e-Learning, e-Education, and Online Training

In the education information retrieval of accounting talent innovation and entrepreneurship training, there may be issues such as inconsistent data quality, missing data, and outdated data, leading to a decrease in the performance of education information retrieval. To this end, a deep learning based accounting talent innovation and entrepreneurship training education information retrieval method is designed. Preprocesses such as text cleaning, Chinese word segmentation, vectorization of text, noise filtering, etc. are implemented for all texts. Aiming at the characteristics of high dimensionality and high sparsity of traditional multi classification text representation and classification methods based on bag of words model features, combined with the advantages of deep learning model to effectively extract high-level features, a deep belief convolution neural network model integrating deep belief network is proposed to extract low dimensionality, dense text high-level feature vector representation and implement text classification. Design a short learning model based on Actor Critic algorithm, use TCN to model the user to obtain the user’s intention, use two fully connected feedforward neural networks to represent the strategy and value function respectively, implement relevance matching for the query and document classification results, and realize the education information retrieval of accounting talent innovation and entrepreneurship training. The test results show that the design method has higher MAP, lower NDCG and ERR, and has good educational information retrieval performance.

In order to promote the benign development of online teaching and improve the quality of online teaching evaluation, this paper proposes an online teaching effect evaluation method of Chinese painting art appreciation course in colleges and universities based on machine learning model. Based on the classroom teaching evaluation model of higher vocational colleges, this paper constructs a teaching evaluation scale, establishes an evaluation index system, and standardizes the evaluation indicators. The Naive Bayesian classification algorithm is applied to train the evaluation index, calculate the weight value of the evaluation index, and automatically give the evaluation result value according to the specific evaluation data. The results show that after the application of the proposed evaluation method, the categories corresponding to the maximum online teaching effect of the Chinese painting art appreciation course in five colleges and universities are general, good, general, poor and very good, which proves the effectiveness of the evaluation method.

In the stage of online education quality evaluation, due to multiple influencing factors and complex indicator parameters, the final evaluation results are prone to significant deviations from the actual situation. In order to improve the accuracy of online education quality evaluation, this article proposes a machine learning model based online education quality evaluation method for vocational e-commerce courses. We designed an education data structure based on BOM and conducted qualitative and quantitative analysis on the factors and parameters that affect the quality of online education. In the construction stage of the evaluation index system, comprehensive indicators are designed from four aspects: school management quality, teacher teaching process, student learning behavior, and academic quality. In the stage of education quality evaluation, the SVM algorithm in machine learning is used to optimize PSO, establish an evaluation optimization model, and train iteratively through parameter optimization to achieve network education quality evaluation. The test results show that the evaluation results of the design method on the quality of online education differ significantly from the actual situation, with a specific error range of 0.02, which improves the accuracy of the evaluation.

In order to improve the accuracy of online teaching data classification results and provide comprehensive technical guidance and help for the standardized implementation of quality education, the machine learning model was introduced to carry out the design and research of online teaching data classification methods, taking the apron control specialty of a university as an example. Collect the basic information of college students majoring in apron control, the information generated in the teaching process, and the phased achievements of professional online teaching, build the online teaching database of college students majoring in apron control, and preprocess the data according to the specifications; The machine learning model is innovatively introduced to visually process the data. The encoding tool is used to transform the data format, so as to achieve the extraction of data characteristics; Calculate the similarity of the online teaching data characteristics of the apron control specialty in colleges and universities, set the classification criteria for the online teaching data of the apron control specialty in colleges and universities, and when the data similarity exceeds the set criteria, divide the data into the same category to complete the design of the classification method. The experimental results show that the designed classification method has a good application effect, and this method can effectively improve the accuracy of the classification results.

Online teaching evaluation requires the use of technical means for data collection and analysis, but current technologies may still have some limitations, such as the inability to accurately capture students’ actual performance or evaluate their comprehensive abilities To this end, a method for evaluating the effectiveness of online English teaching based on machine learning algorithms was studied. Firstly, an evaluation index system was constructed through primary selection and principal component analysis. Then collect and process the evaluation information corresponding to the indicators for data reduction, transformation, and anomaly data detection. Utilizing genetic algorithms to optimize the weights of BP neural networks in machine learning algorithms, finding better combinations of weights, and improving the learning ability and prediction accuracy of neural networks. A method for evaluating the effectiveness of online English education based on genetic algorithm optimized BP neural network has been proposed. The research results indicate that it can effectively evaluate the effectiveness of online English teaching.

