A network education resource sharing system based on big data

Abstract

The invention discloses a network education resource sharing system based on big data, relates to the technical field of network education resource sharing, and solves the technical problem that there are many teaching video resources on the existing network, the quality is uneven, students need to spend a lot of time to screen out high-quality resources suitable for themselves, which increases the difficulty and burden of learning; the information acquisition module of the invention acquires the sharing information of teaching videos; based on the shared information, the actual browsing ratio and evaluation index are analyzed; based on the actual browsing ratio and evaluation index, the quality evaluation value of the teaching video is analyzed and compared with the evaluation threshold range, and the teaching video is classified; based on the category, it is stored in different storage areas; by performing quality evaluation and classification on the teaching video, the user's search time can be saved, the learning efficiency can be improved, and the user experience can be improved at the same time.

Description

Technical Field

[0001] The present invention belongs to the field of network education resource sharing, and specifically is a network education resource sharing system based on big data. Background Art

[0002] The existing education resource sharing platform is connected with the university education resource platform, the public education resource platform and the network transaction resource platform. Users can obtain the education resources they need for retrieval through the education resource sharing platform. Educational resources include teaching videos, teaching documents, etc., which facilitates user learning.

[0003] The invention patent with publication number CN116484052B provides an educational resource sharing system based on big data. The invention is provided with a data storage module, an acquisition module, a generation module and a sharing module. The acquisition module is used to determine the dialect type to which the teaching video belongs. The generation module is used to convert the voice data in the teaching video into text. In addition, the word segmentation category in the text is determined, and the first type of word segmentation is directly added to the teaching video to form subtitles. The semantic vocabulary of the second type of word segmentation is determined, and the semantic vocabulary is added as an annotation to the side of the second type of word segmentation to form subtitles in the teaching video. Through the above process, the dialect category of different teaching videos and the special vocabulary unique to the dialect category are identified, and subtitles are formed after adding annotations, thereby reducing the trouble caused to viewers by special vocabulary in different dialect categories and improving the utilization rate of teaching resources.

[0004] The prior art is to convert the speech of teaching videos into text to prevent special words therein from causing confusion to viewers due to dialect issues, but the teaching videos are not classified; because there are many teaching video resources on the existing network with varying quality, students need to spend a lot of time screening out high-quality resources suitable for them, which increases the difficulty and burden of learning; therefore, the present invention provides a network education resource sharing system based on big data to solve the above technical problems. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a network education resource sharing system based on big data, which is used to solve the technical problem that there are too many teaching video resources on the existing network, the quality is uneven, and students need to spend a lot of time to screen out high-quality resources suitable for themselves, which increases the difficulty and burden of learning.

[0006] To achieve the above-mentioned purpose, the first aspect of the present invention provides a network education resource sharing system based on big data, comprising a sharing module, and an information collection module and an information analysis module connected thereto; the information collection module is in communication connection with the information analysis module;

[0007] Information collection module: used to collect shared information of teaching videos and send the shared information to the information analysis module; the shared information includes browsing history, total browsing volume, video duration and quality characteristics; browsing history includes user ID and browsing duration; quality characteristics include positive reviews, negative reviews and forwarding volume;

[0008] Information analysis module: used to analyze the actual browsing ratio of the corresponding teaching video according to the browsing history, video duration and total browsing volume; calculate the evaluation index of the corresponding teaching video based on the actual browsing volume and quality characteristics; wherein the actual browsing ratio is the ratio of the actual browsing volume to the total browsing volume; and,

[0009] Calculate the quality evaluation value of the corresponding teaching video according to the actual browsing ratio and the evaluation index; classify and store a plurality of the teaching videos according to the quality evaluation value;

[0010] Sharing module: used to obtain matching teaching videos based on the keywords entered by the user and push the matching teaching videos to the user.

