search for and obtain is often more in line with the user's real needs and represents a higher level of interest. Therefore, the information source item is used to distinguish between information that users actively obtain and information that they passively receive, so as to reflect the user's active interest level in the score; likes and dislikes are direct feedback from users on the information. Likes indicate a user's positive interest in the information and increase the score; dislikes indicate a user's negative reaction to the information and reduce the score; because likes and dislikes are mutually exclusive for information, likes and dislikes cannot exist at the same time, so when a like is given, 1+UP i -DOWN i The value of is 2, indicating the user's positive interest. In this case, the interest score will be magnified by 2 times, highlighting the recommendation priority of the information. When there is no like or dislike, the value is 0, indicating the user's negative feedback and it is not recommended to continue recommending the information. When there is no like or dislike, the value is 1, indicating that the user neither likes nor dislikes the information. The score is calculated according to the normal weight, indicating that the user has a neutral attitude towards the information or has no clear positive interest; browsing time is a direct indicator to measure the user's attention to information, but the simple browsing time may be affected by the length of the information. Therefore, in order to remove the interference of information length on browsing time, the browsing time and the number of words in the information are normalized to more accurately reflect the user's interest in the content of the information rather than the influence of the length of the information; similarly, comments are further participation of users in the information. The more words in the comments, the higher the user's participation and attention to the information. Therefore, CW i Used to measure user engagement and To eliminate the impact of information length and make the rating more fair. Taking into account the user's interactive behavior and information source, the interest rating can more accurately reflect the user's interests, thereby achieving more efficient information push.

[0060] Obtain the information type and the information interest score, calculate the interest type score of the user information chain, and set the information types before the interest type score as the chain information type; obtain the information type and the information interest score, calculate the interest type score of the user information chain, and the calculation logic of the interest type score is: obtain all information types of all information in the user information chain, assign the information interest score to the information type corresponding to the i-th information identifier, add up the information interest scores of all the same information types in the user information chain, and obtain the interest type score of each information type, wherein each information includes at least one information type. Specifically, the core goal of the above logic is to calculate the user's interest intensity in different information types, evaluate the user's preference for each information type by comprehensively evaluating the interest score of each information in the user's historical behavior data, and then determine the several information types (chain information types) that the user is most interested in. Through this calculation method, the system can identify the user's interest preferences and provide a basis for subsequent personalized recommendations.

[0061] The present invention is further configured to calculate the promotion matching coefficient according to the user information chain, including:

[0062] When the information to be promoted has not been accessed by any user, the first promotion matching coefficient is calculated based on the matching degree between the information type of the information to be promoted and the chain information type of the user information chain; specifically, the promotion matching coefficient is calculated by comparing the matching degree between the information type of the information and the chain information type in the user information chain. The matching degree is measured by vectorizing the information type and calculating the similarity of the vectors. The vectorized information type can better reflect the semantic characteristics of the information;

[0063] When the information to be promoted is accessed by any user, the user who has accessed the information to be promoted is marked as the accessing user, and the remaining users are marked as the users to be promoted. The user information chain of the accessing user is obtained, and the type matching coefficient is calculated based on the matching degree between the information type of the information to be promoted and the chain information type of the user information chain of the user to be promoted;

[0064] According to all the chain information types of users whose information interest scores of the visiting users for the information to be promoted are greater than the information interest score threshold, an access chain information type set is set, and a user matching coefficient of the chain information type of the user information chain of the user to be promoted and the access chain information type set is calculated;

[0065] A second promotion matching coefficient is calculated according to the type matching coefficient and the user matching coefficient.

[0066] The present invention is further configured to use a Word2Vec model to map words in each information type to a vector space of a fixed dimension, generate an embedding vector corresponding to each information type, and when vectorizing the chain information type of the user information chain and the information type of the information to be promoted, add the embedding vectors of all words in the information type to obtain the vector representation of the chain information type of the user information chain and the information type of the information to be promoted, wherein, since the words in each information type are mapped to a vector space of a fixed dimension, the dimension of the vector representation is the number of all information types;

[0067] When performing the first promotion matching coefficient, the type matching coefficient and the user matching coefficient, it is achieved by calculating the cosine similarity between the vectors; the second promotion matching coefficient is calculated by weighted summing the type matching coefficient and the user matching coefficient, and the weights are all greater than 0, and the sum is 1. Specifically, the core goal of the above calculation logic is to achieve accurate promotion of information through promotion matching coefficients. The promotion matching coefficient is calculated by the degree of matching between the information type in the user information chain and the information type to be promoted, ensuring that the promoted information can be more in line with the user's interests, thereby improving the promotion effect. The logic also introduces the Word2Vec model to vectorize the information type, making the matching calculation more accurate. The information type is vectorized through the Word2Vec model, and the cosine similarity is used for matching, so that the promotion of information can more accurately match the user's interests and ensure the relevance of the promoted information.

