Intelligent recommendation method and device fusing user feedback

Abstract

The invention discloses an intelligent recommendation method and device fusing user feedback, and relates to the technical field of electric digital data processing. The method comprises the following steps: acquiring feedback; obtaining a comprehensive recommendation evaluation index; and obtaining a test result and taking recommended measures. The method comprises the following steps: recommending a short video to a to-be-tested user by using a primary recommendation algorithm, obtaining a corresponding to-be-tested tag, collecting feedback data of the to-be-tested user on the short video of the to-be-tested tag in secondary recommendation, and then obtaining a dominant recommendation evaluation index and a recessive recommendation evaluation index based on the feedback data; and obtaining a comprehensive recommendation evaluation index based on the dominant recommendation evaluation index and the implicit recommendation evaluation index, and comparing the comprehensive recommendation evaluation index with a recommendation evaluation threshold to obtain a test result so as to take recommendation measures, thereby realizing more accurate intelligent recommendation of the user video. The problem of inaccurate user feedback data in user video recommendation in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic digital data processing, and particularly relates to an intelligent recommendation method and device integrating user feedback. Background Art

[0002] In an intelligent recommendation device integrating user feedback, explicit feedback (such as ratings, comments) and implicit feedback (such as clicks, browsing history, purchase records) work together. Through machine learning and data mining technologies, the user profile and item matching degree are optimized. The device uses multi-source data fusion, bias correction, and real-time update algorithms to improve the personalization and accuracy of recommendation results, overcome data sparsity and bias problems, and achieve a comprehensive understanding and accurate prediction of user preferences.

[0003] Existing intelligent recommendation methods and devices mainly include collaborative filtering, content-based recommendation, and hybrid recommendation. Collaborative Filtering (CF) calculates the similarity between users or the similarity between items by analyzing user behaviors (such as ratings, clicks, purchase records), and recommends items liked by other users or similar users. It is divided into user-based collaborative filtering and item-based collaborative filtering. However, this method may face cold start and data sparsity problems; Content-Based Filtering (CBF) recommends based on the user's historical behaviors and item features (such as text, tags, attributes). This method relies on in-depth analysis of item content and clear representation of user interests, and can solve the cold start problem, but is easily restricted by single interest; Hybrid Recommender Systems (HRS) combines multiple recommendation methods (such as collaborative filtering and content recommendation) to play their respective advantages and make up for their respective defects. By integrating explicit and implicit feedback and using technologies such as deep learning and graph neural networks, it can further improve the accuracy and diversity of recommendation and adapt to dynamic user needs.

[0004] However, in practical applications, due to user personal operation problems, for example, accidental touches when swiping short videos affect the types of short videos recommended for users. Therefore, there is a problem of inaccurate user feedback data in user video recommendation in the prior art. Summary of the Invention

[0005] In order to solve the problem of inaccurate user feedback data in user video recommendations in the prior art, a comprehensive recommendation evaluation index obtained through an explicit recommendation evaluation index and an implicit recommendation evaluation index realizes the accurate quantification of the frequency of a user's access to short videos with specific tags and more comprehensively considers user behavior data, thereby optimizing the short video recommendation content of the short video software to be tested, and further realizing the technical problem of more accurate intelligent recommendation of user videos. An embodiment of the present invention provides an intelligent recommendation method and device that integrates user feedback. The technical solution is as follows:

[0006] On the one hand, an intelligent recommendation method that integrates user feedback is provided. This method is implemented by an intelligent recommendation device that integrates user feedback. The method includes:

[0007] Use the main recommendation algorithm to recommend short videos to the user to be tested, and obtain the corresponding tags to be tested; based on the tags to be tested, collect the feedback data of the user to be tested on the short videos with the tags to be tested in the secondary recommendation within a preset time period; the tags to be tested represent the tags corresponding to the short videos for which the user to be tested has the first feedback behavior; the feedback data includes explicit feedback data and implicit feedback data;

[0008] Based on the explicit feedback correction factor in the preset database, evaluate according to the feedback data to obtain an explicit recommendation evaluation index; based on the implicit feedback correction factor in the preset database, evaluate according to the feedback data to obtain an implicit recommendation evaluation index; based on the preset database, obtain a comprehensive recommendation evaluation index according to the explicit recommendation evaluation index and the implicit recommendation evaluation index; the explicit recommendation evaluation index is used to directly reflect the frequency of the user to be tested's access to the short videos with the tags to be tested; the implicit recommendation evaluation index is used to indirectly reflect the frequency of the user to be tested's access to the short videos with the tags to be tested; the comprehensive recommendation evaluation index is used to comprehensively quantify the frequency of the user to be tested's access to the short videos with the tags to be tested;

[0009] Compare the comprehensive recommendation evaluation index with the recommendation evaluation threshold obtained from the preset database to obtain a test result, and take a recommendation measure according to the test result.

[0010] On the other hand, an intelligent recommendation device that integrates user feedback is provided. This device is applied to the intelligent recommendation method that integrates user feedback. The device includes:

[0011] A feedback data acquisition module, which is used to use the main recommendation algorithm to recommend short videos to the user to be tested, and obtain the corresponding tags to be tested; based on the tags to be tested, collect the feedback data of the user to be tested on the short videos with the tags to be tested in the secondary recommendation within a preset time period; the tags to be tested represent the tags corresponding to the short videos for which the user to be tested has the first feedback behavior; the feedback data includes explicit feedback data and implicit feedback data;

[0012] A feedback recommendation evaluation module, configured to evaluate based on the explicit feedback correction factor in a preset database according to the feedback data to obtain an explicit recommendation evaluation index; evaluate based on the implicit feedback correction factor in the preset database according to the feedback data to obtain an implicit recommendation evaluation index; based on the preset database, obtain a comprehensive recommendation evaluation index according to the explicit recommendation evaluation index and the implicit recommendation evaluation index; the explicit recommendation evaluation index is used to directly reflect the access frequency of the user to be tested to the short video of the label to be tested; the implicit recommendation evaluation index is used to indirectly reflect the access frequency of the user to be tested to the short video of the label to be tested; the comprehensive recommendation evaluation index is used to comprehensively quantify the access frequency of the user to be tested to the short video of the label to be tested;

[0013] A recommendation measure adjustment module, configured to compare the comprehensive recommendation evaluation index with a recommendation evaluation threshold obtained from a preset database to obtain a test result, and take recommendation measures according to the test result.

[0014] On the other hand, provided is an intelligent recommendation device integrating user feedback, where the intelligent recommendation device integrating user feedback includes: a processor; a memory, on which computer-readable instructions are stored, and when the computer-readable instructions are executed by the processor, any one of the methods in the above-mentioned intelligent recommendation method integrating user feedback is implemented.

[0015] On the other hand, provided is a computer-readable storage medium, in which at least one instruction is stored, and the at least one instruction is loaded and executed by a processor to implement any one of the methods in the above-mentioned intelligent recommendation method integrating user feedback.

