Teacher-side online teaching resource recommendation method

Abstract

The application relates to a teacher-side online teaching resource recommendation method, which comprises the following steps: extracting key information from videos, courses and teacher information, and storing the information into a retrieval engine; obtaining information of a current teacher or course, and retrieving whether similar teachers or courses exist; if yes, obtaining reference videos of the similar teachers or courses, analyzing and summarizing the reference videos into a general video reference chain, and recommending several related videos according to which section of the reference chain the current teacher is in based on the current behavior of the teacher. The teacher-side online teaching resource recommendation method can recommend multiple courses or videos of teachers with the highest correlation degree with the course of the current teacher according to the searched teacher or video information when the information of the teacher and the current course is obtained, the method can quickly recommend the required videos to the teacher in a complicated video library, is more convenient to use, and saves the preparation time of the teacher.

Description

Technical Field

[0001] This application relates to the field of online education technology, and in particular to a method for recommending online teaching resources on the teacher's side. Background Technology

[0002] In higher education settings, teachers not only have teaching duties but also a significant amount of research work. Therefore, how to improve teaching efficiency while maintaining teaching quality is a classic challenge in higher education. While instructional videos can save teachers a considerable amount of teaching time, producing them requires extensive preparation. Consequently, many teachers choose to reuse publicly available, high-quality instructional videos in their own courses.

[0003] On some educational video platforms, the sheer volume of videos makes the process of searching, reviewing, and selecting them time-consuming for teachers. Therefore, finding the specific educational videos needed from a vast pool of resources has become one of the bottlenecks in reducing teachers' lesson preparation costs.

[0004] Currently, video recommendation in online education typically targets learners, recommending videos based on their learning interests. However, users' learning interests are generally divergent and unstructured, while teachers' teaching objectives and systems are clear, specific, and their processes are usually rigorous and orderly. For example, it's difficult to get students to learn differentials without prior knowledge of derivatives. Therefore, existing recommendation tasks and solutions are not suitable for video recommendations in teachers' lesson preparation scenarios. Summary of the Invention

[0005] Therefore, it is necessary to provide a method for recommending online teaching resources for teachers that can quickly recommend the videos they need from a complex video library, addressing the aforementioned technical issues.

[0006] A method for recommending online teaching resources on the teacher's side, the method comprising:

[0007] Extract key information from videos, courses, and teacher information, and store it in the search engine;

[0008] Retrieve information about the current teacher or course, and search for similar teachers or courses. If so, then...

[0009] The system retrieves referenced videos from similar teachers or courses, analyzes and summarizes them into a video reference chain, and determines which section of the reference chain the teacher is currently in based on their current behavior. Multiple related videos are then recommended based on that section. If not, then...

[0010] The system obtains information about the teacher's current behavior, extracts key knowledge points, and recommends multiple related videos based on this information.

[0011] The recommended videos are sorted by score, and the final predetermined number of videos are selected.

[0012] In one embodiment, the extraction of key information from videos, courses, and teacher information includes:

[0013] Vectorization and key information extraction are performed on the structured and unstructured information of the video.

[0014] Extract course names, schools offering the courses, uploaded materials, referenced videos, and historical search information;

[0015] Extract teachers' historical course offering information, uploaded data, and historical video viewing and citation information.

[0016] In one embodiment, the vectorization processing of structured and unstructured information of the video and the extraction of key information include:

[0017] Select some fields from the structured information of the video as key video information;

[0018] Perform OCR processing on the video to remove duplicate and contained frames, and assign duration to each frame according to a forward rule.

[0019] In one embodiment, the step of selecting a subset of fields from the video structured information as key video information further includes:

[0020] The selected key information is embedded into the video vector representation and stored in the retrieval engine.

[0021] In one embodiment, the OCR processing of the video, which removes duplicate frames and assigns duration to each frame according to a forward-looking rule, further includes:

[0022] Keyword extraction is performed on the deduplicated text, and the weight of the keyword is determined based on the proportion of the video duration to the duration of the frame containing the keyword, and then embedded into the video vector representation.

[0023] In one embodiment, the vectorization processing of structured and unstructured information of the video and the extraction of key information further include:

[0024] The video is processed using ASR (Automatic Representation), keywords are extracted from the text, and the weight of each keyword is determined based on its frequency of occurrence. These keywords are then embedded into the video vector representation.

[0025] In one embodiment, obtaining information about the current teacher or course and searching for the existence of similar teachers or courses includes:

[0026] Assign values ​​to the key information points of the teacher or course;

[0027] Based on the teacher's video behavior and key knowledge points, determine if similar teachers or courses exist. If so, then...

[0028] Find similar teachers or courses and summarize the video reference chain.