[0068] The present invention is further configured to obtain information promotion data of nodes in real time, including: number of information visitors, forwarding rate, average sequence number of information in user information chain, like rate, comment rate and collection rate;

[0069] The promotion influence of computing nodes includes:

[0070] The promotion breadth is calculated according to the number of information visitors and the forwarding rate; the present invention is further configured such that the calculation logic of the promotion breadth is: Among them, Rad is the promotion breadth, m is the information in the node, Gacc i is the number of people who visited the i-th information, Gsh i is the forwarding rate of the ith information; specifically, the promotion breadth is mainly measured by the number of visitors and the forwarding rate. The more visitors and the higher the forwarding rate, the wider the dissemination range of the information, thereby enhancing the influence of the promotion node.

[0071] The promotion depth is calculated based on the sequence number, like rate, comment rate and collection rate of the information in the user information chain; the calculation logic of the promotion depth is: Among them, Dadv is the promotion depth, Num iis the average sequence number of the i-th information in the user information chain, L i is the like rate of the i-th information, C i is the comment rate of the i-th information, F i is the collection rate of the ith information; specifically, the promotion depth is measured by the position of the information in the user chain, the like rate, the comment rate and the collection rate, focusing on the deep interaction of users on the information, and through the weighted accumulation of these interactive behaviors, the in-depth dissemination effect of the information is obtained.

[0072] The promotion influence is calculated according to the promotion breadth and the promotion depth. The promotion breadth and the promotion depth are normalized, and the weighted sum is calculated to obtain the promotion influence, and the weights are all greater than 0, and the sum is 1. Specifically, the core goal of the above calculation logic is to evaluate the promotion influence of the promotion node through the user's behavior data in the information chain. The influence includes two key dimensions: promotion breadth and promotion depth. The promotion breadth focuses on the number of visitors and forwarding rate of the information, reflecting the scope of information dissemination; the promotion depth focuses on the interaction between users and information, reflecting the attractiveness and depth of the information. Finally, the overall promotion influence of the promotion node is calculated by normalizing and weighted summing the promotion breadth and promotion depth. By combining the calculation logic of promotion breadth and promotion depth, the system can more comprehensively evaluate the actual influence of the promotion node, so as to make more accurate choices in promotion decisions. Promotion breadth ensures that the information is widely disseminated, while promotion depth ensures that the information generates deep interaction among users. The combination of the two not only increases the exposure rate of information, but also improves the user's participation and interactivity, thereby improving the overall effect of information promotion.

[0073] The present invention is further configured to synchronize node information of high-quality information according to promotion influence and quality score in combination with the quality score of the information in the node, including:

[0074] The quality score of information is calculated based on the number of visitors, forwarding rate, like rate, comment rate and collection rate of the information. The calculation logic is: Qua i =Gacc i ·(1+Gsh i +L i +C i +F i ), where Qua i Score the quality of the i-th information;

[0075] The comprehensive score of the information is calculated based on the promotion influence and quality score. The calculation logic is: Sco i =Qua i *Inf i , among which Sco i is the comprehensive score of the i-th information, Infi is the promotion influence of the i-th information;

[0076] Information whose comprehensive score is greater than the comprehensive score threshold is set as high-quality information. Specifically, the above calculation logic is intended to calculate the quality score and comprehensive score of information through the interaction data between users and information, including the number of visitors, forwarding rate, like rate, comment rate, and collection rate, so as to identify high-quality and high-quality information. The quality score evaluates the overall quality of information through the number of visitors and the user's interaction data. The number of visitors reflects the basic audience size of the information. The more visitors, the higher the quality score of the information; the interaction data reflects the user's recognition and interest in the information. The higher the value of each interactive data item, the higher the user's recognition of the information, thereby improving the quality score. The comprehensive score is calculated by multiplying the quality score of the information and the promotion influence. The promotion influence is a weighted score of the previously calculated promotion breadth and depth, which is used to reflect the scope and depth of the spread of information among users.

[0077] The present invention is further configured to synchronize node information of high-quality information, including: obtaining a node embedding vector of a node where the high-quality information to be synchronized is located, and the logic for obtaining the node embedding vector is: obtaining embedding vectors of all users in the node, adding and averaging the embedding vectors of all users to obtain the node embedding vector;

[0078] Calculate the matching degree of the node embedding vectors of the users of the remaining nodes and the nodes where the high-quality information to be synchronized is located, which is recorded as the node-user similarity. The calculation method is to calculate the cosine similarity between the vectors.

