Education resource personalized recommendation method and system based on artificial intelligence

Abstract

The invention relates to the technical field of personalized recommendation, and particularly discloses an educational resource personalized recommendation method and system based on artificial intelligence, and the method comprises the following steps: S1, screening video course titles containing keywords according to knowledge point keywords of a user, and obtaining a corresponding video course; s2, acquiring historical watching data conforming to the video class, wherein the historical watching data comprise total watching times, comments and favorable comment rates; calculating the key proportion of the video class through the watching integrity rate and the watching speed curve; s3, calculating a recommendation value of each conforming video course according to the favorable comment rate, the watching integrity rate and the key proportion value, and recommending the video course to the user based on the recommendation value; according to the method, personalized education resource recommendation is realized through multi-dimensional data analysis.

Description

Technical Field

[0001] The present invention relates to the field of personalized recommendation technology, and in particular to a method and system for personalized recommendation of educational resources based on artificial intelligence. Background Art

[0002] Educational resources refer to the sum of various elements and conditions that support educational and teaching activities, encompassing human, material, financial, and information resources. These resources include teaching equipment, teaching funds, and network resources. Network resources, based on the internet or educational cloud platforms, help teachers and students manage learning tasks and projects, simulate real-world classroom environments, and support interactive learning.

[0003] In today's era of information explosion, with the rapid development and widespread adoption of internet technology, online resources are becoming richer and more diverse than ever before. Among them, video courses, a highly engaging and practical learning resource, are experiencing an astonishing growth in both quantity and variety. From academic knowledge and vocational skills training to hobby development and everyday life, video courses offer a wide range of topics, covering nearly every aspect of learning and daily life.

[0004] However, precisely because of the overwhelming abundance and diversity of video course resources, users often feel overwhelmed when faced with such a vast online landscape. When they attempt to search for a video course that meets their specific needs in the vast online world, they find it challenging. Firstly, search engines often return thousands or even tens of thousands of links to video courses, requiring users to sift through and identify them, which is not only inefficient but also exhausting. Secondly, the quality of video courses varies greatly across platforms, with some content not being accurate, complete, or tailored to users' specific needs, further complicating the selection process.

[0005] In this context, a personalized video course recommendation method is urgently needed. This method can deeply understand users' individual needs, such as their learning goals, interests, knowledge level, and study habits. By accurately analyzing and matching this information, it can tailor a list of video course recommendations to meet their unique needs. This eliminates the need for users to blindly search through a vast sea of video courses. Instead, users can more efficiently and conveniently find high-quality video course resources that are truly valuable and suitable for them, thereby improving learning efficiency and effectiveness and better meeting their learning and development needs. Summary of the Invention

[0006] The purpose of the present invention is to provide an artificial intelligence-based personalized recommendation method and system for educational resources to solve the above technical problems.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] A personalized recommendation method for educational resources based on artificial intelligence, comprising the following steps:

[0009] Step S1: Obtain the user's knowledge point keywords and obtain the titles of all video courses, filter out the titles containing the knowledge point keywords from all titles, record them as matching titles, and obtain the video courses corresponding to the matching titles, record them as matching video courses;

[0010] Step S2: Obtain historical viewing data of the matching video course, and obtain the total number of views and all comments of the matching video course, filter out positive comments from all comments, and obtain the favorable comment rate; obtain the viewing completion rate of the matching video course based on the historical viewing data, and obtain the viewing speed curve of historical users; and obtain the key point ratio of the matching video course based on the viewing speed curve;

[0011] Step S3: Obtain a recommendation value of the matching video course based on the praise rate, viewing completion rate, and key point ratio of the matching video course; and recommend each matching video course to the user based on the recommendation value of each matching video course.

[0012] As a further solution of the present invention: the process of screening out the matching titles includes:

[0013] Obtain the text of the knowledge point keyword and record it as the keyword text, and obtain the text of the title and record it as the title text; divide the title text into several words based on natural language processing technology, and obtain the semantic similarity between the keyword text and each word in turn; if there is any word whose semantic similarity with the keyword text exceeds a preset similarity threshold, then the title is a matching title.

[0014] As a further solution of the present invention: the historical viewing data includes all historical users who watched the corresponding video course, as well as the viewing time and viewing speed data of each historical user. The viewing speed data is the real-time speed at which the historical users watched the corresponding video course.