With the expansion of enterprise scale and the increase in information volume, traditional manual retrieval methods are no longer able to meet the needs of rapid and accurate retrieval of enterprise education information. To this end, research is conducted to optimize the design of enterprise education information retrieval models based on machine learning technology. Utilize web scraping techniques to gather comprehensive educational data for the company’s learning and development initiatives, and complete the pre-processing of the initial enterprise education information through word segmentation, semantic tagging, clustering and other steps. The support vector machine technology in machine learning is used to extract the characteristics of enterprise education information, and the output results of enterprise education information retrieval model are obtained through the steps of feature matching, retrieval expansion, etc. Through the model test experiment, it is concluded that the retrieval accuracy and recall rate of the design model are 99.0% and 99.6% respectively, which shows that the design model has good retrieval performance and running performance.

The current detection matrix of English students’ classroom learning status in higher vocational colleges is mostly a one-way processing form, and the detection range is small, resulting in an increase in the mean difference of unit detection. Therefore, this paper proposes a design and verification study on the detection method of English students’ classroom learning status in higher vocational colleges under the improved SSD algorithm. According to the actual detection requirements and the changes in standards, first extract the detection features of English learning status, expand the detection range by using a multi-objective approach, and design MTCNN multi-target detection matrix. Based on this, build a learning status detection model under the improved SSD algorithm, and use multi-level reduction correction to achieve status detection processing. The final test results indicate that the learning status of the selected 6 classes in the English classroom is detected, combined with an improved SSD algorithm. The final unit detection mean difference was well controlled below 1.5, and the detection accuracy for the five types of classroom behaviors remained above 90%, indicating that this learning state detection method has stronger pertinence and reliability, high detection efficiency, controllable errors, and practical application value.

In the context of mass industry and innovative education, educational institutions and educators need to focus on cultivating students’ innovation ability and creativity, and prepare students for their future employment and career development by providing courses and practical opportunities for innovative education. At the same time, enterprises and industries also need to cooperate with educational institutions to jointly promote the development of innovative education and cultivate innovative talents that meet the needs of the mass industry. In order to ensure the effectiveness of English Chinese translation teaching information resource recommendation and improve the accuracy of English Chinese translation teaching information resource recommendation, a deep learning based English Chinese translation teaching information resource recommendation method is proposed. By analyzing students’ demand for teaching information resources in English Chinese translation, convolutional neural networks are used to extract the characteristics of teaching information resources in English Chinese translation. By utilizing the self coding neural network in deep learning methods, the correlation between English Chinese translation teaching information resources is excavated, and a recommendation model for English Chinese translation teaching information resources is constructed to achieve English Chinese translation teaching information resource recommendation. The experimental results show that the method proposed in this paper has a good recommendation effect on teaching information resources for English Chinese translation, and can effectively improve the accuracy of teaching information resource recommendation for English Chinese translation.

In order to improve the performance of Kabaddi athlete posture target tracking, a method of Kabaddi athlete posture target tracking based on machine learning is proposed. The threshold change parameter is calculated by using the obtained athletes’ posture characteristic parameters, and the golden section is introduced to transform it and smooth the features of athletes’ posture are extracted. Constraint loss is added to the local global supervision module of machine learning, and the local features of athlete pose are integrated, and the local parameters of athlete pose are obtained by loss function. Taylor formula was used to calculate the athletes’ pose velocity, and Kalman filter was used to evaluate the joint motion data, and Kabaddi athlete pose model was constructed. The frame difference of the background image is calculated by normalizing the athletes’ pose image, and the athletes’ pose is automatically tracked. The experimental results show that this method can track all nodes of the athletes’ posture, and has good performance in the absolute error, detail loss and tracking lag rate of the athletes’ posture tracking, so it is helpful to improve the athletes’ technical level and improve the training effect.

With the rapid development of information technology, the field of English Chinese translation teaching has accumulated a large amount of information resources. The quantity of these teaching information resources is huge and diverse, and students and teachers face the problem of information overload, making it difficult to find resources that are suitable for their needs. Personalized recommendation technology has emerged to solve the problem of information overload, recommending resources that match users’ personal interests and needs from a vast amount of resources. In response to the problem of poor personalized recommendation effectiveness in the existing personalized recommendation methods for English Chinese translation teaching information resources, this article posits a fresh individualized suggestion method for informative resources in regards to teaching the translation of English to Chinese. This paper constructs a state perception model of information resources for E-C edge teaching based on transfer learning. Based on this, obtain student group information, English Chinese translation teaching information resource information, and resource rating information, cluster English Chinese translation teaching information resources, and construct a personalized recommendation model for English Chinese translation teaching information resources. The experimental results show that the information resource clustering effect of this method is good, the diversity of resource recommendations is better, and the F-Measure value is higher.

