Method and system for content recommendation based on traffic monitoring

Abstract

The application provides a content recommendation method and system based on flow monitoring, and relates to the technical field of content recommendation.The method comprises the following steps: determining relevant content information related to a search keyword; determining a first recommendation list according to the relevance of the relevant content information and the search keyword, the historical browsing volume and the historical interaction quantity; determining a first preset number of to-be-tested content information, obtaining the historical interaction information of the to-be-tested content information, and then determining a key negative factor; determining a second recommendation list according to the historical browsing record of a user, the introduction information of the to-be-tested content information and the key negative factor, and displaying the relevant content information.According to the application, the key negative factor can be found out, the degree of resistance of the user to the key negative factor can be judged, the preferences of the user for various factors can be reflected, the second recommendation list can be obtained, the content information in the second recommendation list is not only related to the search keyword, and the recommendation order takes into account the content information quality and the preferences of the user, and the user experience is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of content recommendation, and in particular to a content recommendation method and system based on traffic monitoring. BACKGROUND

[0002] In the related art, when a user inputs a keyword to search for content, a recommendation list is usually generated according to the relevance of the content to the keyword, and the content is displayed for the user according to the recommendation list. Although this recommendation method can search for content related to the keyword, it does not take into account the user's preference for the content or the quality of the content itself. Therefore, the user may be recommended content that is highly relevant but of poor quality or not of interest to the user, resulting in a poor user experience.

[0003] The information disclosed in the background section of this application is only intended to deepen the understanding of the general background of the application and should not be considered as recognition or in any form as implying that this information constitutes prior art known to those skilled in the art. SUMMARY

[0004] The present application provides a content recommendation method and system based on traffic monitoring, which can solve the technical problem that it is difficult to focus on user preferences and content quality when recommending content for a keyword in the related art.

[0005] According to a first aspect of the present application, a content recommendation method based on traffic monitoring is provided, comprising:

[0006] In the case where a search keyword of a user is received, a plurality of relevant content information related to the search keyword is determined;

[0007] According to the relevance of the plurality of relevant content information to the search keyword, the historical browsing volume and the historical interaction quantity of the plurality of relevant content information, a first recommendation list of the plurality of relevant content information is determined, wherein the historical interaction quantity includes the number of historical evaluation information of the content information;

[0008] In the first recommendation list, a first predetermined number of to-be-tested content information is determined, and historical interaction information of the to-be-tested content information is obtained, wherein the historical interaction information includes the content of the historical evaluation information of the to-be-tested content information;

[0009] According to the historical interaction information, a key negative factor is determined;

[0010] According to the historical browsing record of the user, the introduction information of the to-be-tested content information and the key negative factor, a second recommendation list of the plurality of relevant content information is determined;

[0011] According to the second recommendation list, the plurality of relevant content information is displayed.

[0012] According to the application, in the case of receiving a search keyword of a user, a plurality of relevant content information related to the search keyword is determined, comprising:

[0013] The title, introduction information and attribute tag of each content information are obtained, wherein the attribute tag comprises tag information added by the author of the content information to describe the attribute of the content information;

[0014] The title semantic information of the title, the introduction semantic information of the introduction information and the label semantic information of the attribute tag are obtained through a semantic recognition model;

[0015] The keyword semantic information of the search keyword is obtained through a semantic recognition model;

[0016] For each content information, the first similarity between the keyword semantic information and the title semantic information, the second similarity between the keyword semantic information and the introduction semantic information, and the third similarity between the keyword semantic information and the label semantic information are determined;

[0017] The maximum value of the first similarity, the second similarity and the third similarity is obtained;

[0018] If the maximum value of the content information corresponding to the similarity is greater than or equal to a preset first similarity threshold, the content information is determined as the relevant content information.

[0019] According to the application, according to the relevance of the plurality of relevant content information to the search keyword, the historical browsing volume and the historical interaction quantity of the plurality of relevant content information, a first recommendation list of the plurality of relevant content information is determined, comprising:

[0020] The maximum value of the similarity corresponding to the relevant content information is determined as the relevance of the relevant content information to the search keyword;

[0021] According to the ratio of the historical interaction quantity and the historical browsing volume of the relevant content information, an interaction attraction coefficient of the relevant content information is determined;

[0022] According to the interaction attraction coefficient, the historical interaction quantity, the historical browsing volume and the relevance, a first recommendation coefficient of the relevant content information is determined;

[0023] According to the first recommendation coefficient, a first recommendation list of the plurality of relevant content information is determined.

[0024] According to the application, according to the historical interaction information, a key negative factor is determined, comprising:

[0025] Through an emotion classification model, the emotion classification information of the content of each historical evaluation information of the to-be-tested content information is determined;

[0026] Filtering negative evaluation information from multiple historical evaluation information based on the sentiment classification information;

[0027] Determine the negative semantic information of negative evaluation information through semantic recognition model;

[0028] performing clustering processing on the negative semantic information to obtain a plurality of negative semantic information clusters;

[0029] The cluster centers of multiple negative semantic information clusters are identified as key negative factors.

[0030] According to the present invention, a second recommendation list of multiple related content information is determined based on the user's historical browsing history, the introduction information of the content information to be tested, and the key negative factors, including:

[0031] Determine, based on the user's historical browsing history, the progress of playback of historical content information browsed by the user, the user's interaction history with the historical content information, and the interaction history of the historical content information;

[0032] determining the user's resistance coefficient to the key negative factor based on the playback progress, the user interaction record, the interaction information record, and the key negative factor;

[0033] Determining a publicity negation coefficient of the content information to be tested based on the introduction information of the content information to be tested and the key negative factors;

[0034] Obtain historical interaction information of multiple related content information in the first recommendation list;

[0035] Determining a supplementary recommendation coefficient for each piece of relevant content information based on historical interaction information of the plurality of relevant content information and the key negative factors;

[0036] A second recommendation list of multiple related content information is determined based on the resistance coefficient, the publicity negation coefficient and the supplementary recommendation coefficient.

[0037] According to the present invention, determining the user's resistance coefficient to the key negative factor based on the playback progress, the user interaction record, the interaction information record, and the key negative factor includes:

[0038] According to the interactive information record of the historical content information, select the target historical content information having the i-th key negative factor from the multiple historical content information;

[0039] According to the formula

[0040] ,

[0041] Determine the user's resistance coefficient to the i-th key negative factor wherein, is the played progress of the jth target historical content information existing the ith key negative factor, is the average played progress of the historical content information, is the semantic similarity between the historical evaluation information of the user to the jth target historical content information existing the ith key negative factor and the ith key negative factor, if the user has not made an evaluation to the jth target historical content information existing the ith key negative factor, , is the number of target historical content information existing the ith key negative factor, if() is a conditional function, j≤ , and i, j and k are positive integers.