[0015] As a further solution of the present invention, the process of screening out the positive reviews includes:

[0016] Setting a sentiment tendency, which includes a positive tendency and a negative tendency; obtaining a number of comments as samples, recorded as comment samples, and marking the sentiment tendency of each comment sample to obtain sample data; establishing a learning model based on natural language processing technology, inputting the sample data into the learning model, training the learning model, and obtaining a sentiment tendency analysis model; analyzing the sentiment tendency of the comment through the sentiment tendency analysis model, if the sentiment tendency of the comment is a positive tendency, the comment is a positive comment; if the sentiment tendency of the comment is a negative tendency, the comment is a negative comment.

[0017] As a further solution of the present invention: the process of obtaining the favorable comment rate and the complete viewing rate includes:

[0018] The positive review rate Ar = Pr / Tn, where Pr is the total number of positive reviews and Tn is the total number of views; obtain the viewing time of the historical user and the total duration of the matching video course to obtain the viewing rate of the historical user Vr = t / T, where t represents the viewing time of the historical user and T represents the total duration of the matching video course; set the viewing rate threshold Vr th , and Vr th ≥50%, record the historical users whose viewing rate exceeds the viewing rate threshold as complete browsing users; obtain the total number num of the complete browsing users, and obtain the viewing completion rate Vor=num / Tn.

[0019] As a further solution of the present invention, the process of obtaining the viewing speed curve of the historical users includes:

[0020] Obtain the viewing time of the historical user for the matching video course, and obtain the viewing time period of the historical user for the matching video course based on the viewing time period; select a number of time nodes equally divided in the viewing time period, and the real-time speed is the speed of the user at each time node in the viewing time period, and the speed includes 1x speed, 1.5x speed, 2x speed and 0.5x speed;

[0021] Each time node is numbered, and a coordinate system is established with the number as the horizontal coordinate and the speed as the vertical coordinate; each numbered time node and its corresponding speed are converted into a coordinate point of a corresponding position on the coordinate system, and each coordinate point is connected with a smooth curve to obtain a viewing speed curve.

[0022] As a further solution of the present invention: the process of obtaining the key points proportion of the corresponding video course includes:

[0023] Numbering each coordinate point on the viewing speed curve according to the number of the time node; obtaining the average value of the coordinate points with the same number on the viewing speed curve of each historical user to obtain a new coordinate point with each number, wherein the average value of the coordinate point is the average value of the speed corresponding to the coordinate point; regenerating a new curve based on the new coordinate point with each number, and recording it as the average speed curve;

[0024] Set the dividing line sp = 1, where sp is the speed, place the dividing line and the average speed curve of the corresponding video course in the same coordinate system, intercept the curve segment below the dividing line on the average speed curve, record it as the slow segment, and obtain the total curve length L of the average speed curve all and the length L of the slow segment slow , get the key proportion value Kpr=L slow / L all .

[0025] As a further solution of the present invention: the recommended value of the corresponding video course Where K is the preset correction coefficient, and K>1, Ar ave Represents the average of the praise rates of all matching video courses, Vr ave Indicates the average viewing completion rate of all matching video courses, Kpr ave Represents the average value of the key points ratio of all matching video courses.

[0026] As a further solution of the present invention: a personalized recommendation system for educational resources based on artificial intelligence, comprising:

[0027] Retrieval module: obtains the user's knowledge point keywords and the titles of all video courses, filters out titles containing the knowledge point keywords from all titles, records them as matching titles, and obtains the video courses corresponding to the matching titles, records them as matching video courses;

[0028] Analysis module: Obtain historical viewing data of the matching video course, obtain the total number of views and all comments of the matching video course, filter out positive comments from all comments, and obtain the praise rate; obtain the viewing completion rate of the matching video course based on the historical viewing data, and obtain the viewing speed curve of historical users; obtain the key percentage of the matching video course based on the viewing speed curve;

[0029] Recommendation module: obtain the recommendation value of the matching video course according to the praise rate, viewing completion rate and key point ratio of the matching video course; recommend each matching video course to the user according to the recommendation value of each matching video course.