[0011] Preferably, the actual browsing ratio of the corresponding teaching video is obtained by analyzing the browsing history, video duration and total browsing volume, including:

[0012] The actual number of views of the corresponding teaching video is calculated based on the browsing history and video duration, marked as SL; the total number of views of the teaching video is extracted, marked as ZL;

[0013] The actual browsing ratio BL is calculated by the formula BL=γ×SL / ZL; wherein γ is a proportional coefficient, and γ is 1 by default.

[0014] Preferably, the actual number of views of the corresponding teaching video is calculated based on the browsing history and the video duration, including:

[0015] Extract the user ID and the browsing duration corresponding to the number of times the user ID browses the teaching video from the browsing history; extract the video clips corresponding to the number of browsing durations of the user ID; match the video clips according to the timeline of the teaching video to obtain a reference timeline;

[0016] Extracting effective browsing time from the reference time axis, and calculating the ratio of effective browsing time to video time; when the ratio exceeds a preset coefficient, marking the number of video clips as the actual browsing times of the corresponding user ID;

[0017] The actual number of views of each user ID is summed up to obtain the actual number of views of the teaching video.

[0018] The present invention further improves the accuracy of browsing count statistics by matching video clips according to the time axis. Since users may interrupt or skip multiple times when browsing videos, matching video clips according to the time axis can better reflect the actual browsing behavior of users; in addition, the present invention further obtains the actual browsing count by calculating the ratio of effective browsing time to video time, providing more accurate basic data for subsequent data analysis and mining.

[0019] Preferably, the evaluation index of the corresponding teaching video is calculated based on the actual page views and quality features, including:

[0020] The teaching videos are marked with HL, CL and FL, respectively, for the number of positive reviews, negative reviews and forwardings corresponding to the quality features;

[0021] The evaluation index PG of the teaching video is calculated by the formula PG=θ×HL / CL+η×FL / SL; wherein θ and η are correction coefficients, and θ and η are both set to 1 by default.

[0022] The positive reviews, negative reviews and forwarding data of the present invention are all quantifiable data, which can objectively reflect the user's evaluation of the teaching video; by combining the three-dimensional data of positive reviews, negative reviews and forwarding data to calculate the evaluation index PG, it can comprehensively reflect the comprehensive evaluation of the video, and as a factor in evaluating the quality of the video, it lays the foundation for the subsequent analysis of the video quality.

[0023] Preferably, the step of calculating the quality evaluation value of the corresponding teaching video according to the actual browsing ratio and the evaluation index includes:

[0024] Based on the actual browsing ratio BL and the evaluation index PG, the quality evaluation value GX is calculated by the formula GX=α×BL+β×PG; wherein α and β are weight coefficients, and both are greater than 0.

[0025] The quality evaluation value GX of the present invention takes into account two factors, namely, the actual browsing ratio BL and the evaluation index PG, and can more comprehensively reflect the quality of the teaching video; both the actual browsing ratio BL and the evaluation index PG are objective and quantifiable data, making the quality evaluation value GX more objective; it can provide an objective and comprehensive quality evaluation standard for the teaching video and provide a standard for students to judge; and it is helpful to optimize the allocation and management of online education resources.

[0026] Preferably, the classifying and storing the teaching videos according to the quality evaluation values ​​includes:

[0027] Compare the quality evaluation value GX with the evaluation threshold range, and classify the teaching video into a high-quality video, a good-quality video, or a poor-quality video;

[0028] High-quality videos are stored in the high-quality storage area, good-quality videos are stored in the good-quality storage area, and poor-quality videos are stored in the poor-quality storage area.

[0029] Preferably, the storage area of ​​the teaching video is adjusted according to the comparison result between the quality evaluation value GX of the teaching video within the observation period and the evaluation threshold range.

[0030] The present invention classifies teaching videos according to quality evaluation values ​​and stores them in different storage areas respectively, thereby optimizing storage resource allocation and improving the efficiency of data retrieval and reading; and according to the comparison result between the quality evaluation value GX within the observation period and the evaluation threshold range, the storage area of ​​the teaching video can be dynamically adjusted to adapt to the change of video quality.

