Intelligent network data screening system and method based on computer

Abstract

The invention discloses an intelligent network data screening system and method based on a computer, and relates to the technical field of network data screening. According to the method, the relation between the text nodes is established by fully mining semantic similarity, and the accuracy and accuracy of text classification are improved by fully utilizing the corresponding relation and frequency between semantic information and label information; extracting video information released by other users who pay attention to the main user, enriching personal consumption interests of the main user through the living group of the main user, and improving the richness of the screened consumption interests D of the main user; the consumption interests D of the main user are embedded into the layers one by one, labels which are interested by the user and have conversion values are obtained after screening, dependency in the labels is captured with the same layer as the horizontal direction and different layers as the vertical direction, and the system stability when the label dependency relationship is incomplete or missing is ensured; and the accuracy of screening the labels which are interested by the user and have the conversion value is further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of network data screening, in particular to a computer-based intelligent network data screening system and method. Background Art

[0002] In the field of digital marketing, accurate target customer positioning and accurate screening of network data can help companies better understand consumer needs and preferences and develop more effective marketing strategies. Data mining plays a vital role in digital marketing. It runs through multiple processes of digital marketing, such as: data collection and preprocessing, customer segmentation and positioning, market trend analysis and prediction, marketing strategy customization and optimization, personalized recommendations, and marketing and marketing effect evaluation.

[0003] The interactive behavior data generated by the main user is collected and analyzed, and the videos and content that the user is most interested in are determined from the multiple videos browsed by the user. The products of interest are selected for the customer based on the content. The advertisements of the products are carried out using videos with copy and emotions, so that the videos that the customers are interested in are pushed to the users, which encourages the users to purchase the products in the advertisements and improves the information recall rate of the advertisements.

[0004] In the lives of users, items of interest that focus more on spiritual experience and consumer interests that focus more on basic life needs often exist at the same time. Users are more likely to buy items directed by consumer interests, but the profits of the items are lower. Users are less likely to buy items directed by interests, but the profits of the items are higher. Users will not buy items directed by interests frequently, but they will not always buy enough. Therefore, it is very important to set the push frequency of interest-directed items and consumer interest-directed items. Pushing more or less of a certain type of item is not conducive to improving the information recall rate of the advertisement. The existing technology cannot reasonably screen out tags that users are interested in with conversion value, resulting in a low information recall rate of the advertisement. Summary of the invention

[0005] The purpose of the present invention is to provide a computer-based intelligent network data screening system and method to solve the problems raised in the prior art.

[0006] To achieve the above object, the present invention provides the following technical solution: a computer-based intelligent network data screening method, the method comprising: Step S1, obtaining the interaction behavior data generated by the primary user; Step S2: Analyze the interactive behavior data generated by the primary user and filter out videos that the user is interested in; Step S3: Perform content analysis on the acquired video to filter out tags that have conversion value and are of interest to the main user.

[0007] The step S2 analyzes the interactive behavior data generated by the primary user and screens out videos that the primary user is interested in, including: Step S21, assign values ​​to the browsing progress, browsing times, likes, favorites and comments of the video: the browsing progress range is 0-1; when the browsing times are 1-2, it is recorded as 0.2, when the browsing times are 3-5, it is recorded as 0.5, and when the browsing times are greater than 5, it is recorded as 1; no likes are recorded as 0, and liked are recorded as 0.1; no favorites are recorded as 0, and favorites are recorded as 0.2; no comments are recorded as 0, when the number of words in the comment is 1-3, it is recorded as 0.1, when the number of words in the comment is 4-10, it is recorded as 0.3, and when the number of words in the comment is greater than 10, it is recorded as 0.5; Step S22: Accumulate the values ​​corresponding to the browsing progress, browsing times, likes, favorites and comments to obtain the interest value of the main user. If the interest value of the main user exceeds the interest threshold, the video is determined to be a video of interest to the main user.

[0008] The step S3 of performing content analysis on the acquired video to obtain tags that the primary user is interested in includes: Step S31, analyzing the interactive behavior data generated by users browsing other people's videos, and filtering the tags M={tag M1, tag M2, ..., tag Mn} that the main user is interested in in the video; Step S32, extracting information from videos posted by the main user himself, extracting information from videos posted by other users who follow the main user, and screening the main user's consumption interests D; Step S33: Considering the tags that the main user is interested in and the main user's consumption interest D at the same time, filters the tags that the user is interested in and have conversion value.