[0030] Beneficial effects of the present invention:

[0031] The present invention analyzes the user's knowledge point keywords to screen out video courses that match their needs, and calculates recommendation values based on multi-dimensional data such as the praise rate, viewing completion rate, and key point ratio, thereby providing users with highly personalized educational resource recommendations. By analyzing the viewing speed curve and key point ratio of historical users, the present invention identifies important content in video courses, helping users learn core knowledge points more efficiently and reducing time waste. Positive comments are screened through a sentiment analysis model to ensure the quality of recommended content and user satisfaction, thereby improving the user's learning experience. Historical viewing data and user behavior analysis are used to provide a scientific and objective recommendation basis, avoid subjective bias, and improve the accuracy and reliability of recommendations. By updating historical viewing data and user feedback in real time, the system can dynamically adjust the recommendation strategy to ensure that the recommended content always meets the user's latest needs. Through precise recommendations, users can quickly find high-quality video courses that suit them, improve the utilization rate of educational resources, and reduce resource waste. Based on artificial intelligence and natural language processing technology, the system can efficiently process massive data and is suitable for large-scale user groups and diverse educational resources. In summary, the present invention realizes efficient and accurate personalized recommendation of educational resources through multi-dimensional data analysis and artificial intelligence technology, significantly improving user learning effects and satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The present invention will be further described below with reference to the accompanying drawings.

[0033] Figure 1 This is a flow chart of a personalized recommendation method for educational resources based on artificial intelligence in the present invention. DETAILED DESCRIPTION

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0035] See also Figure 1 As shown, the present invention is a personalized recommendation method for educational resources based on artificial intelligence, comprising the following steps:

[0036] Step S1: Obtain the user's knowledge point keywords and obtain the titles of all video courses, filter out the titles containing the knowledge point keywords from all titles, record them as matching titles, and obtain the video courses corresponding to the matching titles, record them as matching video courses;

[0037] It is understandable that based on natural language processing (NLP) technology, the knowledge point keywords provided by users (such as "machine learning" and "network resources") are matched with the video course titles, the title text is broken down into independent words (such as "network resources" is split into "network" and "resources"), and the semantic similarity between them and the keywords is calculated to ensure that the recommended content is directly related to the user's knowledge point needs and avoid recommending irrelevant courses;

[0038] For example, a user searches for “online resources” and only filters video courses whose titles contain the keyword or related semantics (e.g., “online collaboration tool”, “education cloud platform”);

[0039] In a preferred embodiment of the present invention, the process of screening out the matching titles includes:

[0040] Obtaining the text of the knowledge point keyword, recorded as the keyword text, and obtaining the text of the title, recorded as the title text; dividing the title text into a plurality of words based on natural language processing technology, and sequentially obtaining the semantic similarity between the keyword text and each word; if the semantic similarity between any word and the keyword text exceeds a preset similarity threshold, the title is a matching title;

[0041] It should be noted that based on the principle of text matching, by searching for content that matches the user's knowledge point keywords in the specific text field of the video course title, each title is compared with the keywords to determine whether the title contains the keywords. Different users have different knowledge backgrounds and learning needs. By obtaining the user's knowledge point keywords, we can understand the specific content the user wants to learn. This method of filtering video courses based on user keywords provides users with personalized learning services. Each user can receive customized learning content recommendations based on their needs and interests.

[0042] For example, if a user enters the keyword "deep learning algorithm in artificial intelligence," relevant video courses with titles containing this keyword, such as "Principles and Applications of Deep Learning Algorithms," will be filtered out. This way, users can more quickly find learning resources that match their needs and avoid wasting time on a large number of irrelevant video courses.

[0043] Step S2: Obtain historical viewing data for the matching video course, including all historical users who have viewed the matching video course, as well as viewing duration and viewing speed data for each historical user, where the viewing speed data represents the real-time speed at which the historical user viewed the matching video course; obtain the total number of views and all comments for the matching video course, filter out positive comments from all comments, and obtain a favorable review rate; and obtain a viewing completion rate for the matching video course based on the historical viewing data;

[0044] Based on the viewing speed data, a viewing speed curve of historical users is obtained; based on the viewing speed curves of all historical users, an average speed curve of the corresponding video course is obtained; based on the average speed curve, a key point ratio value of the corresponding video course is obtained;