[0016] The beneficial effects brought by the technical solution provided by the embodiments of the present invention at least include:

[0017] By using the main recommendation algorithm to recommend short videos to the user to be tested, obtaining the corresponding label to be tested, collecting the feedback data of the user to be tested on the short videos of the label to be tested in the secondary recommendation within a preset time period, then obtaining the explicit recommendation evaluation index and the implicit recommendation evaluation index based on the feedback data, and obtaining the comprehensive recommendation evaluation index based on the explicit recommendation evaluation index and the implicit recommendation evaluation index, and finally comparing the comprehensive recommendation evaluation index with the recommendation evaluation threshold obtained from the preset database to obtain a test result, and taking recommendation measures according to the test result, so as to more accurately evaluate the access frequency of the user to the current label short video, and further realize more accurate intelligent recommendation of user videos, effectively solving the problem of inaccurate user feedback data in user video recommendation in the prior art.

[0018] The explicit recommendation evaluation index of the user to be tested is obtained through explicit feedback data and an explicit feedback correction factor, which directly reflects the frequency of the user's access to the short videos with the current tags, thereby improving the accuracy of the explicit feedback evaluation.

[0019] The implicit recommendation evaluation index of the user to be tested is obtained through implicit feedback data and an implicit feedback correction factor, which indirectly reflects the frequency of the user's access to the short videos with the current tags, thereby improving the reliability of the implicit feedback evaluation. Brief Description of the Drawings

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0021] Figure 1 It is a flowchart of an intelligent recommendation method that integrates user feedback provided by an embodiment of the present invention;

[0022] Figure 2 It is a schematic diagram of the change of a comprehensive recommendation evaluation index provided by an embodiment of the present invention;

[0023] Figure 3 It is a block diagram of an intelligent recommendation device that integrates user feedback provided by an embodiment of the present invention;

[0024] Figure 4 It is a schematic diagram of the structure of an intelligent recommendation device that integrates user feedback provided by an embodiment of the present invention. Detailed Embodiments

[0025] The following describes the technical solutions in the present invention with reference to the drawings.

[0026] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as "example" in the present invention should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Exactly speaking, the use of the word "example" is intended to present concepts in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one of the two.

[0027] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, their intended meanings are the same. "Of", "corresponding", and "corresponding to" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, their intended meanings are the same.

[0028] In the embodiments of the present invention, sometimes subscripts such as W 1 may be written in a non-subscript form such as W1. When the difference is not emphasized, their intended meanings are the same.

[0029] To make the technical problems, technical solutions, and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0030] The embodiments of the present invention provide an intelligent recommendation method integrating user feedback. This method can be implemented by an intelligent recommendation device integrating user feedback, and the intelligent recommendation device integrating user feedback can be a terminal or a server. As Figure 1 shown in the flowchart of the intelligent recommendation method integrating user feedback, the processing flow of this method can include the following steps:

[0031] S1. Use the main recommendation algorithm to recommend short videos to the user to be tested, and obtain the corresponding tags to be tested; based on the tags to be tested, collect the feedback data of the user to be tested on the short videos with the tags to be tested in the secondary recommendation within a preset time period; the tags to be tested represent the tags corresponding to the short videos for which the user to be tested has a feedback behavior for the first time; the feedback data includes explicit feedback data and implicit feedback data.

[0032] In a feasible implementation manner, the main recommendation algorithm adopted by the present invention is the collaborative filtering recommendation algorithm, and the type of the collaborative filtering recommendation algorithm adopted is user-based collaborative filtering. Specifically, based on the similarity of user behaviors, other users (referred to as "neighbors") whose behaviors are similar to those of the target user are found, and then the items that these neighbors like but the target user has not yet accessed are recommended. For example, if user A and user B have the same ratings for most movies, then according to user B's ratings, movies that B has watched but A has not can be recommended to user A; the secondary recommendation algorithm is the content matching algorithm, and the core of the content matching algorithm is to recommend content similar to the items that the user has liked historically based on the characteristics of the items (such as text, pictures, audio, etc.). Each item is regarded as being described by a series of characteristics (attributes), and the user's interests are expressed through their interactions with the items (such as clicks, views, purchases, etc.). For example, in a news recommendation device, the articles that the user has read may be related to politics or technology, so the device will recommend more articles of similar themes, or in a short video recommendation device, the user has liked a certain type of video before (such as funny videos or food videos), and the device will recommend other similar videos based on these content characteristics; in addition, the to-be-tested tags represent the types of various videos in the to-be-tested short video software, such as: entertainment, news, sports, and learning, etc., and common to-be-tested short video software includes Douyin, Kuaishou, and Bilibili, etc.; through the method provided by the present invention, not only the access frequency of the user to the short videos of specific tags is accurately quantified, but also the intelligent recommendation of the user's videos is more accurate.

[0033] Optionally, collect the feedback data of the to-be-tested users on the short videos of the to-be-tested tags within a preset time period, including:

[0034] Based on the to-be-tested tags, obtain the explicit feedback data through the log records of the to-be-tested short video software; the explicit feedback data includes the number of user comments, the number of user positive feedbacks, and the number of user negative feedbacks corresponding to the to-be-tested users;

[0035] Based on the to-be-tested tags, obtain the implicit feedback data through the behavior records of the to-be-tested short video software; the implicit feedback data includes the browsing time, the number of searches, and the number of bounce-outs of the to-be-tested users.

[0036] In a feasible implementation manner, by collecting the explicit feedback (comments, likes, collections, reports, "not interested") and implicit feedback (browsing time, number of searches, number of bounce-outs) of users on short videos, the behaviors and interests of users are comprehensively captured; the explicit feedback directly reflects the preferences and attitudes of users, and the implicit feedback provides potential access clues of users. Combining these two types of data enables the to-be-tested short video software to more accurately identify the access situation of users to short videos of specific tags, thereby improving the accuracy and personalization effect of the intelligent recommendation of the to-be-tested short video software, and ultimately enhancing the user experience.

[0037] Optionally, after collecting the feedback data of the user to be tested on the short videos of the tags to be tested in the secondary recommendation within a preset time period, the method further includes:

[0038] Performing data processing on the implicit feedback data;

[0039] Performing data processing on the implicit feedback data includes:

[0040] Querying to obtain the viewing duration according to the viewing time; the viewing duration represents the time interval between the end time and the start time when the user to be tested views the short video of the tag to be tested.

[0041] In a feasible implementation manner, the preset time period represents the set test cycle, usually 24 hours, 12 hours, 8 hours, etc.; by further processing the implicit feedback data (calculating the viewing duration and bounce rate), implicit feedback data that is more convenient for subsequent algorithm calculation is obtained, which helps to further analyze in depth the frequency of the user to be tested accessing the short videos of the tags to be tested; the viewing duration provides the time length of the user's actual viewing content to evaluate the user's attention, while the bounce rate reveals the acceptance and satisfaction of the user with the content of the short videos of the tags to be tested; through the above data processing steps, the short video software to be tested can more accurately evaluate the true access willingness and participation degree of the user to the short videos of specific tags, so as to make the recommendation more in line with the user's preferences and improve the recommendation effect and user experience.

[0042] S2. Based on the explicit feedback correction factor of the preset database, evaluating according to the feedback data to obtain the explicit recommendation evaluation index; based on the implicit feedback correction factor of the preset database, evaluating according to the feedback data to obtain the implicit recommendation evaluation index; based on the preset database, obtaining the comprehensive recommendation evaluation index according to the explicit recommendation evaluation index and the implicit recommendation evaluation index; the explicit recommendation evaluation index is used to directly reflect the frequency of the user to be tested accessing the short videos of the tags to be tested; the implicit recommendation evaluation index is used to indirectly reflect the frequency of the user to be tested accessing the short videos of the tags to be tested; the comprehensive recommendation evaluation index is used to comprehensively quantify the frequency of the user to be tested accessing the short videos of the tags to be tested.