[0009] The step S31 comprises: Step S311, extracting text contained in the video content through sentences and voices in the video to form a text data set X={text 1, text 2, ..., text n}, which represents the text with tag number n, and inputting the text data set into the trained text classification network to obtain the classification results of the text classification network and the prediction output; Step S312: Nodes and word nodes are taken as two types of nodes and word-word, word-text, and text-text are taken as three types of edges; the text contains a word set C={word c 1 , word c 2 , …, word c m}, use the cosine similarity formula to calculate the two text nodes x i 、x j If the cosine similarity between two text nodes is greater than k 1, then add a straight line between the two text nodes and use the cosine similarity value as the weight between the text nodes; if the cosine similarity between the two text nodes is less than or equal to k 1 , then no straight line is added between the two text nodes; A word c i The probability of appearing on the edge is related to another word c j The product of the probabilities of appearing on the edge is P1. In the corpus, word c i and word c j The joint probability of occurrence is P2. The value of the logarithm of P2 with P1 as the base is calculated. When the value of the logarithm of P2 with P1 as the base is greater than 0, it means that the correlation between the two word nodes is high. This value is used as the weight between the two word nodes, and a straight line is added between the two words. When the value of the logarithm of P2 with P1 as the base is less than or equal to 0, it means that the correlation between the two word nodes is low, and a straight line is not added between the two words. The frequency of a word node appearing in a text node is recorded as P3. If P3 is greater than k 2 , then the word is of high importance to the text, P3 is used as the weight between the word node and the text node, and a straight line is added between the text and the word; if P3 is less than or equal to k 2 , then the importance of the word to the text is low, and a straight line is not added between the text and the word. The weights of the edges between the nodes are converted into an adjacency matrix graph; by fully mining the semantic similarity to establish the connection between text nodes, the connection between text and words, the connection between text and text, and the connection between words are improved, and the structural information between each node is fully utilized, and the correspondence and frequency between semantic information and label information are fully utilized to solve the over-smoothing problem, improve the performance of multi-level text classification, and improve the precision and accuracy of text classification.

[0010] Step S313: Use a multi-layer graph convolutional network to extract adjacency matrix graph information, obtain nodes as key nodes after multiple iterations, represent the text of the key nodes using the text of the data set, mark the labels corresponding to the text of the data set, and obtain the labels M={label M1, label M2, …, label Mn} that the user is interested in.

[0011] The step S32 comprises: Step S321, information extraction includes extracting the shooting environment, clothing, language, text and music in the video, extracting the purchase order information of the main user, and obtaining the main user's consumer interest A={consumer interest A1, consumer interest A2, ..., consumer interest An} and other users' consumer interest B={consumer interest B1, consumer interest B2, ..., consumer interest Bn} by analyzing the shooting environment, clothing, language, text and music in the video, sorting the subsets in consumer interest B according to the intensity of consumer interest, the intersection of consumer interest A and consumer interest B is consumer interest C, marking the subset of consumer interest B corresponding to consumer interest C in consumer interest B, recording it as the central subset, and then marking the subset of consumer interest B adjacent to the central subset, and recording the union of the marginal subset, the central subset and consumer interest A as the consumer interest D of the main user. Extract information from videos posted by other users who follow each other with the main user, use the consumption interests of other users who follow each other with the main user to expand and filter the main user's consumption interests, enrich the main user's personal consumption interests through the main user's life group, and improve the richness of the filtered main user's consumption interests D.

[0012] The step S33 comprises: Step S331: Layer and assign weights to tags of interest to users according to the types of tags, embed the main user's consumption interests D into the layers one by one, and assign weights to the main user's consumption interests D one by one according to the intensity of consumption interests; use the same layer as the horizontal direction and different layers as the vertical direction to capture the dependencies in the tags respectively, ensure the system stability when the tag dependency is incomplete or missing, and further improve the accuracy of screening tags of interest to users with conversion value; Step S332: In the same layer, the weight of the tag of interest to the user and the corresponding weight of the consumer interest D assigned to the main user are added together to obtain a comparison value. If the comparison value exceeds a preset threshold, it is a tag of interest to the user with conversion value.