[0045] It should be noted that the total number of views reflects the popularity and user base of the video course, and analyzes the user's interest in different content segments through the real-time speed (such as 1x speed, 1.5x speed) when the user watches;

[0046] In a preferred embodiment of the present invention, the process of screening out the positive reviews includes:

[0047] Setting a sentiment tendency, wherein the sentiment tendency includes a positive tendency and a negative tendency; obtaining a number of comments as samples, recorded as comment samples, and labeling the sentiment tendency of each comment sample to obtain sample data; establishing a learning model based on natural language processing technology, inputting the sample data into the learning model, training the learning model, and obtaining a sentiment tendency analysis model; analyzing the sentiment tendency of the comment using the sentiment tendency analysis model; if the sentiment tendency of the comment is positive, the comment is a positive comment; if the sentiment tendency of the comment is negative, the comment is a negative comment;

[0048] In a preferred embodiment of the present invention, the process of obtaining the favorable review rate includes:

[0049] The positive review rate Ar = Pr / Tn, where Pr is the total number of positive reviews and Tn is the total number of views;

[0050] It is understandable that a natural language processing (NLP) model (such as a sentiment classifier) is used to analyze the sentiment tendency of the reviews, filter out the positive reviews, and calculate the positive review rate (number of positive reviews / total number of reviews);

[0051] In a preferred embodiment of the present invention, the process of obtaining the viewing completion rate includes:

[0052] Get the viewing time of the historical user and the total duration of the matching video course, and get the viewing rate of the historical user Vr=t / T, where t represents the viewing time of the historical user and T represents the total duration of the matching video course; set the viewing rate threshold Vr th , and Vr th ≥50%, record the historical users whose viewing rate exceeds the viewing rate threshold as complete browsing users; obtain the total number num of the complete browsing users, and obtain the viewing completion rate Vor = num / Tn;

[0053] It is understandable that the percentage of users who have watched the video in its entirety is counted, and the completion rate reflects the attractiveness of the course and the user's engagement;

[0054] In a preferred embodiment of the present invention, the process of obtaining the viewing speed curve of the historical user includes:

[0055] Obtain the viewing time of the historical user for the matching video course, and obtain the viewing time period of the historical user for the matching video course based on the viewing time period; select a number of time nodes equally divided in the viewing time period, and the real-time speed is the speed of the user at each time node in the viewing time period, and the speed includes 1x speed, 1.5x speed, 2x speed and 0.5x speed;

[0056] Numbering each time node, establishing a coordinate system with the number as the horizontal coordinate and the speed as the vertical coordinate; converting each numbered time node and its corresponding speed into a coordinate point at a corresponding position on the coordinate system, connecting each coordinate point with a smooth curve to obtain a viewing speed curve;

[0057] In a preferred embodiment of the present invention, the process of obtaining the average speed curve includes:

[0058] Numbering each coordinate point on the viewing speed curve according to the number of the time node; obtaining the average value of the coordinate points with the same number on the viewing speed curve of each historical user to obtain a new coordinate point with each number, wherein the average value of the coordinate point is the average value of the speed corresponding to the coordinate point; regenerating a new curve based on the new coordinate point with each number, and recording it as the average speed curve;

[0059] It is understandable that the viewing speed curve records the speed selection of each historical user at different time points during the video playback process, and the average speed curve is the average of the speed curves of all users to generate the speed change trend of the entire course;

[0060] In a preferred embodiment of the present invention, the process of obtaining the key point ratio of the corresponding video course includes:

[0061] Set the dividing line sp = 1, where sp is the speed, place the dividing line and the average speed curve of the corresponding video course in the same coordinate system, intercept the curve segment below the dividing line on the average speed curve, record it as the slow segment, and obtain the total curve length L of the average speed curve all and the length L of the slow segment slow , get the key proportion value Kpr=L slow / L all ;

[0062] It is understood that the slow-motion segments are historical periods when users slowed down the video, corresponding to the key or difficult content within the video course. Based on the average speed curve, segments with a speed of ≤ 1 (users are more focused) are intercepted and their proportion of the total video length is calculated (slow-motion segment length / total length). This helps identify redundant or inefficient content that needs improvement (parts that users speed up to watch). The proportion of core course content is quantified to provide a "content value" indicator for the recommendation method.