[0042] According to the application, the promotion negative coefficient of the to-be-tested content information is determined according to the introduction information of the to-be-tested content information and the key negative factor, comprising:

[0043] According to the historical interaction information of the to-be-tested content information, the target to-be-selected content information existing the ith key negative factor is screened;

[0044] The introduction information of the target to-be-tested content information is segmented and the stop words are removed to obtain a plurality of introduction words of the introduction information;

[0045] The introduction sentiment classification information of each introduction word is determined through a sentiment classification model;

[0046] According to the introduction sentiment classification information, the positive introduction words of the introduction words are determined;

[0047] The negative words are added to the positive introduction words to obtain negative introduction words;

[0048] The negative introduction semantic information of the negative introduction words is obtained through a semantic recognition model;

[0049] According to the formula

[0050] ,

[0051] the promotion negative coefficient of the kth target to-be-tested content information to the ith key negative factor is determined wherein, is the negative introduction semantic information of the st negative introduction word of the kth target to-be-tested content information, is the ith key negative factor, is and ​a semantic similarity of and M, i and N are positive integers.

[0052] According to the application, the supplementary recommendation coefficient of each related content information is determined according to the historical interaction information of multiple related content information and the key negative factors, and the supplementary recommendation coefficient of each related content information is determined according to the historical interaction information of multiple related content information and the key negative factors, comprising:

[0053] The interaction sentiment classification information of the historical interaction information is determined through the sentiment classification model.

[0054] According to the interaction sentiment classification information, the positive historical interaction information is screened out from the historical interaction information.

[0055] The negative affix is added to the positive historical interaction information to obtain the modified historical interaction information.

[0056] The modified interaction semantic information of the modified historical interaction information is obtained through the semantic recognition model.

[0057] According to the formula

[0058] ,

[0059] The supplementary recommendation coefficient of the tth related content information to the ith key negative factor is determined , wherein is the modified interaction semantic information of the xth historical interaction information of the tth related content information, is the ith key negative factor, is and a semantic similarity of and M, i and N are positive integers. is the number of historical interaction information of the tth related content information, x≤ , and x and are positive integers.

[0060] According to the application, the second recommendation list of multiple related content information is determined according to the conflict coefficient, the propaganda negative coefficient and the supplementary recommendation coefficient, and the second recommendation list of multiple related content information is determined according to the conflict coefficient, the propaganda negative coefficient and the supplementary recommendation coefficient, comprising:

[0061] If the tth related content information is not the target to-be-tested content information, the gain recommendation coefficient of the tth related content information is determined according to the formula

[0062] ,

[0063] The gain recommendation coefficient of the tth related content information is determined , wherein is the number of positive historical interaction information of the tth related content information, is the number of historical interaction information of the tth related content information, an average positive interaction rate of a plurality of related content information, a supplementary recommendation coefficient of the tth related content information for the ith key negative factor, a resistance coefficient of the user to the ith key negative factor, N is the number of key negative factors, i≤N, and i and N are positive integers;

[0064] If the t related content information is a target to-be-tested content information, the gain recommendation coefficient of the tth related content information is determined according to the formula

[0065]

[0066] determining the gain recommendation coefficient of the tth related content information wherein, a promotion negative coefficient of the tth related content information for the ith key negative factor;

[0067] determining a second recommendation list according to the gain recommendation coefficient of each related content information.

[0068] According to a second aspect of the present application, a content recommendation system based on traffic monitoring is provided, comprising:

[0069] a related content information module configured to determine a plurality of related content information related to a search keyword of a user when the search keyword is received;

[0070] a first recommendation list module configured to determine a first recommendation list of the plurality of related content information according to the relevance of the plurality of related content information to the search keyword, the historical browsing volume and the historical interaction quantity of the plurality of related content information, wherein the historical interaction quantity includes the number of historical evaluation information of the content information;

[0071] a historical interaction information module configured to determine a first preset number of to-be-tested content information in the first recommendation list and obtain historical interaction information of the to-be-tested content information, wherein the historical interaction information includes the content of the historical evaluation information of the to-be-tested content information;

[0072] a key negative factor module configured to determine a key negative factor according to the historical interaction information;

[0073] a second recommendation list module configured to determine a second recommendation list of the plurality of related content information according to the historical browsing record of the user, the introduction information of the to-be-tested content information and the key negative factor;

[0074] a display module configured to display the plurality of related content information according to the second recommendation list.

[0075] By adopting the above technical solutions, the present application can achieve the following technical effects: ​

[0076] According to the present application, after obtaining the relevant content information related to the search keyword, the quality of the relevant content information can be indirectly reflected based on the historical browsing volume and the historical interaction quantity of the relevant content information, so as to recommend the relevant content information in combination with the quality of the relevant content information, find the key negative factors of the to-be-tested content information, judge the resistance degree of the user to the key negative factors, reflect the likes and dislikes of the user to various factors of the content, obtain a second recommendation list, so that the content information in the second recommendation list is not only related to the search keyword, but also the recommendation order takes into account the quality of the content information itself and the likes of the user, and the user experience is improved. When determining the resistance coefficient, the influence of the key negative factors on the playing can be determined through the played progress of the historical content information, so as to judge whether the user has a resistance psychology to the key negative factors, and the intensity of the resistance psychology of the user to the key negative factors can be amplified through the interaction record of the user, so that the resistance psychology of the user to the key negative factors can be accurately reflected, and accurate data basis is provided for recommending the content interested by the user. When determining the propaganda negative coefficient, the negative introduction words can be obtained by adding negative affixes to the positive introduction words in the introduction information, and the semantic similarity between the negative introduction words and the key negative factors is solved, so as to judge whether the to-be-tested content information has false propaganda, so as to reduce the recommendation probability of the content with false propaganda and improve the user experience. When determining the supplementary recommendation coefficient, the negative affixes can be added to the positive historical interaction information in the objective comments of the historical user, and the semantic similarity between the modified historical interaction information and the key negative factors is determined, and then the possibility that the key negative factors do not exist in the relevant content information is determined, and the supplementary recommendation coefficient is obtained, which can be used to improve the probability that the relevant content information without the key negative factors is recommended, so as to improve the user experience. When determining the gain recommendation coefficient, the target to-be-tested content information and other relevant content can be classified and calculated, so that the recommendation coefficient of the target to-be-tested content information with false propaganda is further reduced. And the recommendation coefficient of each relevant content information can be adjusted in combination with the quality of the relevant content information, whether there is false propaganda, whether it contains key negative factors, and the resistance psychology of the user to the key negative factors and other factors, so that the relevant content information with good quality, without false propaganda and without negative key factors can be preferentially recommended, so as to improve the user experience.

[0077] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, rather than limiting the present application. Other features and aspects of the present application will be more apparent from the following detailed description of exemplary embodiments with reference to the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0078] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other embodiments can be obtained from these drawings without creative labor.

[0079] Figure 1 An exemplary flowchart of a content recommendation method based on traffic monitoring according to an embodiment of the present application is shown;

[0080] Figure 2 An exemplary flowchart of searching for related content information according to an embodiment of the present application is shown;

[0081] Figure 3 An exemplary flowchart of determining a first recommendation list according to an embodiment of the present application is shown;

[0082] Figure 4 An exemplary flowchart of determining a key negative factor according to an embodiment of the present application is shown;

[0083] Figure 5 An exemplary flowchart of determining a second recommendation list according to an embodiment of the present application is shown;

[0084] Figure 6 An exemplary block diagram of a content recommendation system based on traffic monitoring according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0085] In order to make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0086] The technical solutions of the present application will be described in detail with specific embodiments. The following specific embodiments can be combined with each other, and some embodiments may not be described again for the same or similar concepts or processes.