[0031] Preferably, comparing the quality evaluation value GX with an evaluation threshold range includes:

[0032] When the quality evaluation value GX is lower than the evaluation threshold range, the video quality is poor and is marked as a poor-quality video; when the quality evaluation value GX is within the evaluation threshold range, the video quality is good and is marked as a good-quality video; when the quality evaluation value GX exceeds the evaluation threshold range, the video quality is excellent and is marked as a high-quality video.

[0033] Preferably, obtaining a matching teaching video according to the keyword input by the user includes:

[0034] Obtain the keywords input by the user, and mark the number of the input keywords as D; match the keywords with the video theme features of the teaching video, count the number of successfully matched keywords, and mark them as R; wherein the video theme features include titles, tags or subtitles;

[0035] The matching index K is calculated by the formula K=δ×R / D to determine whether the matching index K is greater than the matching index threshold; if yes, the teaching video is a matching teaching video; if no, the teaching video is not a matching teaching video; wherein the matching teaching videos are arranged from large to small according to the quality evaluation value GX; wherein δ is a proportional coefficient, and δ defaults to 1.

[0036] The present invention obtains matching teaching videos by analyzing the matching index between keywords and teaching video titles. The method provides users with relevant video recommendations quickly and accurately. The searched matching teaching videos are arranged from large to small according to the quality evaluation value GX, so that users can see higher quality videos first, thereby improving user experience.

[0037] Preferably, the keywords may also be obtained through a speech recognition model, including:

[0038] Get the user's voice data;

[0039] The speech data is recognized and analyzed by a speech recognition model to extract keywords; wherein the speech recognition model includes a LAS model or a CTC model.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] 1. The present invention obtains the actual browsing ratio and evaluation index of the teaching video by analyzing the shared information of the teaching video; comprehensively evaluates the teaching video by combining these two factors, and the evaluation result is more representative; and based on the evaluation result, the teaching video is stored in the corresponding storage area; it is helpful to reasonably divide and organize a large number of online teaching videos and quickly screen out high-quality teaching videos for users.

[0042] 2. The present invention collects shared information of teaching videos, analyzes actual browsing ratios and evaluation indexes; based on the actual browsing ratios and evaluation indexes, analyzes and obtains quality evaluation values ​​of teaching videos, and compares them with evaluation threshold ranges to classify teaching videos; based on categories, they are stored in different storage areas; by performing quality evaluation and classification on teaching videos, users can quickly find high-quality video resources and improve learning efficiency; at the same time, this also helps high-quality videos to be more widely disseminated and utilized. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 is a system structure block diagram of the present invention;

[0045] Figure 2 is a flowchart of the method of the present invention;

[0046] Figure 3 It is a flowchart of the quality assessment method of the teaching video of the present invention. DETAILED DESCRIPTION

[0047] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] See also Figure 1-Figure 3, the first aspect of the present invention provides a network education resource sharing system based on big data, including a sharing module, and an information collection module and an information analysis module connected thereto; the information collection module is in communication connection with the information analysis module;

[0049] The information collection module is used to collect the shared information of the teaching video and send the shared information to the information analysis module; the shared information includes browsing history, total browsing volume, video duration and quality characteristics; browsing history includes user ID and browsing duration; quality characteristics include the number of positive reviews, negative reviews and forwarding volume;

[0050] The information analysis module is used to analyze the actual browsing ratio of the corresponding teaching video based on the browsing history, video duration and total browsing volume. The analysis process is as follows:

[0051] Extract the user ID and the browsing duration corresponding to the number of times the user ID browses the teaching video from the browsing history; extract the video clips corresponding to the number of browsing durations of the user ID; match the video clips according to the timeline of the teaching video to obtain a reference timeline;

[0052] Extracting effective browsing time from the reference time axis, and calculating the ratio of effective browsing time to video time; when the ratio exceeds a preset coefficient, marking the number of video clips as the actual browsing times of the corresponding user ID;

[0053] The actual number of views of each user ID is summed up to obtain the actual number of views of the teaching video, which is marked as SL.