[0043] In a feasible implementation manner, the explicit feedback correction factor is the weight factor corresponding to the explicit feedback data of the user to be tested in the preset database, representing the difference between the value indicating the impact of the user's misoperation on the explicit recommendation evaluation index and the value 1. When in use, the explicit feedback correction factor corresponding to the explicit feedback data can be directly obtained from the preset database, and its corresponding relationship can be a pre-set mapping relationship. For example, the sum of the number of user comments, the number of positive feedbacks, and the number of negative feedbacks corresponding to the explicit feedback data forms a mapping set with the explicit recommendation evaluation index corresponding to the explicit feedback data in the preset database. The real-time explicit feedback data is input into the mapping set to obtain the corresponding explicit feedback correction factor, and the mapping relationship therein can be a one-to-one or many-to-one relationship. In the present invention, the value range of the explicit feedback correction factor is [0, 1].

[0044] Optionally, based on the explicit feedback correction factor of the preset database, evaluate according to the feedback data to obtain the explicit recommendation evaluation index, including:

[0045] Number the user to be tested and the preset time period;

[0046] Judge whether the explicit feedback data are all 0; if the explicit feedback data are all 0, record the explicit recommendation evaluation index of the corresponding user to be tested as 0; otherwise, based on the number of the preset time period, obtain the explicit recommendation evaluation index of the user to be tested according to the explicit feedback data and the explicit feedback correction factor;

[0047] Based on the number of the preset time period, obtain the explicit recommendation evaluation index of the user to be tested according to the explicit feedback data and the explicit feedback correction factor, including:

[0048] Judge whether the number of the preset time period is equal to 1; if the number of the preset time period is not equal to 1, obtain the change situation of the explicit feedback data according to the explicit feedback data of the current preset time period and the explicit feedback data of the previous preset time period, and obtain the explicit recommendation evaluation index of the user to be tested based on the obtained change situation of the explicit feedback data and the explicit feedback correction factor; otherwise, obtain the explicit recommendation evaluation index of the user to be tested according to the explicit feedback data of the current preset time period and the explicit feedback correction factor.

[0049] In a feasible implementation manner, the explicit recommendation evaluation index is calculated using the following formula (1):

[0050]

[0051] In the formula, t represents the number of the preset time period, t = 1, 2,... T, T represents the total number of preset time periods, n represents the number of the user to be tested, n = 1, 2,... N, N represents the total number of users to be tested, and τ represents the explicit feedback correction factor. represents the number of user comments of the nth user to be tested in the tth preset time period, represents the number of user comments of the nth user to be tested in the (t - 1)th preset time period, represents the number of positive feedbacks of the nth user to be tested in the tth preset time period, represents the number of positive feedbacks of the nth user to be tested in the (t - 1)th preset time period, represents the number of negative feedbacks of the nth user to be tested in the tth preset time period, represents the number of negative feedbacks of the nth user to be tested in the (t - 1)th preset time period, represents the explicit recommendation evaluation index of the nth user to be tested in the tth preset time period, and e represents the natural constant.

[0052] The present invention comprehensively analyzes the explicit recommendation evaluation index of the user to be tested by combining the explicit feedback data and the explicit feedback correction factor. Due to the differences in the preset time period situations, the algorithm has two forms of performance. The preset time period situation of the first form is that the current tested preset time period is not the first preset time period (i.e., t ≠ 1, that is, t - 1 ≠ 0). Therefore, the algorithm can rely on the explicit feedback data of the previous preset time period, and in the formula, the number of user comments and the number of positive feedbacks of the user are positively correlated with the explicit recommendation evaluation index, while the number of negative feedbacks of the user is negatively correlated with the explicit recommendation evaluation index. The preset time period situation of the second form is that the current tested preset time period is the first preset time period (i.e., t = 1). At this time, there is no explicit feedback data of the previous preset time period, so the initial explicit feedback data are all set to 1 (that is, the explicit feedback data of t - 1 are all 1). The correlation between the explicit recommendation evaluation index and the explicit feedback data in the algorithm is the same as that of the first form. The larger the value of the explicit recommendation evaluation index, the higher the access frequency of the user to be tested to the short video of the tested label. By judging whether the explicit feedback data is 0, it can be identified whether the user has no interaction with the short video of the tested label at all, avoiding the interference of invalid data on the evaluation. When the explicit feedback data is valid, the explicit recommendation evaluation index is calculated in combination with the correction factor, objectively reflecting the explicit feedback of the user to be tested on the content of the short video of the tested label. The reliability and accuracy of the explicit feedback evaluation are improved by the method provided in this embodiment, which helps the short video software to be tested to make more appropriate adjustments during recommendation optimization, thereby improving the user experience.

[0053] Optionally, after obtaining the explicit recommendation evaluation index by evaluating according to the feedback data based on the explicit feedback correction factor of the preset database, the method further includes:

[0054] Judging the explicit recommendation evaluation index and the explicit evaluation threshold obtained from the preset database;

[0055] Judge the explicit recommendation evaluation index against the explicit evaluation threshold obtained from a preset database, including:

[0056] If the explicit recommendation evaluation index is less than the explicit evaluation threshold, send a user feedback questionnaire for investigation within the preset time period of the next paragraph to obtain negative feedback records; according to the negative feedback records of the user to be tested, reduce the push of the short videos with the corresponding tags to be tested;

[0057] If the explicit recommendation evaluation index is greater than or equal to the explicit evaluation threshold, push the short videos with the tags to be tested to the user to be tested at a preset increase ratio within the preset time period of the next paragraph.

[0058] In a feasible implementation, if the explicit recommendation evaluation index is less than the explicit evaluation threshold, send a user feedback questionnaire within the next preset time period and reduce the push of the corresponding short videos with the tags to be tested according to the negative feedback records of the user to be tested. The negative feedback records include report records and "not interested" click records; if the explicit recommendation evaluation index is not less than the explicit evaluation threshold, push the short videos with the tags to be tested to the user to be tested at a preset increase ratio within the next preset time period. The preset increase ratio represents the sum of the preset ratio and the increase ratio. The increase ratio represents the value set for increasing the preset ratio. The preset increase ratio cannot be greater than the push limit value. The push limit value represents the maximum push ratio of a short video with a tag to be tested specified by the short video software to be tested.

[0059] The preset increase ratio is the sum of the preset ratio and the increase ratio. The preset ratio and the increase ratio are set according to the push rules required by the test, while the push limit value is set according to the push regulations of the short video software to be tested. For example, assume the preset ratio is 0.6, the increase ratio is 0.1, and the push limit value is 0.9. Then the preset increase ratio is the sum of the preset ratio and the increase ratio, which is 0.7. If in the next preset time period, the explicit recommendation evaluation index is still not less than the explicit evaluation threshold, then the preset increase ratio is the sum of 0.7 and the increase ratio 0.1, which is 0.8, and so on. When the preset increase ratio reaches 0.9, it indicates that the current push ratio of the short video has reached the maximum limit value specified by the push regulations of the short video platform to be tested, and the preset increase ratio will no longer be increased.