[0087] Figure 1 An exemplary flowchart of a content recommendation method based on traffic monitoring according to an embodiment of the present application is shown, the method comprising:

[0088] Step S1, in the case of receiving a search keyword of a user, determining a plurality of related content information related to the search keyword;

[0089] Step S2, determining a first recommendation list of the plurality of related content information according to the relevance of the plurality of related content information and the search keyword, the historical browsing volume and the historical interaction quantity of the plurality of related content information, wherein the historical interaction quantity comprises the number of historical evaluation information of the content information;

[0090] Step S3, determining a first preset number of to-be-tested content information in the first recommendation list, and obtaining historical interaction information of the to-be-tested content information, wherein the historical interaction information comprises the content of the historical evaluation information of the to-be-tested content information;

[0091] Step S4, determining a key negative factor according to the historical interaction information;

[0092] Step S5, determining a second recommendation list of the plurality of related content information according to the historical browsing record of the user, the introduction information of the to-be-tested content information and the key negative factor;

[0093] Step S6, displaying the plurality of related content information according to the second recommendation list.

[0094] According to the content recommendation method based on traffic monitoring according to the embodiment of the present application, after obtaining the related content information related to the search keyword, the quality of the related content information can be indirectly reflected based on the historical browsing volume and the historical interaction quantity of the related content information, so as to recommend the related content information combined with the quality of the related content information, and the key negative factor of the to-be-tested content information can be found, the resistance degree of the user to the key negative factor can be judged, and the preferences and dislikes of the user to various factors of the content can be reflected, so as to obtain the second recommendation list, so that the content information in the second recommendation list is not only related to the search keyword, but also the recommendation order considers the quality of the content information itself and the preferences of the user, and the user experience is improved.

[0095] According to an embodiment of the present application, in step S1, after the user inputs the search keyword, the server can find the plurality of related content information related to the search keyword, for example, the user opens a video website, inputs the search keyword in the search engine of the website, and the server can push the plurality of videos related to the search keyword to the user.

[0096] Figure 2 An exemplary flowchart of searching related content information according to an embodiment of the present application is shown.

[0097] According to an embodiment of the present application, in step S1, after receiving the search keyword of the user, the plurality of related content information related to the search keyword is determined, comprising:

[0098] In step S11, the title, introduction information and attribute tag of each content information are obtained, wherein the attribute tag includes tag information added by an author of the content information to describe attributes of the content information.

[0099] In step S12, the title semantic information of the title, the introduction semantic information of the introduction information and the tag semantic information of the attribute tag are obtained through a semantic recognition model.

[0100] In step S13, the keyword semantic information of the search keyword is obtained through the semantic recognition model.

[0101] In step S14, for each content information, the first similarity between the keyword semantic information and the title semantic information, the second similarity between the keyword semantic information and the introduction semantic information, and the third similarity between the keyword semantic information and the tag semantic information are determined.

[0102] In step S15, the maximum value of the first similarity, the second similarity and the third similarity is obtained.

[0103] In step S16, if the maximum value of the content information is greater than or equal to a preset first similarity threshold, the content information is determined as the related content information.

[0104] According to an embodiment of the present application, in step S11, each content information can include a title, introduction information and an attribute tag, the introduction information is brief information of the content information, and the attribute tag is tag information added by an author to describe attributes of the content information. In an example, the content information of a video website is a video, each video can have a title and brief information of the video, and the author of the video can also add attribute tags such as "game video", "animation video" and the like to the video.

[0105] According to an embodiment of the present application, in step S12, the semantic recognition model can be a recurrent neural network model and the like, and the present application does not limit the specific type of the semantic recognition model. The title semantic information is obtained by processing the title of the content information through the semantic recognition model, the introduction semantic information is obtained by processing the introduction information of the content information through the semantic recognition model, and the tag semantic information is obtained by processing the attribute tag of the content information through the semantic recognition model. The title semantic information, the introduction semantic information and the tag semantic information are all information in the form of vectors, which are respectively used to express the meanings of the title, the introduction information and the attribute tag.

[0106] According to an embodiment of the present application, in step S13, the keyword semantic information can be obtained by processing the search keyword through the semantic recognition model, and the keyword semantic information is information in the form of a vector, which is used to express the meaning of the keyword.

[0107] According to an embodiment of the present application, in step S14, for each content information, a first similarity between keyword semantic information and title semantic information, a second similarity between keyword semantic information and introduction semantic information, and a third similarity between keyword semantic information and label semantic information of each label information can be determined, for example, a cosine similarity between keyword semantic information and title semantic information can be solved as the first similarity, a cosine similarity between keyword semantic information and introduction semantic information can be solved as the second similarity, and a cosine similarity between keyword semantic information and label semantic information of each label information can be solved as the third similarity.

[0108] According to an embodiment of the present application, in step S15, a maximum value of the first similarity, the second similarity and the third similarity can be obtained as an overall similarity between content involved in the content information and the search keyword, for example, the third similarity between the label information of a certain video and the search keyword is low, and the first similarity between the title and the search keyword is low, but the second similarity between the introduction information of the video and the search keyword is high, it can be considered that the video involves the content that the user wants to search in the introduction content, that is, the video involves the content that the user wants to search, therefore, the maximum value of the first similarity, the second similarity and the third similarity can be used as the overall similarity between the content involved in the video and the search keyword.

[0109] According to an embodiment of the present application, in step S16, if the maximum value of the similarity corresponding to the content information (i.e., the overall similarity between the content involved in the content information and the search keyword) is greater than or equal to a preset first similarity threshold, the content information is determined as relevant content information, that is, it is judged that the content information is related to the search keyword.

[0110] According to an embodiment of the present application, in step S2, based on the relevance of the plurality of relevant content information to the search keyword, the historical browsing amount and the historical interaction number of the plurality of relevant content information, the plurality of relevant content information can be comprehensively sorted to obtain a first recommendation list.

[0111] Figure 3 An exemplary flowchart for determining a first recommendation list according to an embodiment of the present application is shown.

[0112] According to an embodiment of the present application, in step S2, the first recommendation list of the plurality of relevant content information is determined according to the relevance of the plurality of relevant content information to the search keyword, the historical browsing amount and the historical interaction number of the plurality of relevant content information, including:

[0113] In step S21, the maximum value of the similarity corresponding to the relevant content information is determined as the relevance of the relevant content information to the search keyword.

[0114] In step S22, an interaction attraction coefficient of the related content information is determined according to a ratio of the historical interaction quantity and the historical browsing quantity of the related content information;

[0115] In step S23, a first recommendation coefficient of the related content information is determined according to the interaction attraction coefficient, the historical interaction quantity, the historical browsing quantity and the relevance;

[0116] In step S24, a first recommendation list of the related content information is determined according to the first recommendation coefficient.