[0054] The actual number of views of the corresponding teaching video is calculated based on the browsing history and video duration, marked as SL; the total number of views of the teaching video is extracted, marked as ZL;

[0055] The actual browsing ratio BL is calculated by the formula BL=γ×SL / ZL; wherein γ is a proportional coefficient, and γ is 1 by default.

[0056] The actual browsing ratio in the present invention is used to count the ratio of the actual browsing volume of the teaching video to the total browsing volume. If the actual browsing ratio is high, it means that the teaching video has been browsed a high number of times, that is, many users are browsing the teaching video, indicating that the teaching video has a high number of browsings.

[0057] For example, suppose an educational resource sharing system contains relevant teaching videos of various subjects in primary school, junior high school, high school and university. This application will analyze the quality evaluation values ​​of three mathematics videos in junior high school and mark them as teaching video A, teaching video B and teaching video C respectively.

[0058] Take teaching video A as an example for analysis. Assume that the duration of teaching video A is 10 minutes. Extract the video clips corresponding to the browsing duration of each user ID browsing teaching video A from the browsing records.

[0059] Case 1: If a user ID browses teaching video A 3 times, the browsing time and the corresponding video clips are: the first browsing time is 3 minutes, and the corresponding video clips are 0-3 minutes; the second browsing time is 5 minutes, and the corresponding video clips are 2-7 minutes; the third browsing time is 5 minutes, and the corresponding video clips are 0-5 minutes;

[0060] Match the video clips obtained above according to the timeline of the teaching video, and the reference timeline obtained is 0-7 minutes, that is, the effective browsing time of the user ID is 7 minutes, and the ratio of the effective browsing time to the video time is 7 / 10=0.7;

[0061] If the preset coefficient is 0.2, the number of views for the user ID is the actual number of views;

[0062] If the preset coefficient is 0.8, the number of views of the user ID is not the actual number of views.

[0063] The second case: If a user ID browses once and the browsing time is 5 minutes, there is no need to analyze the reference time axis. The browsing time is the effective time, and the ratio of the browsing time to the video time is 0.5;

[0064] If the preset coefficient is 0.2, the number of views for the user ID is the actual number of views;

[0065] If the preset coefficient is 0.8, the number of views of the user ID is not the actual number of views.

[0066] According to the above method, each user ID is analyzed, and the actual browsing times of each user ID are counted to obtain the actual browsing volume of teaching video A. Assuming that the actual browsing volume obtained by analysis is SL=300, if the total browsing volume of teaching video A ZL=600, the actual browsing ratio BL=1×300 / 600=0.5;

[0067] The evaluation index of the corresponding teaching video is calculated based on the actual page views and quality characteristics:

[0068] The teaching videos are marked with HL, CL and FL, respectively, for the number of positive reviews, negative reviews and forwardings corresponding to the quality features;

[0069] The evaluation index PG of the teaching video is calculated by the formula PG=θ×HL / CL+η×FL / SL; wherein θ and η are correction coefficients, and θ and η are both set to 1 by default.

[0070] The evaluation index in the present invention is used to evaluate the quality of teaching videos, and the quality of teaching videos is closely related to the number of positive reviews, negative reviews, and forwarding. If the number of positive reviews, negative reviews, and forwarding are all relatively high, it means that the quality of the teaching video is good, and many users have learned from the teaching video and given it a high evaluation, so the evaluation index of the teaching video is high.

[0071] For example: Taking teaching video A as an example for analysis, the actual number of views SL = 300, assuming that the number of positive reviews HL = 200, the number of negative reviews CL = 80 and the number of reposts FL = 150; the evaluation index PG = 1×200 / 80+1×150 / 300 = 3.

[0072] Based on the actual browsing ratio BL and the evaluation index PG, the quality evaluation value GX is calculated by the formula GX=α×BL+β×PG; wherein α and β are weight coefficients, and both are greater than 0.