[0063] Step S3: Obtaining a recommendation value for the matching video course based on the praise rate, viewing completion rate, and key point ratio of the matching video course; and recommending each matching video course to the user based on the recommendation value of each matching video course;

[0064] As a preferred embodiment of the present invention, the recommended value of the corresponding video course Where K is the preset correction coefficient, and K>1, Ar ave Represents the average of the praise rates of all matching video courses, Vr ave Indicates the average viewing completion rate of all matching video courses, Kpr ave Indicates the average value of the key points of all matching video lessons;

[0065] It is understandable that the quality of a video course is evaluated by considering three key indicators: the praise rate, viewing completion rate, and key percentage of the corresponding video course; the praise rate reflects other users' evaluation of the video course, the viewing completion rate reflects the video course's appeal to users and the effectiveness of the content, and the key percentage measures the extent to which the video course contains the knowledge points required by users. By comparing these indicators with the corresponding average values and combining them with the preset correction coefficient K, the recommendation value R of each corresponding video course is calculated.

[0066] In a preferred embodiment of the present invention, the process of recommending each matching video course to the user according to the recommendation value of each matching video course includes sorting each matching video course from largest to smallest according to the recommendation value of each matching video course, and displaying the sorted matching video courses to the user;

[0067] It can be understood that according to the calculated recommendation value R, the matching video courses are sorted from large to small; the higher the recommendation value, the better the performance of the video course in multiple key indicators and the more it meets the needs of users. Finally, the sorted matching video courses are displayed to users, allowing users to quickly see the best learning resources.

[0068] An AI-based personalized recommendation system for educational resources, including:

[0069] Retrieval module: obtains the user's knowledge point keywords and the titles of all video courses, filters out titles containing the knowledge point keywords from all titles, records them as matching titles, and obtains the video courses corresponding to the matching titles, records them as matching video courses;

[0070] Analysis module: Obtain historical viewing data of the matching video course, obtain the total number of views and all comments of the matching video course, filter out positive comments from all comments, and obtain the praise rate; obtain the viewing completion rate of the matching video course based on the historical viewing data, and obtain the viewing speed curve of historical users; obtain the key percentage of the matching video course based on the viewing speed curve;

[0071] Recommendation module: obtain the recommendation value of the matching video course according to the praise rate, viewing completion rate and key point ratio of the matching video course; recommend each matching video course to the user according to the recommendation value of each matching video course.

[0072] The above is a detailed description of an embodiment of the present invention. However, the content described is only a preferred embodiment of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A personalized recommendation method for educational resources based on artificial intelligence, characterized in that: The following steps are involved: Step S1: Obtain the user's knowledge point keywords and obtain the titles of all video courses, filter out the titles containing the knowledge point keywords from all titles, record them as matching titles, and obtain the video courses corresponding to the matching titles, record them as matching video courses; Step S2: Obtain historical viewing data of the matching video course, and obtain the total number of views and all comments of the matching video course, filter out positive comments from all comments, and obtain the favorable comment rate; obtain the viewing completion rate of the matching video course based on the historical viewing data, and obtain the viewing speed curve of historical users; and obtain the key point ratio of the matching video course based on the viewing speed curve; Step S3: Obtaining a recommendation value for the corresponding video course based on the praise rate, viewing completion rate, and key point ratio of the corresponding video course; Recommend each matching video course to the user based on its recommendation value.

2. The method for personalized recommendation of educational resources based on artificial intelligence according to claim 1, characterized in that: In step S1, the process of screening out the matching titles includes: Obtain the text of the knowledge point keyword and record it as the keyword text, and obtain the text of the title and record it as the title text; divide the title text into several words based on natural language processing technology, and obtain the semantic similarity between the keyword text and each word in turn; if there is any word whose semantic similarity with the keyword text exceeds a preset similarity threshold, then the title is a matching title.

3. The method for personalized recommendation of educational resources based on artificial intelligence according to claim 1, characterized in that: In step S2, the historical viewing data includes all historical users who watched the matching video course, as well as the viewing time and viewing speed data of each historical user. The viewing speed data is the real-time speed at which the historical users watched the matching video course.