[0060] By comprehensively evaluating the explicit recommendation evaluation index and the explicit evaluation threshold, the short video recommendation strategy is dynamically adjusted. For users whose explicit evaluation index is lower than the explicit evaluation threshold, the device collects explicit feedback by sending out questionnaires and improves the user experience by reducing the push volume. For the users to be tested whose explicit evaluation index is higher than the explicit evaluation threshold, the short video software to be tested increases the push volume according to a preset increase ratio, but the push volume is controlled by the push limit. This push method can prevent over-recommendation; the method provided by the present invention optimizes the recommendation frequency and content matching degree, and improves the user satisfaction and participation.

[0061] The explicit evaluation threshold is represented by the minimum value of the range of the explicit recommendation evaluation index that is expected to be achieved by the push of the short video software to be tested. For example, if the range of the explicit recommendation evaluation index that is expected to be achieved by the push of the short video software to be tested is [0.3, 1], then 0.3 is set as the explicit evaluation threshold.

[0062] Optionally, based on the implicit feedback correction factor of the preset database, the feedback data is evaluated to obtain the implicit recommendation evaluation index, including:

[0063] Number the users to be tested and the preset time periods;

[0064] Based on the numbers of the preset time periods, according to the implicit feedback data and the implicit feedback correction factor, obtain the implicit recommendation evaluation index of the users to be tested;

[0065] Based on the numbers of the preset time periods, according to the implicit feedback data and the implicit feedback correction factor, obtain the implicit recommendation evaluation index of the users to be tested, including:

[0066] Judge whether the number of the preset time period is equal to 1; if the number of the preset time period is not equal to 1, then obtain the change situation of the implicit feedback data according to the implicit feedback data of the current preset time period and the implicit feedback data of the previous preset time period respectively, and obtain the implicit recommendation evaluation index of the users to be tested based on the obtained change situation of the implicit feedback data and the implicit feedback correction factor; otherwise, obtain the implicit recommendation evaluation index of the users to be tested according to the implicit feedback data and the implicit feedback correction factor of the preset time period.

[0067] In a feasible implementation manner, the implicit recommendation evaluation index is calculated using the following formula (2):

[0068]

[0069] In the formula, t represents the number of the preset time period, t = 1, 2,... T, T represents the total number of the preset time periods, n represents the number of the users to be tested, n = 1, 2,... N, N represents the total number of the users to be tested, represents the implicit feedback correction factor, represents the browsing duration of the nth user to be tested in the tth preset time period, represents the browsing duration of the nth user to be tested in the (t - 1)th preset time period, represents the number of searches of the nth user to be tested in the tth preset time period, represents the number of searches of the nth user to be tested in the (t - 1)th preset time period, represents the bounce rate of the nth user to be tested in the tth preset time period, represents the bounce rate of the nth user to be tested in the (t - 1)th preset time period, represents the implicit recommendation evaluation index of the nth user to be tested in the tth preset time period, and e represents the natural constant.

[0070] The present invention comprehensively analyzes the implicit recommendation evaluation index of the user to be tested by combining the implicit feedback data and the implicit feedback correction factor. Moreover, the browsing duration, the number of searches, and the bounce rate are interrelated. At the same time, due to the differences in the preset time period situations, the algorithm has two forms of performance. The situation of the preset time period of the first form of performance is that the current tested preset time period is not the first preset time period (i.e., t ≠ 1, that is, t - 1 ≠ 0). Therefore, the algorithm can rely on the implicit feedback data of the previous preset time period, and in the formula, the browsing duration and the number of searches are positively correlated with the implicit recommendation evaluation index, while the bounce rate is negatively correlated with the implicit recommendation evaluation index. The situation of the preset time period of the second form of performance is that the current tested preset time period is the first preset time period (i.e., t = 1). At this time, there is no implicit feedback data of the previous preset time period, and the initial implicit feedback data are all set to 1 (that is, the implicit feedback data of t - 1 are all 1). The correlation between the implicit recommendation evaluation index and the implicit feedback data in the algorithm is the same as that of the first form of performance. The larger the value of the implicit recommendation evaluation index, the higher the access frequency of the user to be tested to the short video with the tested label. At the same time, once the user to be tested uses the short video software to be tested, implicit feedback data (browsing duration, bounce rate) will be generated. At the same time, the value of the number of searches may be 0, but there is no situation where the browsing duration, the number of searches, and the bounce rate are all 0 at the same time, and the browsing duration, the number of searches, and the bounce rate are interrelated. For example, the larger the value of the bounce rate, the smaller the value of the browsing duration may be, and the larger the value of the number of searches, the larger the value of the browsing duration may be. The method provided by the present invention objectively reflects the implicit feedback of the user to be tested on the content of the short video with the tested label, improves the reliability of the implicit feedback evaluation through the method provided by this embodiment, and helps the short video software to be tested dynamically adjust the push strategy.

[0071] The implicit feedback correction factor is the weight factor corresponding to the implicit feedback data of the user to be tested in the preset database, representing the difference between the value indicating the impact of the user's incorrect operation on the implicit recommendation evaluation index and the value 1. When in use, the implicit feedback correction factor corresponding to the implicit feedback data can be directly obtained from the preset database, and its corresponding relationship can be a pre-set mapping relationship. For example, the sum of the search times and bounce times corresponding to the implicit feedback data forms a mapping set with the implicit recommendation evaluation index corresponding to the implicit feedback data in the preset database, and the real-time implicit feedback data is input into the mapping set to obtain the corresponding implicit feedback correction factor, where the mapping relationship can be one-to-one or many-to-one. In the present invention, the value range of the implicit feedback correction factor is [0, 1].

[0072] Optionally, based on the implicit feedback correction factor of the preset database, after evaluating according to the feedback data and obtaining the implicit recommendation evaluation index, the method further includes:

[0073] Judging the implicit recommendation evaluation index and the implicit evaluation threshold obtained from the preset database:

[0074] If the implicit recommendation evaluation index is less than the implicit evaluation threshold, reduce the push of short videos with tags to be tested within the preset time period of the next segment;

[0075] If the implicit recommendation evaluation index is greater than or equal to the implicit evaluation threshold, push short videos with tags to be tested to the user to be tested according to the preset increase ratio within the preset time period of the next segment.

[0076] In a feasible implementation manner, by judging the implicit recommendation evaluation index, the short video push strategy is dynamically adjusted. If the implicit evaluation index is lower than the implicit evaluation threshold, reduce the push of short videos with tags to be tested and instead recommend other tag short videos for testing to increase the frequency of access of the user to be tested to new content; if the implicit evaluation index is above the implicit evaluation threshold, increase the push of short videos with tags to be tested; the relevance and personalization of the recommended content are improved by the method provided by the present invention, thereby improving the user stickiness and viewing experience.

[0077] The implicit evaluation threshold represents the minimum value of the range of the implicit recommendation evaluation index that the short video software to be tested is expected to reach for the push regulation. For example, if the range of the implicit recommendation evaluation index that the short video software to be tested is expected to reach for the push regulation is [0.6, 1], then 0.6 is set as the implicit evaluation threshold.