[0073] Compare the quality evaluation value GX with the evaluation threshold range, and classify the teaching video into a high-quality video, a good-quality video, or a poor-quality video;

[0074] Among them, when the quality evaluation value GX is lower than the evaluation threshold range, the video quality is poor and is marked as a poor-quality video; when the quality evaluation value GX is within the evaluation threshold range, the video quality is good and is marked as a good-quality video; when the quality evaluation value GX exceeds the evaluation threshold range, the video quality is excellent and is marked as a high-quality video.

[0075] The quality evaluation value in the present invention is used to evaluate the comprehensive quality of the teaching video, which is closely related to the actual browsing ratio and the evaluation index. If the actual browsing ratio and the evaluation index are both relatively high, it means that the comprehensive quality of the teaching video is good, and it is learned by many users and is worth promoting.

[0076] For example: Assume α = 0.4, β = 0.6; the quality evaluation value of teaching video A GX = α × BL + β × PG = 0.4 × 0.5 + 0.6 × 3 = 2;

[0077] Assuming that the evaluation threshold range of the quality evaluation value GX is [3,8], it can be seen that the mathematical video A is a poor quality video;

[0078] Teaching Video B and Teaching Video C are analyzed according to the analysis method of Teaching Video A. Assume that the analysis results show that GX=5 for Teaching Video B, which is a good quality video; and GX=11 for Teaching Video C, which is a high quality video.

[0079] High-quality videos are stored in the high-quality storage area, good-quality videos are stored in the good-quality storage area, and poor-quality videos are stored in the poor-quality storage area.

[0080] It should be noted that, according to the comparison result between the quality evaluation value GX of the teaching video in the observation period and the evaluation threshold range, the storage area of ​​the teaching video can be dynamically adjusted to adapt to the change of the video quality. In general, the stored teaching videos can be dynamically evaluated at regular intervals; if the quality of the teaching video changes, its storage area is adjusted.

[0081] For example: after the teaching videos are classified and stored, the quality evaluation values ​​within the observation period are analyzed; assuming that the observation period is 1 month, the actual browsing ratio of teaching video A within this 1 month is BL=0.8, and the evaluation index PG=6, so the quality evaluation value GX=0.4×0.8+0.6×6=3.92; then the teaching video A changes from a poor quality video to a good quality video, so it is transferred from the poor quality storage area to the good quality storage area.

[0082] Sharing module: used to obtain matching teaching videos according to the keywords entered by the user, push the matching teaching videos to the user, and arrange the matching teaching videos from large to small according to the quality evaluation value GX, and the user can choose by himself;

[0083] Specifically, the keywords input by the user are obtained, and the number of the input keywords is marked as D; the keywords are matched with the video theme features of the teaching video, and the number of successfully matched keywords is counted and marked as R; wherein the video theme features include titles, tags or subtitles;

[0084] The matching index K is calculated by the formula K=δ×R / D to determine whether the matching index K is greater than the matching index threshold; if yes, the teaching video is a matching teaching video; if no, the teaching video is not a matching teaching video; wherein the matching teaching videos are arranged from large to small according to the quality evaluation value GX; wherein δ is a proportional coefficient, and δ defaults to 1.

[0085] The video theme features in this application mainly include titles, tags or subtitles. The title is the title drafted by the author when uploading the corresponding teaching video. The tag is the classification tag set by the author for the teaching video when uploading, and of course it can also be the tag automatically recognized by the AI ​​software. The subtitles are the subtitles set by the author for the teaching video, and of course they can also be subtitles automatically recognized by the AI ​​software.

[0086] For example: suppose the user inputs the keywords "congruent triangles" and "proof", D = 2; extract the same keywords in the teaching video title and count the number of the same keywords. If the title of teaching video A is "Proof of congruent triangles", the same keywords are: "congruent triangles" and "proof", the number of the same keywords R = 2, and the matching index K = 2 / 2 = 1;

[0087] The title of teaching video B is "Solving Techniques of Congruent Triangles", so the same keyword is: "Congruent Triangles", the number of the same keywords R = 1, and the matching index K = 1 / 2 = 0.5;

[0088] The title of teaching video C is "quadratic function", which does not have the same keywords;

[0089] Assuming that the matching index threshold is 0.2, teaching video A and teaching video B are matching teaching videos, and are presented to the user in descending order according to the quality evaluation value GX.