4. The method for personalized recommendation of educational resources based on artificial intelligence according to claim 1, characterized in that: In step S2, the process of screening out the positive reviews includes: Setting a sentiment tendency, which includes a positive tendency and a negative tendency; obtaining a number of comments as samples, recorded as comment samples, and marking the sentiment tendency of each comment sample to obtain sample data; establishing a learning model based on natural language processing technology, inputting the sample data into the learning model, training the learning model, and obtaining a sentiment tendency analysis model; analyzing the sentiment tendency of the comment through the sentiment tendency analysis model, if the sentiment tendency of the comment is a positive tendency, the comment is a positive comment; if the sentiment tendency of the comment is a negative tendency, the comment is a negative comment.

5. The method for personalized recommendation of educational resources based on artificial intelligence according to claim 3, characterized in that: In step S2, the process of obtaining the favorable comment rate and the complete viewing rate includes: The positive review rate Ar = Pr / Tn, where Pr is the total number of positive reviews and Tn is the total number of views; obtain the viewing time of the historical user and the total duration of the matching video course to obtain the viewing rate of the historical user Vr = t / T, where t represents the viewing time of the historical user and T represents the total duration of the matching video course; set the viewing rate threshold Vr th , and Vr th ≥50%, record the historical users whose viewing rate exceeds the viewing rate threshold as complete browsing users; obtain the total number num of the complete browsing users, and obtain the viewing completion rate Vor=num / Tn.

6. The method for personalized recommendation of educational resources based on artificial intelligence according to claim 3, characterized in that: In step S2, the process of obtaining the viewing speed curve of the historical user includes: Obtain the viewing time of the historical user for the matching video course, and obtain the viewing time period of the historical user for the matching video course based on the viewing time period; select a number of time nodes equally divided in the viewing time period, and the real-time speed is the speed of the user at each time node in the viewing time period, and the speed includes 1x speed, 1.5x speed, 2x speed and 0.5x speed; Each time node is numbered, and a coordinate system is established with the number as the horizontal coordinate and the speed as the vertical coordinate; each numbered time node and its corresponding speed are converted into a coordinate point of a corresponding position on the coordinate system, and each coordinate point is connected with a smooth curve to obtain a viewing speed curve.

7. The method for personalized recommendation of educational resources based on artificial intelligence according to claim 6, characterized in that: In step S2, the process of obtaining the key points percentage of the corresponding video course includes: Numbering each coordinate point on the viewing speed curve according to the number of the time node; obtaining the average value of the coordinate points with the same number on the viewing speed curve of each historical user to obtain a new coordinate point with each number, wherein the average value of the coordinate point is the average value of the speed corresponding to the coordinate point; regenerating a new curve based on the new coordinate point with each number, and recording it as the average speed curve; Set the dividing line sp = 1, where sp is the speed, place the dividing line and the average speed curve of the corresponding video course in the same coordinate system, intercept the curve segment below the dividing line on the average speed curve, record it as the slow segment, and obtain the total curve length L of the average speed curve all and the length L of the slow segment slow , get the key proportion value Kpr=L slow / L all .

8. The method for personalized recommendation of educational resources based on artificial intelligence according to claim 7, characterized in that: In step S3, the recommendation value of the corresponding video course Where K is the preset correction coefficient, and K>1, Ar ave It represents the average of the praise rate of all matching video courses, Vr ave Indicates the average viewing completion rate of all matching video courses, Kpr ave Represents the average value of the key points ratio of all matching video courses.

9. An artificial intelligence-based personalized recommendation system for educational resources, characterized by: include: Retrieval module: obtains the user's knowledge point keywords and the titles of all video courses, filters out titles containing the knowledge point keywords from all titles, records them as matching titles, and obtains the video courses corresponding to the matching titles, records them as matching video courses; Analysis module: Obtain historical viewing data of the matching video course, obtain the total number of views and all comments of the matching video course, filter out positive comments from all comments, and obtain the praise rate; obtain the viewing completion rate of the matching video course based on the historical viewing data, and obtain the viewing speed curve of historical users; obtain the key percentage of the matching video course based on the viewing speed curve; Recommendation module: obtain the recommendation value of the matching video course according to the praise rate, viewing completion rate and key point ratio of the matching video course; Recommend each matching video course to the user based on its recommendation value.

Images