[0117] According to one embodiment of the present application, in step S21, as described above, the maximum value of the first similarity, the second similarity and the third similarity can be used as the overall similarity between the content involved in the content information and the search keyword, and thus can be used as the relevance between the related content information and the search keyword.

[0118] According to one embodiment of the present application, in step S22, the ratio of the historical interaction quantity and the historical browsing quantity can be used to describe the situation that the user is interested in the related content information after watching the related content information, for example, the user is interested in a certain detail of the related content information after watching, and then leaves a message to discuss, which increases the number of historical evaluation information of the related content information. The higher the ratio of the historical interaction quantity and the historical browsing quantity, the higher the possibility that the related content information contains content that attracts the user, and thus the ratio can be determined as the interaction attraction coefficient of the related content information.

[0119] According to one embodiment of the present application, in step S23, a first recommendation coefficient of each related content information can be determined according to the interaction attraction coefficient, the historical interaction quantity, the historical browsing quantity and the relevance, that is, in addition to considering the relevance between the related content information and the search keyword, the first recommendation coefficient is determined by considering the interaction attraction coefficient, the historical interaction quantity, the historical browsing quantity and other parameters which can reflect the quality of the related content information itself, and the first recommendation coefficient is the basis for ranking the related content information, and the related content information with a higher first recommendation coefficient is arranged in a higher position in the first recommendation list, therefore, the recommendation order in the first recommendation list is not only related to the relevance between the related content information and the search keyword, but also related to the quality of the related content information itself. In an example, an average value of the historical browsing quantity of the plurality of related content information can be solved, and a first ratio of the historical browsing quantity of one related content information to the average value of the historical browsing quantity can be solved, as a ratio of the attraction of the related content information to browsing to the average attraction of the plurality of related content information to browsing, that is, the relative attraction of the related content information in attracting users to watch, which can indirectly reflect the quality of the related content information. In an example, an average value of the historical interaction quantity of the plurality of related content information can be solved, and a second ratio of the historical interaction quantity of one related content information to the average value of the historical interaction quantity can be solved, as a ratio of the attraction of the related content information to comments to the average attraction of the plurality of related content information to comments, that is, the relative attraction of the related content information in attracting users to comment, which can indirectly reflect the quality of the related content information. The interaction attraction coefficient, the first ratio, the second ratio and the relevance can be weighted and summed to obtain the first recommendation coefficient of the related content information.

[0120] According to one embodiment of the present application, in step S24, after obtaining the first recommendation coefficient of each related content information, the first recommendation list can be obtained by ranking according to the first recommendation coefficient, so that the ranking of the related content in the first recommendation list is not only related to the relevance between the related content information and the search keyword, but also related to the quality of the related content information itself, thereby improving the quality of the recommended content.

[0121] According to one embodiment of the present application, in step S3, the first preset number of related content information with the highest first recommendation score in the first recommendation list can be obtained as the to-be-tested content information, and the first preset number can be 10% of the total number of related content information in the first recommendation list, and the present application does not limit this. The first recommendation score of the to-be-tested content information is relatively high, that is, the to-be-tested content information has high relevance to the search keyword and has good quality itself, but the to-be-tested content information does not consider the user preference and aversion factors. When the to-be-tested content information is preferentially recommended, if the to-be-tested content information has user-averse or resistant factors, the user may prefer to watch the averse content, thereby causing the use experience to decrease. Therefore, the user preference and aversion factors can be analyzed, and the order of the first recommendation list can be adjusted based on this, and the content related to the search keyword, having good quality and being preferred by the user can be preferentially recommended.

[0122] According to one embodiment of the present application, in step S4, whether the to-be-tested content information with a high order in the first recommendation list has a key negative factor can be found, that is, the factors that can be averse to the user in the to-be-tested content information are found. If the user-averse factors exist, the order of the first recommendation list is changed, so that the related content information preferred by the user is preferentially recommended.

[0123] Figure 4 An exemplary flowchart for determining the key negative factor according to an embodiment of the present application is shown.

[0124] According to one embodiment of the present application, in step S4, the key negative factor is determined according to the historical interaction information, and the determining includes:

[0125] In step S41, the sentiment classification information of the content of each historical evaluation information of the to-be-tested content information is determined through a sentiment classification model.

[0126] In step S42, the negative evaluation information is screened from the plurality of historical evaluation information according to the sentiment classification information.

[0127] In step S43, the negative semantic information of the negative evaluation information is determined through a semantic recognition model.

[0128] In step S44, the negative semantic information is clustered to obtain a plurality of negative semantic information clustering clusters.

[0129] In step S45, the clustering center of the plurality of negative semantic information clustering clusters is determined as the key negative factor.

[0130] According to an embodiment of the present application, in step S41, the sentiment analysis model can be a recurrent neural network model, which can be used to analyze the sentiment information expressed by the text, so as to determine whether the text belongs to praise, derogatory or no praise or derogatory meaning. The sentiment analysis model can be used to classify the sentiment information of the content of each historical evaluation information of the to-be-tested content information, for example, to classify whether each historical evaluation information is praise, derogatory or no praise or derogatory meaning.

[0131] According to an embodiment of the present application, in step S42, negative evaluation information with derogatory meaning can be screened out from the plurality of historical evaluation information of the plurality of to-be-tested content information.

[0132] According to an embodiment of the present application, in step S43, the negative evaluation information with derogatory meaning is processed by the semantic recognition model to obtain negative semantic information, which can be used to determine which negative factors the negative evaluation information has, in other words, which factors affect the historical user to make negative evaluation on the to-be-tested content information.

[0133] According to an embodiment of the present application, in step S44, the negative semantic information can be clustered to obtain a plurality of negative semantic information clusters, and the clustering process can include K-means clustering process. The specific way of clustering process is not limited by the present application, and the number of cluster centers (i.e. the number of negative semantic information clusters) can be set artificially, for example, it can be set to 3, then 3 negative semantic information clusters can be obtained, and 3 key negative factors can also be obtained in subsequent processing, that is, 3 key reasons for the to-be-tested content information to obtain derogatory evaluation. Of course, the number of cluster centers can also be determined by using methods such as silhouette coefficient, and other parameters of the clustering process can be set to default values.

[0134] According to an embodiment of the present application, in step S45, the cluster centers of the plurality of negative semantic information clusters can be determined as key negative factors. The key negative factors are information in the form of vectors, and the dimension of the vector is the same as the vector dimension of the negative semantic information.

[0135] According to an embodiment of the present application, in step S5, based on the historical browsing records of the user, the introduction information of the to-be-tested content information and the key negative factors, it can be determined comprehensively which factors the user is more disgusted with, so as to adjust the order of the first recommendation list to preferentially recommend the related content information that the user is interested in.

[0136] Figure 5 An exemplary flowchart for determining the second recommendation list according to an embodiment of the present application is shown.

[0137] According to one embodiment of the present application, in step S5, a second recommendation list of the plurality of related content information is determined according to the historical browsing record of the user, introduction information of the to-be-tested content information, and the key negative factor, and includes:

[0138] In step S51, a played progress of the historical content information browsed by the user, a user interaction record of the historical content information, and an interaction information record of the historical content information are determined according to the historical browsing record of the user.