[0013] A computer-based intelligent network data screening system, the system comprises a data acquisition module, a data screening module 1 and a data screening module 2 connected to each other; The data acquisition module is used to obtain user-generated interactive behavior data through the device used by the user. The user-generated interactive behavior data includes videos posted by other users and browsed by the main user, videos posted by the main user himself, and the interactive behavior log between the main user and others.

[0014] The data screening module 1 is used to analyze the interactive behavior data generated by the main user and screen out videos that the user is interested in; The data screening module 2 is used to perform content analysis on the acquired videos and screen out tags that have conversion value and are of interest to the main user.

[0015] The data screening module 1 comprises: The main user's interest value calculation module is used to assign values ​​to the browsing progress, number of views, likes, collections and comments of the video, and accumulate the corresponding values ​​of the browsing progress, number of views, likes, collections and comments to obtain the main user's interest value; The main user's interest value screening module is used to screen the main user's interest value. If the main user's interest value exceeds an interest threshold, the video is determined to be a video that the main user is interested in.

[0016] The data screening module 2 comprises: The interest tag extraction module is used to analyze the interactive behavior data generated by users browsing other people's videos and filter the tags M={tag M1, tag M2, ..., tag Mn} that the main user is interested in in the video; The consumption interest extraction module is used to extract information from videos posted by the main user himself, and to extract information from videos posted by other users who follow the main user, so as to screen the consumption interest D of the main user; The value tag extraction module is used to simultaneously consider the tags that the main user is interested in and the main user's consumption interest D, and filter the tags that the user is interested in with conversion value.

[0017] Compared with the prior art, the present invention has the following beneficial effects: The present invention establishes the connection between text nodes by fully mining semantic similarity, improves the connection between text and words, the connection between text and text, and the connection between words, fully utilizes the structural information between each node, fully utilizes the correspondence and frequency between semantic information and label information, solves the over-smoothing problem, improves the performance of multi-level text classification, and improves the precision and accuracy of text classification; extracts information from videos posted by other users who follow each other with the main user, expands and filters the main user's consumer interests by using the consumer interests of other users who follow each other with the main user, enriches the main user's personal consumer interests through the main user's life group, and improves the richness of the main user's consumer interests D that are filtered out; embeds the main user's consumer interests D into the layers one by one, obtains user-interested tags with conversion value after filtering, captures the dependencies in the tags respectively with the same layer as the horizontal direction and different layers as the vertical direction, ensures the system stability when the tag dependency is incomplete or missing, and further improves the accuracy of filtering user-interested tags with conversion value. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A flowchart of a computer-based intelligent network data screening method of the present invention; Figure 2The present invention is a composition diagram of a computer-based intelligent network data screening system. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not 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.

[0020] Example: Figure 1 As shown, the present invention provides a computer-based intelligent network data screening method, the method comprising: Step S1, obtaining the interaction behavior data generated by the primary user; Step S2: Analyze the interactive behavior data generated by the primary user and filter out videos that the user is interested in; Step S3: Perform content analysis on the acquired video to filter out tags that have conversion value and are of interest to the main user.

[0021] The step S2 analyzes the interactive behavior data generated by the primary user and screens out videos that the primary user is interested in, including: Step S21, assign values ​​to the browsing progress, browsing times, likes, favorites and comments of the video: the browsing progress range is 0-1; when the browsing times are 1-2, it is recorded as 0.2, when the browsing times are 3-5, it is recorded as 0.5, and when the browsing times are greater than 5, it is recorded as 1; no likes are recorded as 0, and liked are recorded as 0.1; no favorites are recorded as 0, and favorites are recorded as 0.2; no comments are recorded as 0, when the number of words in the comment is 1-3, it is recorded as 0.1, when the number of words in the comment is 4-10, it is recorded as 0.3, and when the number of words in the comment is greater than 10, it is recorded as 0.5; Step S22: Accumulate the values ​​corresponding to the browsing progress, browsing times, likes, favorites and comments to obtain the interest value of the main user. If the interest value of the main user exceeds the interest threshold, the video is determined to be a video of interest to the main user.