[0079] Obtain the user overlap rate of the remaining nodes and the node where the high-quality information to be synchronized is located, recorded as the node user overlap rate;

[0080] When the node user similarity between the remaining nodes and the node where the high-quality information to be synchronized is located is greater than the similarity threshold and the node user overlap rate is less than the overlap threshold, information synchronization is performed. Specifically, the high-quality information is propagated among multiple nodes through the node information synchronization mechanism. To ensure that high-quality information can be shared between nodes with similar user characteristics, the similarity and user overlap rate between nodes are first calculated. When the similarity reaches a certain threshold and the user overlap rate is low, information synchronization is performed to expand the scope of influence of high-quality information while avoiding repeated promotion among the same users; through node embedding vector and cosine similarity calculation, the system can push high-quality information to nodes with similar user characteristics, ensuring that the promotion content is more in line with user interests and achieving accurate promotion; by controlling the user overlap rate, the system can avoid repeated information push between the same users, improve user experience, and reduce interference caused by information redundancy; the node information synchronization mechanism can automatically propagate high-quality information between qualified nodes, thereby expanding the coverage of information and improving the efficiency of information dissemination.

[0081] The present invention is further configured to include: before promoting information, matching the historical behavior data of the user, and not promoting the information if the information has been visited in the past. Specifically, before promoting information, by matching the historical behavior data of the user, the user who has visited the information is filtered out to avoid repeated promotion to the user.

[0082] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0083] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0084] In this application, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can be represented by: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple.

[0085] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0086] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0087] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0088] In the several embodiments provided in the present application, it should be understood that the disclosed system can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0089] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0090] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0091] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks or optical disks.

[0092] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A decentralized information promotion method based on big data, characterized in that: include: Obtaining the user's historical behavior data, analyzing and generating a user information chain, wherein the user information chain includes an information identifier, an information interest score, a chain information type, and a pointer, wherein the pointer points to the next accessed information according to the order in which the information is accessed; For the information to be promoted, a promotion matching coefficient is calculated according to the user information chain, and the information to be promoted is promoted to users whose promotion matching coefficient is greater than a threshold; In the information promotion system, a decentralized network is set up, and any node in the decentralized network can perform information promotion tasks, obtain the node's information promotion data in real time, calculate the node's promotion influence, sort all nodes in descending order according to the node's promotion influence, and combine the quality score of the information in the node to synchronize node information with high-quality information according to the promotion influence and quality score.

2. The method for decentralized information promotion based on big data according to claim 1, characterized in that: Obtain the user's historical behavior data and analyze and generate the user information chain, including: Obtaining historical behavior data of users, including information identifiers, timestamps, information types, and interaction data of historically browsed information. The interaction data includes browsing time, information sources, and interaction data. The information sources include search acquisition and promotion acquisition. The interaction data includes likes, comments, favorites, and dislikes. Within a preset time window, the information identifiers are sorted in order according to the order of timestamps to generate a user information chain; Calculate the information interest score according to browsing time, information source and interaction data, and set the information interest score as an attribute of the information identifier; The information type and the information interest score are obtained, the interest type score of the user information chain is calculated, and the information types with the top N interest type scores are set as chain information types.

3. The decentralized information promotion method based on big data according to claim 2 is characterized in that: The information interest score is calculated based on browsing time, information source and interaction data. The calculation logic of the information interest score is: Among them, Sii(IList i ) is the information interest score of the information corresponding to the i-th information identifier in the user information chain IList, ISo i is the information source of the information corresponding to the i-th information identifier. When the information source is search acquisition, the value is 1. When the information source is promotion acquisition, the value is 0.

8. i and DOWN i is the likes and dislikes of the information corresponding to the i-th information identifier, with a value of 0 or 1. It is 1 when there is a like or dislike, and 0 when there is no like or dislike. i is the browsing time of the information corresponding to the i-th information identifier, IW i is the number of words of the information corresponding to the i-th information identifier, CW i is the number of words in the comments of the information corresponding to the i-th information identifier, β1 and β2 are weight coefficients, both greater than 0; The information type and the information interest score are obtained, and the interest type score of the user information chain is calculated. The calculation logic of the interest type score is as follows: all information types of all information in the user information chain are obtained, and the information type of the information corresponding to the i-th information identifier is assigned the information interest score. The information interest scores of all the same information types in the user information chain are added together to obtain the interest type score of each information type, wherein each information includes at least one information type.