[0029] In one embodiment, the recommendation of multiple related videos based on videos of similar teachers or courses includes:

[0030] Determine which section of the citation chain the current teacher's or course's video is in;

[0031] Based on the node in the reference chain where the current teacher or course's video is located, recommend the previous and next two videos.

[0032] In one embodiment, the step of retrieving videos from the video library based on the key information and recommending multiple videos with the highest relevance includes:

[0033] Video retrieval based on semantics and keywords;

[0034] The three videos with the highest scores are selected based on the similarities in key information between the retrieved videos and the current teacher's videos.

[0035] In one embodiment, the method further includes:

[0036] Deduplicat videos from different recommended paths;

[0037] Different weights are assigned to different paths to obtain a weighted score for a video across all paths, and then the three videos with the highest scores are recommended.

[0038] The aforementioned method for recommending online teaching resources on the teacher's side extracts key information about videos, courses, and teachers and stores it in a retrieval engine. When teacher and current course information is obtained, the retrieval engine searches and recommends courses or videos from multiple teachers with the highest relevance to the current teacher's course based on the searched teacher or video information. This method can quickly recommend the videos that teachers need from a complex video library, making it more convenient to use and saving teachers' lesson preparation time. Attached Figure Description

[0039] Figure 1 This is a step diagram of a teacher-side online teaching resource recommendation method according to an embodiment of this application;

[0040] Figure 2 This is a step diagram of a teacher-side online teaching resource recommendation method according to another embodiment of this application;

[0041] Figure 3This is a schematic diagram of the architecture of an online teaching resource recommendation method for teachers according to one embodiment of this application. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0043] It should be noted that when a component is referred to as being "fixed to" or "set on" another component, it can be directly on the other component or there may be an intermediate component. When a component is considered to be "connected to" another component, it can be directly connected to the other component or there may be an intermediate component present. The terms "vertical," "horizontal," "upper," "lower," "left," "right," and similar expressions used in this application's specification are for illustrative purposes only and do not represent the only possible implementation.

[0044] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0045] In this application, unless otherwise expressly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature and the second feature are in indirect contact through an intermediate medium. Furthermore, "above," "over," and "on top" the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply indicates that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply indicates that the first feature is at a lower horizontal level than the second feature.

[0046] Unless otherwise defined, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The term "and / or" as used in this application includes any and all combinations of one or more of the associated listed items.

[0047] like Figure 1 As shown, in one embodiment, a method for recommending online teaching resources on the teacher's side includes the following steps:

[0048] Step S110: Extract key information from videos, courses, and teacher information, and store it in the search engine.

[0049] Specifically, relevant information about videos, courses, and teachers is extracted, including structured and unstructured information about the videos; course names, schools offering the courses, uploaded materials, video viewing history, and historical search information; and teachers' historical course offerings, uploaded materials, and historical video viewing and citation information. This information is then stored in the search engine for easier retrieval.

[0050] Step S120: Obtain information about the current teacher or course, and search for similar teachers or courses. If so, proceed.

[0051] Specifically, it searches for teachers or courses that have similarities to the current teacher or course, and then determines whether those similarities constitute similarity.

[0052] Step S130: Obtain referenced videos of similar teachers or courses, analyze and summarize them into a video reference chain, and determine which section of the reference chain the teacher is currently in based on the teacher's current behavior, recommending multiple related videos based on that section; otherwise, proceed with...

[0053] Specifically, the system analyzes and processes video referencing behavior from similar or other teachers to obtain video referencing chains for similar teachers' courses. It also analyzes which section of this video referencing chain the current teacher's behavior falls within and recommends multiple related videos based on that section. Alternatively, it analyzes and processes video referencing behavior from similar or other courses to obtain video referencing chains for similar courses, analyzes which section of this video referencing chain the current teacher's behavior falls within, and recommends multiple related videos based on that section.

[0054] Step S140: Obtain the teacher's current behavior information, extract key knowledge points and other information from it, and recommend multiple related videos based on this information.

[0055] Specifically, analyze teachers' current lesson preparation activities (uploading materials, searching keywords) and extract key knowledge points and other information from them.

[0056] Step S150: Sort the multiple recommended videos by score and select the final predetermined number of videos.

[0057] The aforementioned method for recommending online teaching resources on the teacher's side extracts key information about videos, courses, and teachers and stores it in a retrieval engine. When teacher and current course information is obtained, the retrieval engine searches and recommends courses or videos from multiple teachers with the highest relevance to the current teacher's course based on the searched teacher or video information. This method can quickly recommend the videos that teachers need from a complex video library, making it more convenient to use and saving teachers' lesson preparation time.

[0058] like Figure 2 As shown, in one embodiment, a method for recommending online teaching resources on the teacher's side includes the following steps:

[0059] Step S210: Select some fields from the video structured information as key video information.