[0078] Optionally, based on the preset database, obtaining the comprehensive recommendation evaluation index according to the explicit recommendation evaluation index and the implicit recommendation evaluation index includes:

[0079] Obtaining the recommendation evaluation weight from the preset database; the recommendation evaluation weight includes the explicit evaluation weight and the implicit evaluation weight;

[0080] Calculate according to the explicit recommendation evaluation index, the implicit recommendation evaluation index, and the recommendation evaluation weight to obtain the comprehensive recommendation evaluation index of the user to be tested;

[0081] The calculation formula for the comprehensive recommendation evaluation index is as follows in formula (3):

[0082]

[0083] In the formula, ρ n represents the explicit evaluation weight of the nth user to be tested, and δ n represents the implicit evaluation weight of the nth user to be tested, represents the explicit recommendation evaluation index of the nth user to be tested in the tth preset time period, represents the implicit recommendation evaluation index of the nth user to be tested in the tth preset time period, represents the comprehensive recommendation evaluation index of the nth user to be tested in the tth preset time period, and π represents the irrational number.

[0084] In a feasible implementation manner, the present invention comprehensively analyzes the explicit recommendation evaluation index, the implicit recommendation evaluation index, and the recommendation evaluation weight to obtain the comprehensive recommendation evaluation index of the user to be tested.

[0085] The explicit recommendation evaluation index and the implicit recommendation evaluation index are independent of each other but jointly affect the comprehensive recommendation evaluation index, and as the explicit recommendation evaluation index and the implicit recommendation evaluation index increase, the comprehensive recommendation evaluation index also increases. The explicit recommendation evaluation index, the implicit recommendation evaluation index, and the comprehensive recommendation evaluation index are positively correlated. As Figure 2 shown, it is a schematic diagram of the change of the comprehensive recommendation evaluation index provided by the present invention. It can be seen from the figure that when both the explicit recommendation evaluation index and the implicit recommendation evaluation index change in the increasing direction, the image shows an upward trend, further verifying that the explicit recommendation evaluation index, the implicit recommendation evaluation index, and the comprehensive recommendation evaluation index are positively correlated; calculating the comprehensive recommendation evaluation index through the proportion of the explicit evaluation weight and the implicit evaluation weight enables the recommendation device of the short video software to be tested to more comprehensively consider user behavior, thereby optimizing the short video recommendation content of the short video software to be tested.

[0086] Assume that the explicit evaluation weight is 0.5 and the implicit evaluation weight is 0.5. The data change table of the comprehensive recommendation evaluation index of the user to be tested is obtained by combining the explicit recommendation evaluation index and the implicit recommendation evaluation index, as shown in Table 1 (Data Change Table of Comprehensive Recommendation Evaluation Index) specifically:

[0087] Table 1

[0088]

[0089] As can be seen from Table 1, the explicit recommendation evaluation index and the implicit recommendation evaluation index reflect the different levels of access frequency of the user to be tested to the short video content of the tag to be tested. For example, the explicit recommendation evaluation index of the user to be tested with the preset time period number 3 is 0.795, while the implicit recommendation evaluation index is 0.128, and the comprehensive recommendation evaluation index is 0.663. The explicit recommendation evaluation index of the user to be tested with the preset time period number 5 is 0.963, while the implicit recommendation evaluation index is 0.828, and the comprehensive recommendation evaluation index is 0.830. ， It indicates that the access frequency of the user to be tested to the short video of the tag to be tested in the 5th preset time period is higher than that in the 3rd preset time period. Generally, the comprehensive recommendation indexes in the table are all in the range of 0.663 to 0.830, indicating that the short video software to be tested can effectively integrate the explicit feedback data and the implicit feedback data to accurately quantify the user's interest.

[0090] The explicit evaluation weight is the weight corresponding to the explicit recommendation evaluation index in the preset database, which represents the value of the influence degree of the explicit recommendation evaluation index on the comprehensive recommendation evaluation index. When used, the weight corresponding to the explicit recommendation evaluation index can be directly obtained from the preset database, and its corresponding relationship can be a pre-set mapping relationship. For example, the explicit recommendation evaluation index and the weight corresponding to the comprehensive recommendation evaluation index in the preset database form a mapping set, and the real-time explicit recommendation evaluation index is input into the mapping set to obtain the corresponding weight, and the mapping relationship therein can be one-to-one or many-to-one relationship. In the present invention, the value range of the explicit evaluation weight is [0,1], and the sum of the implicit evaluation weight and the explicit evaluation weight in the present invention is 1.

[0091] S3. Compare the comprehensive recommendation evaluation index with the recommendation evaluation threshold obtained from the preset database to obtain the test result, and take recommendation measures according to the test result.

[0092] Optionally, comparing the comprehensive recommendation evaluation index with the recommendation evaluation threshold obtained from the preset database to obtain the test result, and taking recommendation measures according to the test result includes:

[0093] Compare the comprehensive recommendation evaluation index with the recommendation evaluation threshold obtained from the preset database:

[0094] If the comprehensive recommendation evaluation index is less than the recommendation evaluation threshold, the test result is that the user accidentally touches the video tag, and the main recommendation algorithm is used to recommend short videos to the user to be tested in the next preset time period;

[0095] If the comprehensive recommendation evaluation index is greater than or equal to the recommendation evaluation threshold, the test result is the user's interested video tag, and the secondary recommendation is used to recommend short videos to the user to be tested in the next preset time period.

[0096] In a feasible implementation manner, the method provided by the present invention effectively improves the relevance of the short video recommendation content, reduces mis-recommendations, and improves the user experience and stickiness.

[0097] Specifically, the recommendation evaluation threshold is obtained from a preset database. In a specific embodiment, the recommendation evaluation threshold represents the minimum value of the range of the comprehensive recommendation evaluation index that the short video software to be tested is expected to reach for pushing regulations. For example, if the range of the comprehensive recommendation evaluation index that the short video software to be tested is expected to reach for pushing regulations is [0.85, 1], then 0.85 is set as the recommendation evaluation threshold.

[0098] The beneficial effects brought by the technical solution provided by the embodiment of the present invention at least include:

[0099] By using the main recommendation algorithm to perform short video recommendations for the user to be tested, obtaining the corresponding tags to be tested, and collecting the feedback data of the user to be tested on the short videos with the tags to be tested in the secondary recommendation within a preset time period, then obtaining the explicit recommendation evaluation index and the implicit recommendation evaluation index based on the feedback data, and obtaining the comprehensive recommendation evaluation index based on the explicit recommendation evaluation index and the implicit recommendation evaluation index, and finally comparing the comprehensive recommendation evaluation index with the recommendation evaluation threshold obtained from the preset database to obtain the test result, and taking recommendation measures according to the test result, so as to more accurately evaluate the frequency of the user's access to the short videos with the current tags, and then realize more accurate intelligent recommendation of the user's videos, effectively solving the problem of inaccurate user feedback data in the existing user video recommendation.

[0100] The explicit recommendation evaluation index of the user to be tested is obtained through the explicit feedback data and the explicit feedback correction factor, which directly reflects the frequency of the user's access to the short videos with the current tags, and thus improves the accuracy of the explicit feedback evaluation.

[0101] The implicit recommendation evaluation index of the user to be tested is obtained through the implicit feedback data and the implicit feedback correction factor, which indirectly reflects the frequency of the user's access to the short videos with the current tags, and thus improves the reliability of the implicit feedback evaluation.