4. The decentralized information promotion method based on big data according to claim 1 is characterized in that: Calculating the promotion matching coefficient according to the user information chain includes: When the information to be promoted has not been accessed by any user, a first promotion matching coefficient is calculated based on the matching degree between the information type of the information to be promoted and the link information type of the user information chain; When the information to be promoted is accessed by any user, the user who has accessed the information to be promoted is marked as the accessing user, and the remaining users are marked as the users to be promoted. The user information chain of the accessing user is obtained, and the type matching coefficient is calculated based on the matching degree between the information type of the information to be promoted and the chain information type of the user information chain of the user to be promoted; According to all the chain information types of users whose information interest scores of the visiting users for the information to be promoted are greater than the information interest score threshold, a set of access chain information types is set, and a user matching coefficient of the chain information type of the user information chain of the user to be promoted and the set of access chain information types is calculated; A second promotion matching coefficient is calculated according to the type matching coefficient and the user matching coefficient.

5. The method for decentralized information promotion based on big data according to claim 4, characterized in that: Use the Word2Vec model to map the words in each information type to a vector space with a fixed dimension, generate an embedding vector corresponding to each information type, and when vectorizing the chain information type of the user information chain and the information type of the information to be promoted, add the embedding vectors of all words in the information type to obtain the vector representation of the chain information type of the user information chain and the information type of the information to be promoted, where the words in each information type are mapped to a vector space with a fixed dimension, so the dimension of the vector representation is the number of all information types; When performing the first promotion matching coefficient, the type matching coefficient and the user matching coefficient, it is achieved by calculating the cosine similarity between vectors; the second promotion matching coefficient is calculated by weighted summing the type matching coefficient and the user matching coefficient, the weights are all greater than 0, and the sum is 1.

6. The method for decentralized information promotion based on big data according to claim 1, characterized in that: Obtain information promotion data of nodes in real time, including: number of information visitors, forwarding rate, average sequence number of information in user information chain, like rate, comment rate and collection rate; The promotion influence of computing nodes includes: Calculate the promotion breadth based on the number of people who visit the information and the forwarding rate; Calculate the promotion depth based on the information's sequence number, like rate, comment rate, and collection rate in the user's information chain; The promotion influence is calculated according to the promotion breadth and the promotion depth.

7. The decentralized information promotion method based on big data according to claim 1 is characterized in that: The calculation logic of the promotion breadth is: Among them, Rad is the promotion breadth, m is the information in the node, Gacc i is the number of people who visited the i-th information, Gsh i is the forwarding rate of the i-th information; The calculation logic of the promotion depth is: Among them, Dadv is the promotion depth, Num i is the average sequence number of the i-th information in the user information chain, L i is the like rate of the i-th information, C i is the comment rate of the i-th information, F i is the collection rate of the i-th information; The promotion breadth and the promotion depth are normalized, and weighted sum is performed to obtain the promotion influence, where the weights are all greater than 0 and the sum is 1.

8. The method for decentralized information promotion based on big data according to claim 7, characterized in that: Combined with the quality score of the information in the node, the high-quality information is synchronized with the node information according to the promotion influence and quality score, including: The quality score of information is calculated based on the number of visitors, forwarding rate, like rate, comment rate and collection rate of the information. The calculation logic is: Qua i =Gacc i ·(1+Gsh i +L i +C i +F i ), where Qua i Score the quality of the i-th information; The comprehensive score of the information is calculated based on the promotion influence and quality score. The calculation logic is: Sco i =Qua i *Inf i , among which Sco i is the comprehensive score of the i-th information, Inf i is the promotion influence of the i-th information; Information with a comprehensive score greater than the comprehensive score threshold is set as high-quality information.

9. The method for decentralized information promotion based on big data according to claim 8, characterized in that: Synchronizing node information of high-quality information, including: obtaining a node embedding vector of the node where the high-quality information to be synchronized is located, and the logic for obtaining the node embedding vector is: obtaining the embedding vectors of all users in the node, adding and averaging the embedding vectors of all users to obtain the node embedding vector; Calculate the matching degree of the node embedding vectors of the users of the remaining nodes and the nodes where the high-quality information to be synchronized is located, which is recorded as the node user similarity. The calculation method is to calculate the cosine similarity between the vectors. Obtain the user overlap rate of the remaining nodes and the node where the high-quality information to be synchronized is located, which is recorded as the node user overlap rate; When the node user similarity between the remaining nodes and the node where the high-quality information to be synchronized is located is greater than the similarity threshold and the node user overlap rate is less than the overlap threshold, information synchronization is performed.

10. The method for decentralized information promotion based on big data according to claim 1, characterized in that: Also includes: Before promoting information, the user's historical behavior data is matched. If the information has been visited in the past, it will no longer be promoted.

Images