[0139] In step S52, a resistance coefficient of the user to the key negative factor is determined according to the played progress, the user interaction record, the interaction information record, and the key negative factor.

[0140] In step S53, a propaganda negative coefficient of the to-be-tested content information is determined according to the introduction information of the to-be-tested content information and the key negative factor.

[0141] In step S54, a historical interaction information of the plurality of related content information in the first recommendation list is obtained.

[0142] In step S55, a supplementary recommendation coefficient of each related content information is determined according to the historical interaction information of the plurality of related content information and the key negative factor.

[0143] In step S56, the second recommendation list of the plurality of related content information is determined according to the resistance coefficient, the propaganda negative coefficient, and the supplementary recommendation coefficient.

[0144] According to one embodiment of the present application, in step S51, the played progress and the user interaction record of the historical content information browsed by the user in the past can be obtained from the historical browsing record of the user, so as to help determine whether the user resists the key negative factor.

[0145] According to one embodiment of the present application, in step S52, the resistance coefficient can be used to describe the resistance degree of the user to the key negative factor. For example, by analyzing the historical browsing record of the user, it can be determined that the user makes a derogatory evaluation on the historical content information with the same or similar negative factor, or the played progress of the historical content information with the same or similar negative factor is less, which indicates that the user is not interested in the negative factor, or even resists it.

[0146] According to one embodiment of the present application, in step S52, the resistance coefficient of the user to the key negative factor is determined according to the played progress, the user interaction record, the interaction information record, and the key negative factor, and includes: according to the interaction information record of the historical content information, screening target historical content information with the i-th key negative factor from the plurality of historical content information; and determining the resistance coefficient of the user to the i-th key negative factor according to formula (1) ,

[0147] (1),

[0148] wherein, is the played progress of the jth target historical content information existing the ith key negative factor, is the average played progress of the historical content information, is the semantic similarity between the historical evaluation information of the jth target historical content information existing the ith key negative factor and the semantic of the ith key negative factor, if the user has not made an evaluation on the jth target historical content information existing the ith key negative factor, then , is the number of the target historical content information existing the ith key negative factor, if() is a conditional function, j≤ , and i, j and are positive integers.

[0149] According to one embodiment of the present application, the interactive information records (including not only the historical evaluation information of the user, but also the evaluation information of other users) of the historical content information in the historical browsing records of the user can be processed by the semantic recognition model to obtain the semantic vector of each interactive information record, and the semantic similarity (for example, cosine similarity) between the semantic vector and the key negative factor is determined. If there is an interactive information record with a semantic vector and a semantic similarity between the key negative factor higher than or equal to a threshold value (for example, 0.8) in multiple interactive information records of a historical content information, it is determined that the historical content information exists a key negative factor. Whether each historical content information exists the ith key negative factor can be determined by the above-mentioned manner.

[0150] According to one embodiment of the present application, in formula (1), is a conditional function, and the value of the conditional function is in the case of , otherwise, the value of the conditional function is 0. If the played progress of the jth target historical content information existing the ith key negative factor exceeds the average played progress, it indicates that the user's playing of the historical content information is not affected by the key negative factor, that is, the influence of the ith key negative factor on playing the jth target historical content information is 0, and the intensity of the user's resistance to the ith negative factor is also 0. The value of the conditional function can be determined as 0. If the played progress of the jth target historical content information existing the ith key negative factor is lower than the average played progress, it indicates that the ith key negative factor has an influence on playing the jth target historical content information, that is, the user has an aversion or resistance to the ith key negative factor, resulting in a lower played progress of the target historical content information, The relative difference is greater when the played progress of the target historical content information is lower, indicating that the i-th key negative factor has a greater impact on playing the j-th target historical content information. The calculation method of the amplification coefficient is similar to the semantic similarity between the semantic vector of the interactive information record and the semantic of the key negative factor, which will not be described herein again, and The amplification coefficient of the impact of the i-th key negative factor on playing the j-th target historical content information, that is, The higher the amplification coefficient is, the more likely the user is to make an evaluation similar to the i-th key negative factor on the target historical content information, in other words, the more likely the user is to have a lower played progress due to the i-th key negative factor, and the stronger the user's resistance to the i-th key negative factor, therefore, The amplification coefficient can be used as an indicator of the strength of the user's resistance to the i-th key negative factor. The average value of the strength of the user's resistance to the i-th key negative factor determined based on a plurality of target historical content information can be used as a resistance coefficient of the user to the i-th key negative factor, and the higher the resistance coefficient is, the stronger the user's resistance to the i-th key negative factor, and the number of recommendations of content containing the i-th key negative factor should also be reduced when recommending content to the user.

[0151] In this way, the impact of the key negative factor on playing can be determined through the played progress of the historical content information, so as to determine whether the user has resistance to the key negative factor, and the strength of the user's resistance to the key negative factor can also be amplified through the user's interactive record, so as to accurately reflect the user's resistance to the key negative factor, and provide accurate data basis for recommending content that the user is interested in.

[0152] According to an embodiment of the present application, in step S53, the to-be-tested content information can have introduction information, which is usually a positive introduction to the to-be-tested content information, and it can be determined whether the introduction information is in conflict with the key negative factor, if in conflict, the introduction information of the to-be-tested content information is not true, and the probability of the to-be-tested content information being recommended should be reduced. The propaganda negative coefficient can be used to describe the possibility of the introduction information being not true.

[0153] According to one of the embodiments of the present application, in step S53, the promotion negative coefficient of the to-be-tested content information is determined according to the introduction information of the to-be-tested content information and the key negative factors, and the method comprises: screening the target to-be-selected content information with the i-th key negative factor according to the historical interaction information of the to-be-tested content information; performing word segmentation and stop word removal processing on the introduction information of the target to-be-tested content information to obtain a plurality of introduction words of the introduction information; determining the introduction sentiment classification information of each introduction word through a sentiment classification model; determining the positive introduction words in the introduction words according to the introduction sentiment classification information; adding negative affixes to the positive introduction words to obtain negative introduction words; obtaining the negative introduction semantic information of the negative introduction words through a semantic recognition model; and determining the promotion negative coefficient of the k-th target to-be-tested content information with respect to the i-th key negative factor according to formula (2) ,

[0154] (2),

[0155] wherein, is the negative introduction semantic information of the s-th negative introduction word of the k-th target to-be-tested content information, is the i-th key negative factor, is and is the semantic similarity, N is the number of key negative factors, M is the number of target to-be-tested content information, max(*) is the maximum value function, s≤M, i≤N, and s, M, i and N are all positive integers.

[0156] According to one of the embodiments of the present application, the method for screening the target to-be-selected content information is similar to the above method for screening the target historical content information, and will not be described here. The introduction information of the to-be-tested content information is usually a paragraph of text including a large number of words, and the introduction information can be preprocessed by word segmentation and stop word removal to obtain a plurality of introduction words. The introduction sentiment classification information of each introduction word can be determined through a sentiment classification model, and the words with positive meanings, i.e., the positive introduction words, can be screened from the introduction sentiment classification information. Negative affixes are added to the positive introduction words to obtain negative introduction words, i.e., the words with positive meanings are modified to words with negative meanings, for example, “clear” is modified to “unclear”, and the negative introduction words can be processed through a semantic recognition model to obtain the negative introduction semantic information.