[0102] Figure 3 It is a block diagram of an intelligent recommendation device that fuses user feedback shown according to an exemplary embodiment. This device is used for the intelligent recommendation method that fuses user feedback. Refer to Figure 3 , this device includes a feedback data acquisition module 310, a feedback recommendation evaluation module 320, and a recommendation measure adjustment module 3300. Among them:

[0103] The feedback data acquisition module 310 is used to recommend short videos to the user to be tested using the main recommendation algorithm, and obtain the corresponding tags to be tested; based on the tags to be tested, collect the feedback data of the user to be tested on the short videos with the tags to be tested in the secondary recommendation within a preset time period; the tags to be tested represent the tags corresponding to the short videos for which the user to be tested has the first feedback behavior; the feedback data includes explicit feedback data and implicit feedback data;

[0104] The feedback recommendation evaluation module 320 is used to evaluate based on the explicit feedback correction factor in the preset database according to the feedback data to obtain the explicit recommendation evaluation index; evaluate based on the implicit feedback correction factor in the preset database according to the feedback data to obtain the implicit recommendation evaluation index; based on the preset database, obtain the comprehensive recommendation evaluation index according to the explicit recommendation evaluation index and the implicit recommendation evaluation index; the explicit recommendation evaluation index is used to directly reflect the access frequency of the user to be tested to the short videos with the tags to be tested; the implicit recommendation evaluation index is used to indirectly reflect the access frequency of the user to be tested to the short videos with the tags to be tested; the comprehensive recommendation evaluation index is used to comprehensively quantify the access frequency of the user to be tested to the short videos with the tags to be tested;

[0105] The recommendation measure adjustment module 330 is used to compare the comprehensive recommendation evaluation index with the recommendation evaluation threshold obtained from the preset database to obtain the test result, and take recommendation measures according to the test result.

[0106] Optionally, the feedback data acquisition module 310 is further used for:

[0107] Based on the tags to be tested, obtain the explicit feedback data through the log records of the short video software to be tested; the explicit feedback data includes the number of user comments, the number of positive feedbacks, and the number of negative feedbacks corresponding to the user to be tested;

[0108] Based on the tags to be tested, obtain the implicit feedback data through the behavior records of the short video software to be tested; the implicit feedback data includes the browsing time, the number of searches, and the number of bounce times of the user to be tested.

[0109] Optionally, the feedback data acquisition module 310 is further used for:

[0110] Perform data processing on the implicit feedback data;

[0111] Performing data processing on the implicit feedback data includes:

[0112] Query the browsing duration according to the browsing time; the browsing duration represents the time interval between the end time and the start time when the user to be tested browses the video with the tags to be tested.

[0113] Optionally, the feedback recommendation evaluation module 320 is further used for:

[0114] Number the test users and the preset time periods;

[0115] Determine whether the explicit feedback data are all 0; if the explicit feedback data to be processed are all 0, record the explicit recommendation evaluation index of the corresponding test user as 0; otherwise, based on the number of the preset time period, obtain the explicit recommendation evaluation index of the test user according to the explicit feedback data and the explicit feedback correction factor;

[0116] Obtaining the explicit recommendation evaluation index of the test user according to the explicit feedback data and the explicit feedback correction factor based on the number of the preset time period includes:

[0117] Determine whether the number of the preset time period is equal to 1; if the number of the preset time period is not equal to 1, obtain the change situation of the explicit feedback data according to the explicit feedback data of the current preset time period and the explicit feedback data of the previous preset time period, and obtain the explicit recommendation evaluation index of the test user based on the obtained change situation of the explicit feedback data and the explicit feedback correction factor; otherwise, obtain the explicit recommendation evaluation index of the test user according to the explicit feedback data of the current preset time period and the explicit feedback correction factor.

[0118] Optionally, the feedback recommendation evaluation module 320 is further configured to:

[0119] Judge the explicit recommendation evaluation index against the explicit evaluation threshold obtained from the preset database;

[0120] Judging the explicit recommendation evaluation index against the explicit evaluation threshold obtained from the preset database includes:

[0121] If the explicit recommendation evaluation index is less than the explicit evaluation threshold, send a user feedback questionnaire for investigation within the preset time period of the next segment to obtain negative feedback records; according to the negative feedback records of the test user, reduce the push of the short videos with the test tags corresponding to the negative feedback records;

[0122] If the explicit recommendation evaluation index is greater than or equal to the explicit evaluation threshold, push the short videos with the test tags to the test user according to the preset increase ratio within the preset time period of the next segment.

[0123] Optionally, the feedback recommendation evaluation module 320 is further configured to:

[0124] Number the test users and the preset time periods;

[0125] Based on the number of the preset time period, obtain the implicit recommendation evaluation index of the test user according to the implicit feedback data and the implicit feedback correction factor;

[0126] Based on the number of the preset time period, obtain the implicit recommendation evaluation index of the user to be tested according to the implicit feedback data and the implicit feedback correction factor, including:

[0127] Judge whether the number of the preset time period is equal to 1; if the number of the preset time period is not equal to 1, obtain the change situation of the implicit feedback data according to the implicit feedback data of the current preset time period and the implicit feedback data of the previous preset time period, and obtain the implicit recommendation evaluation index of the user to be tested based on the obtained change situation of the implicit feedback data and the implicit feedback correction factor; otherwise, obtain the implicit recommendation evaluation index of the user to be tested according to the implicit feedback data and the implicit feedback correction factor of the preset time period.

[0128] Optionally, the feedback recommendation evaluation module 320 is further configured to:

[0129] Judge the implicit recommendation evaluation index and the implicit evaluation threshold obtained from the preset database:

[0130] If the implicit recommendation evaluation index is less than the implicit evaluation threshold, reduce the push of short videos with the to-be-tested tags within the next preset time period;

[0131] If the implicit recommendation evaluation index is greater than or equal to the implicit evaluation threshold, push short videos with the to-be-tested tags to the user to be tested according to the preset increase ratio within the next preset time period.

[0132] Optionally, the feedback recommendation evaluation module 320 is further configured to:

[0133] Obtain the recommendation evaluation weight from the preset database; the recommendation evaluation weight includes the explicit evaluation weight and the implicit evaluation weight;

[0134] Calculate according to the explicit recommendation evaluation index, the implicit recommendation evaluation index and the recommendation evaluation weight to obtain the comprehensive recommendation evaluation index of the user to be tested;

[0135] The calculation formula of the comprehensive recommendation evaluation index is as follows in formula (1):

[0136]

[0137] In the formula, ρ n represents the explicit evaluation weight of the nth user to be tested, δ n represents the implicit evaluation weight of the nth user to be tested, represents the explicit recommendation evaluation index of the nth user to be tested in the tth preset time period, represents the implicit recommendation evaluation index of the nth user to be tested in the tth preset time period, represents the comprehensive recommendation evaluation index of the nth user to be tested in the tth preset time period, and π represents the irrational number.

[0138] Optionally, the recommended measure adjustment module 330 is further configured to:

[0139] Compare the comprehensive recommended evaluation index with the recommended evaluation threshold obtained from the preset database:

[0140] If the comprehensive recommended evaluation index is less than the recommended evaluation threshold, the test result is that the user accidentally touches the video tag, and the main recommendation algorithm is used to recommend short videos to the user to be tested within the preset time period of the next paragraph;

[0141] If the comprehensive recommended evaluation index is greater than or equal to the recommended evaluation threshold, the test result is the user's interested video tag, and the secondary recommendation is used to recommend short videos to the user to be tested within the preset time period of the next paragraph.