[0090] Keywords can also be obtained through speech recognition models, including:

[0091] Get the user's voice data;

[0092] The speech data is recognized and analyzed by a speech recognition model to extract keywords; wherein the speech recognition model includes a LAS model or a CTC model.

[0093] The LAS model (Listen, Attend and Spell) is an end-to-end neural network model for automatic speech recognition (ASR) tasks. The model consists of three parts: "listen", "attend" and "spell". Specifically, the LAS model contains the functions of the acoustic model, the vocalization model, and the language model, that is, the functions of the previous three modules are included in one neural network.

[0094] In the LAS model, the listener is a pyramid-shaped recurrent network encoder whose input is the spectrum of the filter bank. The main function of this part is to convert the input speech signal into a high-level representation, which can remove the noise in the speech and only retain the sequence information related to speech recognition.

[0095] The speller part is an attention-based recurrent network decoder, which is responsible for converting the high-level representation output by the listener part into the final character sequence. Compared with models such as DNN-HMM and CTC, the advantage of the LAS model is that it does not make independent assumptions about the probability distribution of the output character sequence.

[0096] In general, the LAS model is an effective speech recognition model. Its end-to-end nature enables it to generate character sequences directly from speech signals without relying on multiple independent modules like traditional speech recognition systems.

[0097] The CTC (Connectionist Temporal Classification) model is a deep learning model for speech recognition. Its main goal is to directly map speech signals to corresponding text sequences to achieve classification of temporal problems.

[0098] In speech recognition, traditional models usually require frame-level annotation of speech signals, which is very difficult for large-scale data processing. The introduction of the CTC model relaxes this one-to-one correspondence restriction requirement. It only requires one input sequence (speech signal) and one output sequence (text sequence) for training. This greatly simplifies the speech recognition process and makes it possible to train on large-scale data sets.

[0099] The working principle of the CTC model can be summarized in the following steps: first, the speech signal is input into the neural network to obtain the output of each frame; then, a many-to-one function is defined to merge the repeated characters in the output sequence to form a unique sequence; finally, by optimizing the objective function (usually the cross entropy loss function), the output of the neural network is made as close to the real text sequence as possible.

[0100] The advantage of the CTC model is that it can automatically align speech and text without explicit frame-level annotation. This makes it highly practical in large-scale speech recognition tasks. In addition, with the continuous development of deep learning technology, end-to-end acoustic models based on CTC are becoming a research hotspot. They can integrate acoustic models and language models into a neural network, further improving the performance of speech recognition.

[0101] In general, the CTC model is a time series classification model based on a neural network. It can directly map speech signals to text sequences to achieve automatic alignment and end-to-end speech recognition.

[0102] The speech recognition model in the present invention can be constructed based on the LAS model or the CTC model; of course, it can also be constructed through other artificial intelligence models.

[0103] The reference for building a speech recognition model through other artificial intelligence models is as follows: The process of building a speech recognition model is to extract a number of speech data and corresponding speech texts from historical data, divide the speech data into several frames, extract the feature parameters in each frame of speech, and normalize the feature parameters; wherein the feature parameters include Mel frequency cepstral coefficients and linear predictive coding;

[0104] The processed speech data is used as standard input data, and the speech text is used as standard output data. The artificial intelligence model is trained with the standard input data and standard output data to obtain a speech recognition model.