[0157] According to one of the embodiments of the present application, in formula (2), may be and The cosine similarity of the two is higher, the semantic similarity of the negative introduction word and the i th key negative factor is higher, which indicates that the semantic of the negative introduction word is closer to the i th key negative factor, and the possibility of false propaganda of the positive introduction word corresponding to the negative introduction word is greater. The maximum value of the semantic similarity of the plurality of negative introduction words and the i th key negative factor is greater, the possibility of existence of false propaganda in the introduction information is greater. For example, there is a word of “clear image quality” in the introduction information, and the i th key negative factor is used to represent unclear image quality, so there is false propaganda in the introduction information.

[0158] In this way, the negative introduction word can be obtained by adding the negative affix to the positive introduction word in the introduction information, and the semantic similarity between the negative introduction word and the key negative factor is solved, so as to determine whether the to-be-tested content information exists false propaganda, thereby reducing the recommendation probability of the content existing false propaganda and improving user experience.

[0159] According to an embodiment of the present application, in step S54, the historical interaction information of the plurality of related content information can be obtained, and in step S55, the supplementary recommendation coefficient of each related content information is determined, the supplementary recommendation coefficient can improve the probability of the related content information without the key negative factor being recommended, thereby reducing the probability of the related content information with the key negative factor being preferentially recommended, and improving user experience.

[0160] According to an embodiment of the present application, in step S55, the supplementary recommendation coefficient of each related content information is determined according to the historical interaction information of the plurality of related content information and the key negative factor, comprising: determining the interaction emotion classification information of the historical interaction information through the emotion classification model; screening out the positive historical interaction information in the historical interaction information according to the interaction emotion classification information; adding the negative affix to the positive historical interaction information to obtain the modified historical interaction information; obtaining the modified interaction semantic information of the modified historical interaction information through the semantic recognition model; determining the supplementary recommendation coefficient of the t th related content information for the i th key negative factor according to formula (3) ,

[0161] (3),

[0162] wherein, the modified interaction semantic information of the x th historical interaction information of the t th related content information, the i th key negative factor, the semantic similarity of and the number of historical interaction information of the t th related content information, x≤ ​, and x and are positive integers.

[0163] According to one embodiment of the present application, the interaction sentiment classification information of the historical interaction information can be determined by the sentiment classification model, that is, it is determined whether each historical interaction information is positive or negative. Then, the positive historical interaction information can be screened based on the interaction sentiment classification information. The positive historical interaction information can be added with a negative affix (if the positive historical interaction information is a positive word, the positive word is added with a negative affix, and if the positive historical interaction information is a sentence including multiple positive words, each positive word in the positive historical interaction information can also be added with a negative affix), to obtain negative modified historical interaction information, and the modified interaction semantic information of the modified historical interaction information is obtained by the semantic recognition model.

[0164] According to one embodiment of the present application, in formula (3), may be and the cosine similarity of the two, the higher the cosine similarity, the higher the semantic similarity between the modified historical interaction information and the i-th key negative factor, indicating that the semantic of the modified historical interaction information is closer to the i-th key negative factor, and the positive historical interaction information corresponding to the modified historical interaction information can better indicate that the related content information does not have the problem corresponding to the i-th key negative factor. is the maximum value of the semantic similarity between the modified historical interaction information and the i-th key negative factor, the larger the maximum value, the higher the possibility that the related content information does not have the problem corresponding to the i-th key negative factor. For example, there is a comment of “clear image quality” in the historical interaction information, and the i-th key negative factor is used to represent unclear image quality. Since the historical interaction information is the objective comment of the related content information by multiple historical users, the possibility that the related content information does not have the problem of unclear image quality is high. The probability that the related content information without the key negative factor is recommended can be improved by the supplementary recommendation coefficient.

[0165] In this way, the positive historical interaction information in the objective comment of the historical user can be added with a negative affix, and the semantic similarity between the modified historical interaction information and the key negative factor is determined, and then the possibility that the related content information does not have the key negative factor is determined, and the supplementary recommendation coefficient is obtained, which can be used to improve the probability that the related content information without the key negative factor is recommended, to improve the user experience.

[0166] According to one embodiment of the present application, in step S56, the first recommendation coefficient of each related content information can be adjusted based on the above determined conflict coefficient, propaganda negative coefficient and supplementary recommendation coefficient, so as to preferentially recommend the content with good quality and interesting to users.

[0167] According to one embodiment of the present application, in step S56, the second recommendation list of the plurality of related content information is determined according to the conflict coefficient, the propaganda negative coefficient and the supplementary recommendation coefficient, comprising: if the tth related content information is not the target to-be-tested content information, the gain recommendation coefficient of the tth related content information is determined according to formula (4) ,

[0168] (4),

[0169] wherein, is the number of positive historical interaction information of the tth related content information, is the number of historical interaction information of the tth related content information, is the average positive interaction rate of the plurality of related content information, is the supplementary recommendation coefficient of the tth related content information for the ith key negative factor, is the conflict coefficient of the user to the ith key negative factor, N is the number of key negative factors, i≤N, and i and N are positive integers; if the tth related content information is the target to-be-tested content information, the gain recommendation coefficient of the tth related content information is determined according to formula (5) ,

[0170] (5),

[0171] wherein, is the propaganda negative coefficient of the tth related content information for the ith key negative factor; the second recommendation list is determined according to the gain recommendation coefficient of each related content information.

[0172] According to one embodiment of the present application, in formula (4), is the gain coefficient based on the supplementary recommendation coefficient, as described above, the higher the value is, the greater the possibility that the key negative factor does not exist in the related content information is, and the probability of being recommended can be improved, therefore, the coefficient greater than 1 can be used to increase the probability of the related content information without the key negative factor being recommended. is the decay coefficient based on the conflict coefficient of the user to the ith key negative factor, the higher the value is, the stronger the conflict psychology of the user to the ith key negative factor is, and the probability of being preferentially recommended should be reduced, therefore, the coefficient less than 1 can be used to reduce the probability of being preferentially recommended. is the adjustment coefficient of the tth related content information for the ith key negative factor, The average value of the adjustment coefficients of the tth relevant content information for various key negative factors can be used to adjust the first recommendation score of the tth relevant content information based on whether various key negative factors exist in the relevant content information and the user's resistance to various key negative factors. The positive interaction rate of the tth relevant content information, i.e., the proportion of historical users who make positive evaluations on the tth relevant content information, The average value of the positive interaction rates of various relevant content information, The relative positive interaction rate of the tth relevant content information relative to the whole, if the value is high (for example, greater than 1), it indicates that the probability of the relevant content information being positively evaluated by the user is high, and the quality of the relevant content information is relatively good. The relative positive interaction rate can be multiplied by and to obtain the gain recommendation coefficient of the tth relevant content information, so that a comprehensive adjustment coefficient (i.e., the gain recommendation coefficient) can be obtained based on the quality of the relevant content, whether it contains key negative factors, and the user's resistance to negative key factors, improving the probability that the relevant content information with good quality and without key negative factors is preferentially recommended, and improving the user experience.