[0022] The step S3 of performing content analysis on the acquired video to obtain tags that the primary user is interested in includes: Step S31, analyzing the interactive behavior data generated by users browsing other people's videos, and filtering the tags M={tag M1, tag M2, ..., tag Mn} that the main user is interested in in the video; Step S32, extracting information from videos posted by the main user himself, extracting information from videos posted by other users who follow the main user, and screening the main user's consumption interests D; Step S33: Considering the tags that the main user is interested in and the main user's consumption interest D at the same time, filters the tags that the user is interested in and have conversion value.

[0023] The step S31 comprises: Step S311, extracting text contained in the video content through sentences and voices in the video to form a text data set X={text 1, text 2, ..., text n}, which represents the text with tag number n, and inputting the text data set into the trained text classification network to obtain the classification results of the text classification network and the prediction output; Step S312: Nodes and word nodes are taken as two types of nodes and word-word, word-text, and text-text are taken as three types of edges; the text contains a word set C={word c 1 , word c 2 , …, word c m}, use the cosine similarity formula to calculate the two text nodes x i 、x j If the cosine similarity between two text nodes is greater than k 1 , then add a straight line between the two text nodes and use the cosine similarity value as the weight between the text nodes; if the cosine similarity between the two text nodes is less than or equal to k 1 , then no straight line is added between the two text nodes; A word c i The probability of appearing on the edge is related to another word c j The product of the probabilities of appearing on the edge is P1. In the corpus, word c i and word c j The joint probability of occurrence is P2. The value of the logarithm of P2 with P1 as the base is calculated. When the value of the logarithm of P2 with P1 as the base is greater than 0, it means that the correlation between the two word nodes is high. This value is used as the weight between the two word nodes, and a straight line is added between the two words. When the value of the logarithm of P2 with P1 as the base is less than or equal to 0, it means that the correlation between the two word nodes is low, and a straight line is not added between the two words. The frequency of a word node appearing in a text node is recorded as P3. If P3 is greater than k 2 , then the word is of high importance to the text, P3 is used as the weight between the word node and the text node, and a straight line is added between the text and the word; if P3 is less than or equal to k 2, then the importance of the word to the text is low, and a straight line is not added between the text and the word. The weights of the edges between the nodes are converted into an adjacency matrix graph; by fully mining the semantic similarity to establish the connection between text nodes, the connection between text and words, the connection between text and text, and the connection between words are improved, and the structural information between each node is fully utilized, and the correspondence and frequency between semantic information and label information are fully utilized to solve the over-smoothing problem, improve the performance of multi-level text classification, and improve the precision and accuracy of text classification.

[0024] Step S313: Use a multi-layer graph convolutional network to extract adjacency matrix graph information, obtain nodes as key nodes after multiple iterations, represent the text of the key nodes using the text of the data set, mark the labels corresponding to the text of the data set, and obtain the labels M={label M1, label M2, …, label Mn} that the user is interested in.

[0025] The step S32 comprises: Step S321, information extraction includes extracting the shooting environment, clothing, language, text and music in the video, extracting the purchase order information of the main user, and obtaining the main user's consumer interest A={consumer interest A1, consumer interest A2, ..., consumer interest An} and other users' consumer interest B={consumer interest B1, consumer interest B2, ..., consumer interest Bn} by analyzing the shooting environment, clothing, language, text and music in the video, sorting the subsets in consumer interest B according to the intensity of consumer interest, the intersection of consumer interest A and consumer interest B is consumer interest C, marking the subset of consumer interest B corresponding to consumer interest C in consumer interest B, recording it as the central subset, and then marking the subset of consumer interest B adjacent to the central subset, and recording the union of the marginal subset, the central subset and consumer interest A as the consumer interest D of the main user. Extract information from videos posted by other users who follow each other with the main user, use the consumption interests of other users who follow each other with the main user to expand and filter the main user's consumption interests, enrich the main user's personal consumption interests through the main user's life group, and improve the richness of the filtered main user's consumption interests D.

[0026] The step S33 comprises: Step S331: Layer and assign weights to tags of interest to users according to the types of tags, embed the main user's consumption interests D into the layers one by one, and assign weights to the main user's consumption interests D one by one according to the intensity of consumption interests; use the same layer as the horizontal direction and different layers as the vertical direction to capture the dependencies in the tags respectively, ensure the system stability when the tag dependency is incomplete or missing, and further improve the accuracy of screening tags of interest to users with conversion value; Step S332: In the same layer, the weight of the tag of interest to the user and the corresponding weight of the consumer interest D assigned to the main user are added together to obtain a comparison value. If the comparison value exceeds a preset threshold, it is a tag of interest to the user with conversion value.