[0142] The beneficial effects brought by the technical solution provided by the embodiment of the present invention at least include:

[0143] By using the main recommendation algorithm to recommend short videos to the user to be tested, obtaining the corresponding tags to be tested, and collecting the feedback data of the user to be tested on the short video tags in the secondary recommendation within the preset time period, then obtaining the explicit recommended evaluation index and the implicit recommended evaluation index based on the feedback data, and obtaining the comprehensive recommended evaluation index based on the explicit recommended evaluation index and the implicit recommended evaluation index, and finally comparing the comprehensive recommended evaluation index with the recommended evaluation threshold obtained from the preset database to obtain the test result, and taking recommended measures according to the test result, so as to more accurately evaluate the frequency of the user's access to the short video of the current tag, and then realize more accurate intelligent recommendation of the user's video, effectively solving the problem of inaccurate user feedback data in the existing user video recommendation.

[0144] The explicit recommended evaluation index of the user to be tested is obtained through the explicit feedback data and the explicit feedback correction factor, which directly reflects the frequency of the user's access to the short video of the current tag, and then improves the accuracy of the explicit feedback evaluation.

[0145] The implicit recommended evaluation index of the user to be tested is obtained through the implicit feedback data and the implicit feedback correction factor, which indirectly reflects the frequency of the user's access to the short video of the current tag, and then improves the reliability of the implicit feedback evaluation.

[0146] Figure 4 It is a schematic structural diagram of an intelligent recommendation device that integrates user feedback provided by an embodiment of the present invention, as Figure 4 shown. The intelligent recommendation device that integrates user feedback may include the intelligent recommendation device that integrates user feedback shown above Figure 3 shown. Optionally, the intelligent recommendation device 410 that integrates user feedback may include a first processor 2001.

[0147] Optionally, the intelligent recommendation device 410 that incorporates user feedback may further include a memory 2002 and a transceiver 2003.

[0148] Among them, the first processor 2001 is connected to the memory 2002 and the transceiver 2003, such as through a communication bus.

[0149] Next, in conjunction with Figure 4 Specific introductions will be made to the various components of the intelligent recommendation device 410 that incorporates user feedback:

[0150] Among them, the first processor 2001 is the control center of the intelligent recommendation device 410 that incorporates user feedback. It can be a single processor or a collective term for multiple processing elements. For example, the first processor 2001 is one or more central processing units (CPUs), or it can be an application specific integrated circuit (ASIC), or it can be one or more integrated circuits configured to implement the embodiments of the present invention. For example: one or more digital signal processors (DSPs), or one or more field programmable gate arrays (FPGAs).

[0151] Optionally, the first processor 2001 can execute various functions of the intelligent recommendation device 410 that incorporates user feedback by running or executing software programs stored in the memory 2002 and by calling data stored in the memory 2002.

[0152] In a specific implementation, as an embodiment, the first processor 2001 may include one or more CPUs, such as Figure 4 the CPU0 and CPU1 shown in

[0153] In a specific implementation, as an embodiment, the intelligent recommendation device 410 that incorporates user feedback may also include multiple processors, such as Figure 4 the first processor 2001 and the second processor 2004 shown in

[0154] Among them, the memory 2002 is used to store the software program for implementing the solution of the present invention and is controlled by the first processor 2001 for execution. The specific implementation manner can refer to the above method embodiment and will not be elaborated here.

[0155] Optionally, the memory 2002 can be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or can also be an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but not limited thereto. The memory 2002 can be integrated with the first processor 2001 or exist independently and is coupled to the first processor 2001 through the interface circuit of the intelligent recommendation device 410 that integrates user feedback ( Figure 4 not shown in the figure), and the embodiments of the present invention do not make specific limitations on this.

[0156] The transceiver 2003 is used to communicate with a network device or with a terminal device.

[0157] Optionally, the transceiver 2003 can include a receiver and a transmitter ( Figure 4 not shown separately in the figure). Among them, the receiver is used to implement the receiving function, and the transmitter is used to implement the sending function.

[0158] Optionally, the transceiver 2003 can be integrated with the first processor 2001 or exist independently and is coupled to the first processor 2001 through the interface circuit of the intelligent recommendation device 410 that integrates user feedback ( Figure 4 not shown in the figure), and the embodiments of the present invention do not make specific limitations on this.

[0159] It should be noted that Figure 4 the structure of the intelligent recommendation device 410 that integrates user feedback shown in the figure does not constitute a limitation on the router. The actual knowledge structure recognition device may include more or fewer components than shown in the figure, or combine some components, or have different component arrangements.

[0160] In addition, for the technical effects of the intelligent recommendation device 410 that incorporates user feedback, reference may be made to the technical effects of the intelligent recommendation method that incorporates user feedback described in the foregoing method embodiments, and details thereof will not be elaborated herein.

[0161] It should be understood that the first processor 2001 in the embodiments of the present invention may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0162] It should also be understood that the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable ROM (PROM), an erasable programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of random access memory (RAM) are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0163] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that contains one or more collections of available media. The available media can be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0164] It should be understood that the term "and / or" in this document is merely a description of the association relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. Additionally, the character " / " in this document generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, which can be understood specifically with reference to the context before and after.

[0165] In the present invention, "at least one" means one or more, and "a plurality" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can represent: a, b, c, a - b, a - c, b - c, or a - b - c, where a, b, and c can be single or multiple.

[0166] It should be understood that in various embodiments of the present invention, the magnitudes of the sequence numbers of the above processes do not imply the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0167] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in connection with the embodiments disclosed herein can be implemented in electronic hardware, or in a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Skilled professionals can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present invention.

[0168] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the devices, apparatuses, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0169] In several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0170] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0171] In addition, the functional units in each embodiment of the present invention can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0172] When the above-mentioned function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0173] As described above, the above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. An intelligent recommendation method integrating user feedback, characterized in that: include; Use the main recommendation algorithm to recommend short videos to the test user, and obtain the corresponding test tag; based on the test tag, collect the feedback data of the test user on the short video with the test tag in the secondary recommendation within a preset time period; the test tag represents the tag corresponding to the short video for which the test user has the first feedback behavior; the feedback data includes explicit feedback data and implicit feedback data; Based on the explicit feedback correction factor of the preset database, the feedback data is evaluated to obtain the explicit recommendation evaluation index; based on the implicit feedback correction factor of the preset database, the feedback data is evaluated to obtain the implicit recommendation evaluation index; Based on the preset database, a comprehensive recommendation evaluation index is obtained according to the explicit recommendation evaluation index and the implicit recommendation evaluation index; the explicit recommendation evaluation index is used to directly reflect the frequency of visits to the short video with the test label by the user to be tested; the implicit recommendation evaluation index is used to indirectly reflect the frequency of visits to the short video with the test label by the user to be tested; the comprehensive recommendation evaluation index is used to comprehensively quantify the frequency of visits to the short video with the test label by the user to be tested; The comprehensive recommendation evaluation index is compared with the recommendation evaluation threshold obtained from a preset database to obtain a test result, and a recommended measure is taken according to the test result.

2. The intelligent recommendation method integrating user feedback according to claim 1, characterized in that: The step of collecting feedback data of the users to be tested on the short videos with the tags to be tested in the recommendation within a preset time period based on the tags to be tested includes: Based on the tag to be tested, explicit feedback data is obtained through the log record of the short video software to be tested; the explicit feedback data includes the number of user comments, the number of user positive feedback, and the number of user negative feedback corresponding to the user to be tested; Based on the tag to be tested, implicit feedback data is obtained through the behavior record of the short video software to be tested; the implicit feedback data includes the browsing time, search times and bounce times of the user to be tested.