[0060] Specifically, the structured information of the video consists of relatively accurate fields manually filled in when the video is published to the online teaching platform. These fields include the video name and the original course information in which the video was initially published. The original course information includes information about the teacher offering the course, the school, and the chapter list of the video within the original course.

[0061] Step S220: The selected key information is embedded into the video vector representation and stored in the retrieval engine.

[0062] Specifically, after analyzing this structured information, some fields were selected as key features of the video, and after normalization and mapping operations, they were embedded into the corresponding positions in the video.

[0063] Step S230: Perform OCR (Optical Character Recognition) processing on the video, remove duplicate frames and contained frames, and assign duration to each frame according to the forward rule.

[0064] Specifically, a significant portion of video information processing involves extracting key information from the text. For example, research on OCR text revealed that if a teacher explains a slide in a teaching video, that slide will be fully displayed for approximately 5 seconds (meaning all content on the slide is displayed without any subsequent animations or added content). Therefore, we chose to extract frames from the video every 5 seconds.

[0065] In the process of obtaining keywords from keyframes, a certain PPT page may exist in multiple frames, or a certain PPT page may exist in multiple frames as it is gradually and completely displayed. Therefore, when obtaining OCR text, the text results need to be processed in a certain way.

[0066] The process involves evaluating two frames. If they are identical, the first frame is removed. If they partially overlap, the different parts are loaded into the second frame, and the first frame is removed. After obtaining the deduplicated keyframes, each keyframe is assigned a duration using a forward-looking approach, based on the keyframe interval. For example, if the first frame is removed, the duration of the second frame is the difference between the duration of the first and second frames.

[0067] Step S240: Extract keywords from the deduplicated text, determine the weight of the keywords based on the proportion of the video duration to the duration of the frames containing the keywords, and embed them into the video vector representation.

[0068] Specifically, the weight of each keyword in a keyframe within the video is determined based on the effective video duration and the duration of each keyframe. The formula for calculating the weight of each keyword in a keyframe within the video is as follows:

[0069]

[0070] The effective video duration is the time of the last keyframe minus the duration of the intro and outro.

[0071] Step S250: Perform ASR (Automatic Speech Recognition) processing on the video, extract keywords from the text, determine the weight of the keywords based on their frequency of occurrence, and embed them into the video vector representation.

[0072] Specifically, for the ASR result text, keywords are extracted directly from all extracted sentences. Weights are assigned based on information such as the frequency of the extracted keywords in the sentences.

[0073] It should be noted that after obtaining the above processed information, it is embedded into the corresponding position in the video, and the vector identifier of the video is more reasonable.

[0074] Step S260: Extract the course name, the school offering the course, the uploaded materials, the videos, and the historical search information.

[0075] Specifically, for course names, irrelevant characters are first removed to retain the key course name, such as "21-Finance-Advanced Mathematics", which originally meant the advanced mathematics course offered by the Finance Department in 2021. After processing, the pure course name "Advanced Mathematics" is obtained.

[0076] If the course already has uploaded materials, extract the text from the materials and further extract the keyword cloud information related to the course.

[0077] If a course has referenced videos, obtain the course information based on the referenced videos, cluster the relevant series of courses, sort them from largest to smallest, and assign three labels and weights to the course based on the size ratio of the top three clusters.

[0078] Step S270 involves extracting the teacher's historical course offering information, uploaded data, and historical video viewing and citation information.

[0079] Specifically, if the teacher has information on history courses, the videos cited in the history courses are deduplicated to obtain a video citation sequence.

[0080] If the teacher has uploaded materials, extract the text from the materials, and then extract a word cloud of relevant keywords for the teacher.

[0081] By collecting historical viewing information from tracking data, the weight of videos with short viewing durations and no references is appropriately reduced.

[0082] Step S280: Store the above key information into the search engine.

[0083] Specifically, the processed information is used as input to the course encoder to obtain a vector representation of the course, and the relevant information is stored in the retrieval engine.

[0084] Step S290: Assign values ​​to the key information scoring points of the teacher or course.

[0085] Specifically, scores are assigned to each key piece of information based on its relevance.

[0086] Step S2100: Based on the teacher's video behavior and key knowledge point information, determine if there are similar teachers or courses. If so, proceed.

[0087] Specifically, when a similar video shares multiple key pieces of information with the current video, the score of the similar video is the sum of the scores of these multiple key pieces of information. At the same time, it is determined whether the sum of the scores is greater than a set threshold, thereby determining whether the "similar video" is truly similar.

[0088] Step S2110: Obtain similar teachers or courses and summarize the video reference chain.

[0089] Specifically, if similar teachers and courses exist, the video sequences referenced by the teachers and courses are deduplicated and normalized to obtain video reference chains.

[0090] Step S2120: Determine which section of the reference chain the current teacher's or course's video is in.