[0027] Figure 2 , a computer-based intelligent network data screening system, the system includes a data acquisition module, a data screening module 1 and a data screening module 2 connected to each other; The data acquisition module is used to obtain user-generated interactive behavior data through the device used by the user. The user-generated interactive behavior data includes videos posted by other users and browsed by the main user, videos posted by the main user himself, and the interactive behavior log between the main user and others.

[0028] The data screening module 1 is used to analyze the interactive behavior data generated by the main user and screen out videos that the user is interested in; The data screening module 2 is used to perform content analysis on the acquired videos and screen out tags that have conversion value and are of interest to the main user.

[0029] The data screening module 1 comprises: The main user's interest value calculation module is used to assign values ​​to the browsing progress, number of views, likes, collections and comments of the video, and accumulate the corresponding values ​​of the browsing progress, number of views, likes, collections and comments to obtain the main user's interest value; The main user's interest value screening module is used to screen the main user's interest value. If the main user's interest value exceeds an interest threshold, the video is determined to be a video that the main user is interested in.

[0030] The data screening module 2 comprises: The interest tag extraction module is used to analyze the interactive behavior data generated by users browsing other people's videos and filter the tags M={tag M1, tag M2, ..., tag Mn} that the main user is interested in in the video; The consumption interest extraction module is used to extract information from videos posted by the main user himself, and to extract information from videos posted by other users who follow the main user, so as to screen the consumption interest D of the main user; The value tag extraction module is used to simultaneously consider the tags that the main user is interested in and the main user's consumption interest D, and filter the tags that the user is interested in with conversion value.

[0031] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A computer-based intelligent network data screening method, characterized in that: The method includes: Step S1, obtaining the interaction behavior data generated by the primary user; Step S2: Analyze the interactive behavior data generated by the primary user and filter out videos that the user is interested in; Step S3: Perform content analysis on the acquired video to filter out tags that have conversion value and are of interest to the main user.

2. The computer-based intelligent network data screening method according to claim 1, characterized in that: The step S2 analyzes the interactive behavior data generated by the primary user and screens out videos that the primary user is interested in, including: Step S21, assigning values ​​to the browsing progress, browsing times, likes, collections and comments of the video; Step S22: Accumulate the values ​​corresponding to the browsing progress, browsing times, likes, favorites and comments to obtain the interest value of the main user. If the interest value of the main user exceeds the interest threshold, the video is determined to be a video of interest to the main user.

3. The computer-based intelligent network data screening method according to claim 1, characterized in that: The step S3 of performing content analysis on the acquired video to obtain tags that the primary user is interested in includes: Step S31, analyzing the interactive behavior data generated by users browsing other people's videos, and filtering the tags M={tag M1, tag M2, ..., tag Mn} that the main user is interested in in the video; Step S32, extracting information from videos posted by the main user himself, extracting information from videos posted by other users who follow the main user, and screening the main user's consumption interests D; Step S33: Considering the tags that the main user is interested in and the main user's consumption interest D at the same time, filters the tags that the user is interested in and have conversion value.

4. The computer-based intelligent network data screening method according to claim 3, characterized in that: The step S31 comprises: Step S311, extracting text contained in the video content through sentences and voices in the video to form a text data set X={text 1, text 2, ..., text n}, which represents the text with tag number n, and inputting the text data set into the trained text classification network to obtain the classification results of the text classification network and the prediction output; Step S312: Nodes and word nodes are taken as two types of nodes and word-word, word-text, and text-text are taken as three types of edges; the text contains a word set C = {word c1, word c2, ..., word c m }, use the cosine similarity formula to calculate the two text nodes x i 、x j If the cosine similarity between two text nodes is greater than k1, a straight line is added between the two text nodes, and the cosine similarity value is used as the weight between the text nodes; if the cosine similarity between two text nodes is less than or equal to k1, no straight line is added between the two text nodes; A word c i The probability of appearing on the edge is related to another word c j The product of the probabilities of appearing on the edge is P1. In the corpus, word c i and word c j The joint probability of occurrence is P2. The value of the logarithm of P2 with P1 as the base is calculated. When the value of the logarithm of P2 with P1 as the base is greater than 0, it means that the correlation between the two word nodes is high. This value is used as the weight between the two word nodes, and a straight line is added between the two words. When the value of the logarithm of P2 with P1 as the base is less than or equal to 0, it means that the correlation between the two word nodes is low, and a straight line is not added between the two words. The frequency of a word node appearing in a text node is recorded as P3. If P3 is greater than k2, the word is highly important to the text. P3 is used as the weight between the word node and the text node, and a straight line is added between the text and the word. If P3 is less than or equal to k2, the word is not important to the text, and a straight line is not added between the text and the word. The weights of the edges between the nodes are converted into an adjacency matrix graph. Step S313: Use a multi-layer graph convolutional network to extract adjacency matrix graph information, obtain nodes as key nodes after multiple iterations, represent the text of the key nodes using the text of the data set, mark the labels corresponding to the text of the data set, and obtain the labels M={label M1, label M2, …, label Mn} that the user is interested in.