[0173] According to one embodiment of the present application, in formula (5), The decay coefficient based on the propaganda denial coefficient, in the case that the tth relevant content information is the target to-be-tested content information, The greater the value is, the higher the possibility that the tth relevant content information has false propaganda is, and the probability of being preferentially recommended should be reduced, therefore, the probability of the tth relevant content information being preferentially recommended is reduced by the coefficient less than 1. In the case that the tth relevant content information is the target to-be-tested content information The adjustment coefficient of the tth relevant content information for the i-th key negative factor. The quality of the relevant content information, whether there is false propaganda, whether it contains key negative factors, and the user's resistance to negative key factors can be comprehensively considered to obtain the gain recommendation coefficient .

[0174] According to one embodiment of the present application, the first recommendation score of each relevant content information is multiplied by the respective gain recommendation coefficient, and the second recommendation coefficient of each relevant content information can be obtained, the second recommendation list is obtained by sorting according to the second recommendation coefficient, and the relevant content information with good quality, without false propaganda, and without negative key factors can be preferentially recommended to improve the user experience.

[0175] In this way, the target to-be-tested content information and other related content can be classified and calculated, so that the recommendation coefficient of the target to-be-tested content information with false propaganda is further reduced. The recommendation coefficient of each related content information can be adjusted comprehensively according to the quality of the related content information, whether there is false propaganda, whether the related content information contains a key negative factor, and the resistance psychology of the user to the key negative factor, so that the related content information with good quality, without false propaganda and without negative key factors can be preferentially recommended, thereby improving the user experience.

[0176] According to one embodiment of the present application, in step S6, the user can be sequentially presented with the multiple related content information according to the order in the second recommendation list, and the related content information with good quality, without false propaganda and without negative key factors can be preferentially recommended, thereby improving the user experience.

[0177] According to the content recommendation method based on traffic monitoring according to the embodiment of the present application, after obtaining the relevant content information related to the search keyword, the quality of the relevant content information can be indirectly reflected based on the historical browsing volume and the historical interaction quantity of the relevant content information, so as to recommend the relevant content information in combination with the quality of the relevant content information, find the key negative factors of the to-be-tested content information, judge the resistance degree of the user to the key negative factors, reflect the likes and dislikes of the user to various factors of the content, obtain the second recommendation list, so that the content information in the second recommendation list is not only related to the search keyword, but also the recommendation order takes into account the quality of the content information itself and the likes of the user, and the user experience is improved. When determining the resistance coefficient, the influence of the key negative factors on the playing can be determined through the played progress of the historical content information, so as to judge whether the user has the resistance psychology to the key negative factors, and the intensity of the resistance psychology of the user to the key negative factors can be amplified through the interaction record of the user, so that the resistance psychology of the user to the key negative factors can be accurately reflected, and accurate data basis is provided for recommending the content interested by the user. When determining the propaganda negative coefficient, the negative introduction words can be obtained by adding negative affixes to the positive introduction words in the introduction information, and the semantic similarity between the negative introduction words and the key negative factors is solved, so as to judge whether the to-be-tested content information has false propaganda, so as to reduce the recommendation probability of the content with false propaganda and improve the user experience. When determining the supplementary recommendation coefficient, the negative affixes can be added to the positive historical interaction information in the objective comments of the historical user, and the semantic similarity between the modified historical interaction information and the key negative factors is determined, and then the possibility that the key negative factors do not exist in the relevant content information is determined, and the supplementary recommendation coefficient is obtained, which can be used to improve the probability that the relevant content information without the key negative factors is recommended, so as to improve the user experience. When determining the gain recommendation coefficient, the target to-be-tested content information and other relevant content can be classified and calculated, so that the recommendation coefficient of the target to-be-tested content information with false propaganda is further reduced. And the recommendation coefficient of each relevant content information can be adjusted in combination with the quality of the relevant content information, whether there is false propaganda, whether the key negative factors are included, and the resistance psychology of the user to the key negative factors and other factors, so that the relevant content information with good quality, without false propaganda and without negative key factors can be preferentially recommended, so as to improve the user experience.

[0178] Figure 6 An exemplary block diagram of a content recommendation system based on traffic monitoring according to an embodiment of the present application is shown, which comprises:

[0179] A relevant content information module is configured to determine a plurality of relevant content information related to the search keyword of the user when the search keyword of the user is received.

[0180] The first recommendation list module is configured to determine a first recommendation list of the plurality of related content information according to the relevance of the plurality of related content information to the search keyword, the historical browsing volume and the historical interaction quantity of the plurality of related content information, wherein the historical interaction quantity comprises the quantity of historical evaluation information of the content information.

[0181] The historical interaction information module is configured to determine a first preset quantity of to-be-tested content information in the first recommendation list, and obtain historical interaction information of the to-be-tested content information, wherein the historical interaction information comprises the content of the historical evaluation information of the to-be-tested content information.

[0182] The key negative factor module is configured to determine a key negative factor according to the historical interaction information.

[0183] The second recommendation list module is configured to determine a second recommendation list of the plurality of related content information according to the historical browsing record of the user, the introduction information of the to-be-tested content information and the key negative factor.

[0184] The display module is configured to display the plurality of related content information according to the second recommendation list.

[0185] The present application can be a method, device, system and / or computer program product. The computer program product can include a computer readable storage medium on which is loaded computer readable program instructions for executing various aspects of the present application.

[0186] Those skilled in the art should understand that the above-described embodiments of the present application shown in the description and drawings are only used as examples and do not limit the present application. The purpose of the present application has been fully and effectively achieved. The function and structural principle of the present application has been shown and explained in the embodiments, and the implementation of the present application can be modified or changed without departing from the principle.

[0187] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit it; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A content recommendation method based on traffic monitoring, characterized in that: include: Upon receiving a user's search keyword, determining a plurality of relevant content information related to the search keyword; Determining a first recommendation list of the plurality of related content information based on the relevance of the plurality of related content information to the search keyword, the historical page views and the historical interaction counts of the plurality of related content information, wherein the historical interaction counts include the number of historical evaluation information of the content information; In the first recommendation list, a first preset number of content information to be tested is determined, and historical interaction information of the content information to be tested is obtained, wherein the historical interaction information includes content of historical evaluation information of the content information to be tested; Determining key negative factors based on the historical interaction information; Determining a second recommendation list of multiple related content information based on the user's historical browsing history, the introduction information of the content information to be tested, and the key negative factors; Displaying a plurality of related content information according to the second recommendation list; Based on the historical interaction information, key negative factors are determined, including: Determine the sentiment classification information of each historical evaluation information of the content information to be tested through the sentiment classification model; Filtering negative evaluation information from multiple historical evaluation information based on the sentiment classification information; Determine the negative semantic information of negative evaluation information through semantic recognition model; performing clustering processing on the negative semantic information to obtain a plurality of negative semantic information clusters; The cluster centers of multiple negative semantic information clusters are identified as key negative factors.