3. The intelligent recommendation method integrating user feedback according to claim 2, characterized in that: After collecting the feedback data of the tested users on the short video with the tested label in the secondary recommendation within the preset time period, the method further includes: Performing data processing on the implicit feedback data; The processing of the implicit feedback data includes: The browsing time is obtained by querying the browsing time; the browsing time represents the time interval between the end time and the start time of the tested user browsing the tested tagged video.

4. The intelligent recommendation method integrating user feedback according to claim 2, characterized in that: The explicit feedback correction factor based on the preset database is evaluated according to the feedback data to obtain the explicit recommendation evaluation index, including: Number the users to be tested and the preset time periods; Determine whether the explicit feedback data are all 0; if the explicit feedback data are all 0, record the explicit recommendation evaluation index of the corresponding user to be tested as 0; otherwise, based on the number of the preset time period, obtain the explicit recommendation evaluation index of the user to be tested according to the explicit feedback data and the explicit feedback correction factor; The method of obtaining the explicit recommendation evaluation index of the user to be tested based on the number of the preset time period and the explicit feedback data and the explicit feedback correction factor includes: Determine whether the number of the preset time period is equal to 1; if the number of the preset time period is not equal to 1, obtain the change of the explicit feedback data according to the explicit feedback data of the preset time period of the current segment and the explicit feedback data of the preset time period of the previous segment, and obtain the explicit recommendation evaluation index of the user to be tested based on the obtained change of the explicit feedback data and the explicit feedback correction factor; otherwise, obtain the explicit recommendation evaluation index of the user to be tested according to the explicit feedback data of the preset time period of the current segment and the explicit feedback correction factor.

5. The intelligent recommendation method integrating user feedback according to claim 4, characterized in that: After the explicit feedback correction factor based on the preset database is evaluated according to the feedback data to obtain the explicit recommendation evaluation index, the method further includes: The explicit recommendation evaluation index is compared with an explicit evaluation threshold obtained from a preset database; The step of judging the explicit recommendation evaluation index against an explicit evaluation threshold value obtained from a preset database includes: If the explicit recommendation evaluation index is less than the explicit evaluation threshold, a user feedback questionnaire is sent to conduct a survey within the next preset time period to obtain a negative feedback record; according to the negative feedback record of the user to be tested, the short video with the tag to be tested corresponding to the negative feedback record is pushed less; If the explicit recommendation evaluation index is greater than or equal to the explicit evaluation threshold, the short video with the tag to be tested is pushed to the user to be tested according to a preset increase ratio within a preset time period of the next section.

6. The intelligent recommendation method integrating user feedback according to claim 2, characterized in that: The implicit feedback correction factor based on the preset database is evaluated according to the feedback data to obtain the implicit recommendation evaluation index, including: Number the users to be tested and the preset time periods; Based on the number of the preset time period, according to the implicit feedback data and the implicit feedback correction factor, obtaining the implicit recommendation evaluation index of the user to be tested; The number based on the preset time period, according to the implicit feedback data and the implicit feedback correction factor, obtains the implicit recommendation evaluation index of the user to be tested, including: Determine whether the number of the preset time period is equal to 1; if the number of the preset time period is not equal to 1, obtain the change of implicit feedback data according to the implicit feedback data of the preset time period of the current section and the implicit feedback data of the preset time period of the previous section, and obtain the implicit recommendation evaluation index of the user to be tested based on the obtained change of implicit feedback data and the implicit feedback correction factor; otherwise, obtain the implicit recommendation evaluation index of the user to be tested according to the implicit feedback data of the preset time period and the implicit feedback correction factor.

7. The intelligent recommendation method integrating user feedback according to claim 6, characterized in that: After the implicit feedback correction factor based on the preset database is evaluated according to the feedback data to obtain the implicit recommendation evaluation index, the method further includes: The implicit recommendation evaluation index is compared with the implicit evaluation threshold obtained from the preset database: If the implicit recommendation evaluation index is less than the implicit evaluation threshold, the push of short videos with the test tag is reduced in the next preset time period; If the implicit recommendation evaluation index is greater than or equal to the implicit evaluation threshold, the short video with the tag to be tested is pushed to the user to be tested according to a preset increase ratio within a preset time period of the next section.

8. The intelligent recommendation method integrating user feedback according to claim 6, characterized in that: The method of obtaining a comprehensive recommendation evaluation index based on the preset database and the explicit recommendation evaluation index and the implicit recommendation evaluation index includes: Obtaining recommended evaluation weights from a preset database; the recommended evaluation weights include explicit evaluation weights and implicit evaluation weights; Calculate the comprehensive recommendation evaluation index of the user to be tested according to the explicit recommendation evaluation index, the implicit recommendation evaluation index and the recommendation evaluation weight; The calculation formula of the comprehensive recommendation evaluation index is as follows (1): In the formula, ρ n represents the explicit evaluation weight of the nth user to be tested, δ n represents the implicit evaluation weight of the nth user to be tested, represents the explicit recommendation evaluation index of the nth user to be tested in the tth preset time period, represents the implicit recommendation evaluation index of the nth user to be tested in the tth preset time period, represents the comprehensive recommendation evaluation index of the nth user to be tested in the tth preset time period, and π represents an irrational number.

9. The intelligent recommendation method integrating user feedback according to claim 8, characterized in that: The taking of recommended measures according to the test results includes: Compare the comprehensive recommendation evaluation index with the recommendation evaluation threshold obtained from the preset database: If the comprehensive recommendation evaluation index is less than the recommendation evaluation threshold, the test result is that the user accidentally touches the video tag, and the main recommendation algorithm is used to recommend short videos to the user to be tested in the next preset time period; If the comprehensive recommendation evaluation index is greater than or equal to the recommendation evaluation threshold, the test result is a user interest video tag, and the secondary recommendation is used to recommend short videos to the user to be tested within the next preset time period.

10. An intelligent recommendation device integrating user feedback, wherein the intelligent recommendation device integrating user feedback is used to implement the intelligent recommendation method integrating user feedback as claimed in any one of claims 1 to 9, characterized in that: The device comprises: The feedback data acquisition module is used to recommend short videos to the test user using the main recommendation algorithm and obtain the corresponding test tag; based on the test tag, collect the feedback data of the test user on the short video with the test tag in the secondary recommendation within a preset time period; the test tag represents the tag corresponding to the short video for which the test user has the first feedback behavior; the feedback data includes explicit feedback data and implicit feedback data; A feedback recommendation evaluation module is used to evaluate the feedback data based on the explicit feedback correction factor of the preset database to obtain an explicit recommendation evaluation index; to evaluate the feedback data based on the implicit feedback correction factor of the preset database to obtain an implicit recommendation evaluation index; based on the preset database, a comprehensive recommendation evaluation index is obtained according to the explicit recommendation evaluation index and the implicit recommendation evaluation index; the explicit recommendation evaluation index is used to directly reflect the frequency of visits to the short video with the test label by the user to be tested; the implicit recommendation evaluation index is used to indirectly reflect the frequency of visits to the short video with the test label by the user to be tested; the comprehensive recommendation evaluation index is used to comprehensively quantify the frequency of visits to the short video with the test label by the user to be tested; The recommended measure adjustment module is used to compare the comprehensive recommended evaluation index with the recommended evaluation threshold obtained from a preset database to obtain a test result, and take recommended measures according to the test result.

Images