5. The computer-based intelligent network data screening method according to claim 4, characterized in that: The step S32 comprises: Step S321, information extraction includes extracting the shooting environment, clothing, language, text and music in the video, extracting the purchase order information of the main user, and obtaining the main user's consumer interest A={consumer interest A1, consumer interest A2, ..., consumer interest An} and other users' consumer interest B={consumer interest B1, consumer interest B2, ..., consumer interest Bn} by analyzing the shooting environment, clothing, language, text and music in the video, sorting the subsets in consumer interest B according to the intensity of consumer interest, the intersection of consumer interest A and consumer interest B is consumer interest C, marking the subset of consumer interest B corresponding to consumer interest C in consumer interest B, recording it as the central subset, and then marking the subset of consumer interest B adjacent to the central subset, and recording the union of the marginal subset, the central subset and consumer interest A as the consumer interest D of the main user.

6. The computer-based intelligent network data screening method according to claim 5, characterized in that: The step S33 comprises: Step S331: The tags that the user is interested in are layered and weighted according to the types of tags, the consumption interests D of the main user are embedded in the layers one by one, and the consumption interests D of the main user are weighted one by one according to the intensity of the consumption interests; Step S332: In the same layer, the weight of the tag of interest to the user and the corresponding weight of the consumer interest D assigned to the main user are added together to obtain a comparison value. If the comparison value exceeds a preset threshold, it is a tag of interest to the user with conversion value.

7. A computer-based intelligent network data screening system, applied to a computer-based intelligent network data screening method according to any one of claims 1 to 6, characterized in that: The system comprises a data acquisition module, a data screening module 1 and a data screening module 2 which are connected to each other; The data acquisition module is used to obtain user-generated interactive behavior data through the device used by the user. The user-generated interactive behavior data includes videos posted by other users and browsed by the main user, videos posted by the main user himself, and the interactive behavior log between the main user and others.

8. The computer-based intelligent network data screening system according to claim 7, characterized in that: The data screening module 1 is used to analyze the interactive behavior data generated by the main user and screen out videos that the user is interested in; The data screening module 2 is used to perform content analysis on the acquired videos and screen out tags that have conversion value and are of interest to the main user.

9. The computer-based intelligent network data screening system according to claim 8, characterized in that: The data screening module 1 comprises: The main user's interest value calculation module is used to assign values ​​to the browsing progress, number of views, likes, collections and comments of the video, and accumulate the corresponding values ​​of the browsing progress, number of views, likes, collections and comments to obtain the main user's interest value; The main user's interest value screening module is used to screen the main user's interest value. If the main user's interest value exceeds an interest threshold, the video is determined to be a video that the main user is interested in.

10. The computer-based intelligent network data screening system according to claim 9, characterized in that: The data screening module 2 comprises: The interest tag extraction module is used to analyze the interactive behavior data generated by users browsing other people's videos and filter the tags M={tag M1, tag M2, ..., tag Mn} that the main user is interested in in the video; The consumption interest extraction module is used to extract information from videos posted by the main user himself, and to extract information from videos posted by other users who follow the main user, so as to screen the consumption interest D of the main user; The value tag extraction module is used to simultaneously consider the tags that the main user is interested in and the main user's consumption interest D, and filter the tags that the user is interested in with conversion value.

Images