2. The content recommendation method based on traffic monitoring according to claim 1, characterized in that: Upon receiving a user's search keyword, determining a plurality of relevant content information related to the search keyword includes: Acquire the title, introduction information and attribute tag of each content information, wherein the attribute tag includes tag information added by the author of the content information to the content information for describing the attributes of the content information; Using a semantic recognition model, obtain the title semantic information of the title, obtain the introduction semantic information of the introduction information, and obtain the label semantic information of the attribute label; Obtain keyword semantic information of search keywords through semantic recognition model; For each piece of content information, determining a first similarity between the keyword semantic information and the title semantic information, a second similarity between the keyword semantic information and the introduction semantic information, and a third similarity between the keyword semantic information and the tag semantic information; Obtaining a maximum similarity among the first similarity, the second similarity, and the third similarity; If the maximum similarity value corresponding to the content information is greater than or equal to a preset first similarity threshold, the content information is determined to be relevant content information.

3. The content recommendation method based on traffic monitoring according to claim 2, characterized in that: Determining a first recommendation list of the plurality of related content information based on the relevance of the plurality of related content information to the search keyword, the historical page views and the historical interaction counts of the plurality of related content information includes: Determining the maximum similarity value corresponding to the relevant content information as the relevance between the relevant content information and the search keyword; Determine the interaction attraction coefficient of the relevant content information based on the ratio of the historical interaction number and the historical pageview number of the relevant content information; Determining a first recommendation coefficient for relevant content information based on the interaction attraction coefficient, the number of historical interactions, the historical page views, and the relevance; A first recommendation list of multiple related content information is determined according to the first recommendation coefficient.

4. The content recommendation method based on traffic monitoring according to claim 1, characterized in that: Determining a second recommendation list of multiple related content information based on the user's historical browsing history, the introduction information of the content information to be tested, and the key negative factors, includes: Determine, based on the user's historical browsing history, the progress of playback of historical content information browsed by the user, the user's interaction history with the historical content information, and the interaction history of the historical content information; determining the user's resistance coefficient to the key negative factor based on the playback progress, the user interaction record, the interaction information record, and the key negative factor; Determining a publicity negation coefficient of the content information to be tested based on the introduction information of the content information to be tested and the key negative factors; Obtain historical interaction information of multiple related content information in the first recommendation list; Determining a supplementary recommendation coefficient for each piece of relevant content information based on historical interaction information of the plurality of relevant content information and the key negative factors; A second recommendation list of multiple related content information is determined based on the resistance coefficient, the publicity negation coefficient and the supplementary recommendation coefficient.

5. The content recommendation method based on traffic monitoring according to claim 4, characterized in that: Determining the user's resistance coefficient to the key negative factor based on the playback progress, the user interaction record, the interaction information record, and the key negative factor, including: According to the interactive information record of the historical content information, select the target historical content information having the i-th key negative factor from the multiple historical content information; According to the formula ， Determine the user's resistance coefficient to the i-th key negative factor ,in, is the playback progress of the jth target historical content information with the i-th key negative factor, is the average playback progress of historical content information, is the semantic similarity between the user's historical evaluation information of the jth target historical content information with the i-th key negative factor and the i-th key negative factor. If the user has not evaluated the jth target historical content information with the i-th key negative factor, then , is the number of target historical content information with the i-th key negative factor, if (*) is a conditional function, j≤ , and i, j and All are positive integers.

6. The content recommendation method based on traffic monitoring according to claim 4, characterized in that: Determine the publicity negation coefficient of the content to be tested based on the introduction information of the content to be tested and the key negative factors, including: Based on the historical interaction information of the content to be tested, the target content to be selected is screened for the presence of the i-th key negative factor; Segmenting the introduction information of the target content information and removing stop words to obtain multiple introduction words of the introduction information; Determine the introductory sentiment classification information of each introductory word through the sentiment classification model; According to the introduction sentiment classification information, positive introduction words are determined among the introduction words; Add negative affixes to positive introductory words to obtain negative introductory words; Obtain negative introduction semantic information of negative introduction words through semantic recognition model; According to the formula ， Determine the negative coefficient of the kth target content information for the ith key negative factor ,in, is the negative introduction semantic information of the sth negative introduction word of the kth target content information to be tested, is the i-th key negative factor, for and where N is the number of key negative factors, max(*) is the maximum value function, i≤N, and s, M, i and N are all positive integers.

7. The content recommendation method based on traffic monitoring according to claim 4, characterized in that: Based on the historical interaction information of the plurality of relevant content information and the key negative factors, a supplementary recommendation coefficient for each relevant content information is determined, including: Determine the interactive sentiment classification information of historical interactive information through the sentiment classification model; According to the interaction sentiment classification information, positive historical interaction information is filtered out from the historical interaction information; Add negative affixes to positive historical interaction information to obtain modified historical interaction information; Obtaining modification interaction semantic information of modification history interaction information through a semantic recognition model; According to the formula ， Determine the supplementary recommendation coefficient of the tth relevant content information for the ith key negative factor ,in, is the modified interaction semantic information of the xth historical interaction information of the tth related content information, is the i-th key negative factor, for and The semantic similarity of is the number of historical interactive information of the t-th related content information, x≤ , and x and All are positive integers.

8. The content recommendation method based on traffic monitoring according to claim 6, characterized in that: According to the resistance coefficient, the publicity negation coefficient, and the supplementary recommendation coefficient, a second recommendation list of multiple related content information is determined, including: If the tth related content information is not the target content information to be tested, then according to the formula ， Determine the gain recommendation coefficient of the tth relevant content information ,in, is the number of positive historical interaction information of the t-th related content information, is the number of historical interactive information of the t-th related content information, is the average positive interaction rate of multiple related content information, is the supplementary recommendation coefficient of the t-th relevant content information for the i-th key negative factor, is the user's resistance coefficient to the i-th key negative factor, N is the number of key negative factors, i≤N, and both i and N are positive integers; If t relevant content information is the target content information to be tested, then according to the formula ， Determine the gain recommendation coefficient of the tth relevant content information ,in, is the publicity negation coefficient of the t-th relevant content information for the i-th key negative factor; A second recommendation list is determined according to the gain recommendation coefficients of the respective related content information.

9. A content recommendation system based on traffic monitoring, used to execute the method according to any one of claims 1 to 8, characterized in that: include: A related content information module is configured to, upon receiving a user's search keyword, determine a plurality of related content information related to the search keyword; A first recommendation list module is configured to determine a first recommendation list of multiple related content information based on the relevance of the multiple related content information to the search keyword, the historical page views and the historical interaction counts of the multiple related content information, wherein the historical interaction counts include the number of historical evaluation information of the content information; A historical interaction information module is configured to determine a first preset number of pieces of content information to be tested in the first recommendation list, and obtain historical interaction information of the content information to be tested, wherein the historical interaction information includes historical evaluation information of the content information to be tested; a key negative factor module, configured to determine key negative factors based on the historical interaction information; A second recommendation list module is used to determine a second recommendation list of multiple related content information based on the user's historical browsing history, the introduction information of the content information to be tested, and the key negative factors; A display module is used to display multiple related content information according to the second recommendation list.

Images