[0091] Step S2130: Recommend the previous and next two videos based on the node in the reference chain where the current teacher's or course's video is located; otherwise, execute...

[0092] Specifically, based on the actual situation, the recommended videos can be set up according to the node in the reference chain where the current teacher or course video is located, with the number of sections before and after each recommendation being customizable.

[0093] Step S2140: Search the video based on semantics and keywords.

[0094] Step S2150: Based on the similarities in key information between the retrieved videos and the current teacher's videos, obtain the three videos with the highest scores.

[0095] Specifically, when a similar video shares multiple key pieces of information with the current video, the score of the similar video is the sum of the scores for these multiple key pieces of information. This method is used to select the three similar videos with the highest total scores.

[0096] Step S2160: Deduplicate videos from different recommended paths.

[0097] Specifically, if the videos recommended by different paths are duplicates, the duplicate videos will be removed.

[0098] Step S2170: Assign different weights to different paths to obtain a weighted score of a video across all paths, and then recommend the three videos with the highest scores.

[0099] The above-mentioned method for recommending online teaching resources on the teacher's side has the following architecture diagram: Figure 3 As shown, by extracting key information from videos, courses, and teachers and storing it in a retrieval engine, when teacher and current course information is obtained, the retrieval engine will search for cases where similar teachers exist, similar courses exist, and no similar teachers or courses exist. It will then score the retrieved courses or videos based on key information, and after deduplication, recommend the three videos with the highest scores, improving both the accuracy and coverage of the recommendations. This method can quickly recommend the videos teachers need from a complex video library, making it more convenient to use and saving teachers' lesson preparation time.

[0100] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0101] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for recommending online teaching resources on the teacher's side, characterized in that, The method includes: Extract key information from videos, courses, and teacher information, and store it in the search engine; Retrieve information about the current teacher or course, and search for similar teachers or courses. If so, then... The system retrieves referenced videos from similar teachers or courses, analyzes and summarizes them into a video reference chain, and determines which section of the reference chain the teacher is currently in based on their current behavior. Multiple related videos are then recommended based on that section. If not, then... The system obtains information about the teacher's current behavior, extracts key knowledge points from it, and recommends multiple related videos based on this information. The recommended videos are sorted by score, and the final predetermined number of videos are selected. The extraction of key information from videos, courses, and teacher information includes: Vectorization and key information extraction are performed on the structured and unstructured information of the video. Extract course names, schools offering the courses, uploaded materials, referenced videos, and historical search information; Extract teachers' historical course offering information, uploaded data, and historical video viewing and citation information; The vectorization processing of structured and unstructured information in the video and the extraction of key information include: Select some fields from the structured information of the video as key video information; Perform OCR processing on the video to remove duplicate and contained frames, and assign duration to each frame according to the forward rule; The process of selecting certain fields from the video's structured information as key video information further includes: The selected key information is embedded into the video vector representation and stored in the retrieval engine; The process of performing OCR processing on the video, removing duplicate frames, and assigning duration to each frame according to a forward-looking rule, further includes: Keyword extraction is performed on the deduplicated text, and the weight of the keyword is determined based on the proportion of the video duration to the duration of the frame containing the keyword, and then embedded into the video vector representation; The vectorization processing and key information extraction of the structured and unstructured information of the video also includes: The video is processed using ASR (Automatic Representation), keywords are extracted from the text, and the weight of each keyword is determined based on its frequency of occurrence. These keywords are then embedded into the video vector representation.

2. The method for recommending online teaching resources on the teacher's side according to claim 1, characterized in that, The process of obtaining information about the current teacher or course and searching for similar teachers or courses includes: Assign values ​​to the key information points of the teacher or course; Based on the teacher's video behavior and key knowledge points, determine if similar teachers or courses exist. If so, then... Find similar teachers or courses and summarize the video reference chain.

3. The method for recommending online teaching resources on the teacher's side according to claim 2, characterized in that, We recommend several related videos based on videos by similar teachers or courses, including: Determine which section of the citation chain the current teacher's or course's video is in; Based on the node in the reference chain where the current teacher or course's video is located, recommend the previous and next two videos.

4. The method for recommending online teaching resources on the teacher's side according to claim 1, characterized in that, Based on the teacher's current behavior information, key knowledge points are extracted to search the video library and recommend the most relevant videos, including: Video retrieval based on semantics and keywords; The three videos with the highest scores are selected based on the similarities in key information between the retrieved videos and the current teacher's videos.

5. The method for recommending online teaching resources on the teacher's side according to claim 4, characterized in that, The method further includes: Deduplicat videos from different recommended paths; Different weights are assigned to different paths to obtain a weighted score for a video across all paths, and then the three videos with the highest scores are recommended.

Images