[0105] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0106] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A network education resource sharing system based on big data, characterized in that: It includes a sharing module, and an information collection module and an information analysis module connected thereto; the information collection module is in communication connection with the information analysis module; Information collection module: used to collect shared information of teaching videos and send the shared information to the information analysis module; the shared information includes browsing history, total browsing volume, video duration and quality characteristics; browsing history includes user ID and browsing duration; quality characteristics include positive reviews, negative reviews and forwarding volume; Information analysis module: used to analyze the actual browsing ratio of the corresponding teaching video according to the browsing history, video duration and total browsing volume; the actual browsing ratio is the ratio of the actual browsing volume to the total browsing volume; Extract the user ID and the browsing duration corresponding to the number of times the user ID browses the teaching video from the browsing history; extract the video clips corresponding to the number of browsing durations of the user ID; match the video clips according to the timeline of the teaching video to obtain a reference timeline; Extracting effective browsing time from the reference time axis, and calculating the ratio of effective browsing time to video time; when the ratio exceeds a preset coefficient, marking the number of video clips as the actual browsing times of the corresponding user ID; The actual number of views of each user ID is summed up to get the actual number of views of the teaching video, which is marked as SL; By formula The actual browsing ratio BL is calculated; where ZL is the total number of views of the teaching video; is the proportionality coefficient, and The default value is 1. Calculate the evaluation index of the corresponding teaching video based on the actual page views and quality characteristics; and, Calculate the quality evaluation value of the corresponding teaching video according to the actual browsing ratio and the evaluation index; classify and store a plurality of the teaching videos according to the quality evaluation value; Sharing module: used to obtain matching teaching videos based on the keywords entered by the user and push the matching teaching videos to the user.

2. According to the big data-based online education resource sharing system of claim 1, it is characterized in that: The evaluation index of the corresponding teaching video calculated based on the actual page views and quality features includes: The teaching videos are marked with HL, CL and FL, respectively, for the number of positive reviews, negative reviews and forwardings corresponding to the quality features; By formula The evaluation index PG of the teaching video is calculated; among them, , is the correction factor, and , The default value is 1.

3. According to the big data-based online education resource sharing system of claim 2, it is characterized in that: The quality evaluation value of the corresponding teaching video is calculated according to the actual browsing ratio and the evaluation index, including: Based on the actual browsing ratio BL and the evaluation index PG, the formula The quality evaluation value GX is calculated; where: , is the weight coefficient, and its value is greater than 0.

4. According to the big data-based online education resource sharing system of claim 3, it is characterized in that: The classifying and storing the teaching videos according to the quality evaluation values ​​includes: Compare the quality evaluation value GX with the evaluation threshold range, and classify the teaching video into a high-quality video, a good-quality video, or a poor-quality video; High-quality videos are stored in the high-quality storage area, good-quality videos are stored in the good-quality storage area, and poor-quality videos are stored in the poor-quality storage area.

5. According to the big data-based online education resource sharing system of claim 4, it is characterized in that: According to the comparison result between the quality evaluation value GX of the teaching video in the observation period and the evaluation threshold range, the storage area of ​​the teaching video is adjusted.

6. The network education resource sharing system based on big data according to claim 4 is characterized in that: The comparing the quality evaluation value GX with the evaluation threshold range includes: When the quality evaluation value GX is lower than the evaluation threshold range, the video quality is poor and is marked as a poor-quality video; when the quality evaluation value GX is within the evaluation threshold range, the video quality is good and is marked as a good-quality video; when the quality evaluation value GX exceeds the evaluation threshold range, the video quality is excellent and is marked as a high-quality video.

7. The network education resource sharing system based on big data according to claim 3 is characterized in that: The step of obtaining a matching teaching video according to the keyword input by the user includes: Obtain the keywords input by the user, and mark the number of the input keywords as D; match the keywords with the video theme features of the teaching video, count the number of successfully matched keywords, and mark them as R; wherein the video theme features include titles, tags or subtitles; By formula Calculate the matching index K, and determine whether the matching index K is greater than the matching index threshold; if yes, the teaching video is a matching teaching video; if no, the teaching video is not a matching teaching video; wherein the matching teaching videos are arranged from large to small according to the quality evaluation value GX; wherein, is the proportionality coefficient, and The default is 1.

8. The big data-based online education resource sharing system according to claim 7 is characterized in that: The keywords may also be obtained through a speech recognition model, including: Get the user's voice data; The speech data is recognized and analyzed by a speech recognition model to extract keywords; wherein the speech recognition model includes a LAS model or a CTC model.

Images