The current evaluation matrix of medical rehabilitation teaching quality is mostly one-way, and the scope of evaluation is limited, resulting in an increase in the average difference of evaluation. Therefore, the design and research of the evaluation method of medical rehabilitation teaching quality based on historical big data decision tree classification is proposed. According to the actual measurement and analysis, set the basic teaching quality evaluation indicators, use the multi-level form, break the limitation of the evaluation range, develop the multi-level evaluation matrix, design the decision tree classification evaluation structure, build the historical big data decision tree classification evaluation model, and use the top-level improvement analysis to achieve quality evaluation. The final test results show that the analysis of the test results has been completed: after five cycles of measurement, the quality of medical rehabilitation teaching has been evaluated for five items, namely, professional ethics, teaching ability, teaching methods, teaching arrangements, and teaching effects. The final evaluation mean difference has been well controlled below 1.5, indicating that this evaluation method is highly targeted and stable, it has practical application value.

The acquisition of behavior recognition in college English smart classroom teaching helps to achieve real-time monitoring and alerting of student behavior, providing important guarantees for improving the quality and effectiveness of college English smart classroom teaching. To this end, a method for identifying abnormal behavior in college English smart classroom teaching based on deep learning is proposed. On the basis of obtaining a large amount of data on teaching behavior in college English smart classrooms, various abnormal teaching behavior characteristics such as playing with mobile phones, watching screens, whispering, and distraction were extracted and integrated from it. The Convolutional neural network in the deep learning algorithm is used to train and recognize these features, so as to achieve the purpose of identifying abnormal teaching behaviors. The experimental results show that this method can fully utilize deep learning methods to achieve recognition of teaching abnormal behavior, with high recognition accuracy, short time consumption, and high recognition accuracy. It helps teachers discover and solve abnormal situations in teaching in a timely manner, improving teaching effectiveness and students’ learning quality.

Under the premise of protecting the privacy of teaching resources, resource sharing is an effective measure to improve the utilization rate of teaching resources. Therefore, this study proposes a sharing method of teaching resources for economics majors based on Federated learning. Firstly, collect teaching resources for economics majors and complete the integration of resources based on their classification results. Then, set the (Transaction Layer Protocol) TLP protocol as the resource sharing protocol, use the Federated learning algorithm to implement encryption processing for text/data, image/video resources, and complete the directional transmission of resources through the selected transmission channel, so as to achieve resource security sharing. The test results show that the average shared resource loss and error rate of this method are 0.27 GB and 1.37%, respectively, and the shared task execution time does not exceed 2 s, indicating that compared with traditional sharing methods, the sharing performance of this method is better.

In order to improve the accuracy of difficulty recognition in college piano teaching scores, a method based on SVM algorithm for difficulty recognition in college piano teaching scores was designed. Construct a mapping space for the difficulty level features of college piano teaching scores, and extract the difficulty level features of college piano teaching scores. Based on SVM algorithm, a difficulty recognition model for college piano teaching score is constructed, and the difficulty level of college piano teaching score is mapped to the feature space for recognition, eliminating linear errors in difficulty recognition of college piano teaching score and improving the accuracy of difficulty recognition of college piano teaching score. The experimental results show that the proposed method has higher accuracy in identifying the difficulty of college piano teaching scores, and can effectively shorten the time for identifying the difficulty of college piano teaching scores.

The online learning platform provides new opportunities for tourism majors to obtain information. However, the diversity and universality of learning resources have led to Exponential growth of data, making it difficult for students to find resources that meet their own needs. To address this issue, this study proposes a personalized recommendation method for tourism professional online education resources based on machine learning. This method combines TF-IDF weight and location information weight on the basis of TextRank algorithm to generate user interest labels and attribute labels for tourism professional online education resources, thereby establishing a user interest model and resource attribute description. By optimizing the K-means clustering algorithm using genetic algorithm, the recommended online education resources for tourism majors are divided into different resource groups. Next, calculate the distance between the user interest model and the center of each resource cluster, and select the resource cluster closest to the user interest model as the recommendation result. Finally, by calculating the similarity between resources, the resources in the resource cluster are sorted to generate a personalized recommendation list. The research results indicate that the recommendation method based on machine learning has a good recommendation effect in personalized recommendation of online education resources for tourism majors. This method effectively utilizes user interest models and resource attribute descriptions to provide personalized learning resource recommendations that meet students' needs, thereby optimizing the learning process and improving learning effectiveness.

In order to improve the classification and sharing performance of teaching resources, an automatic classification and sharing method of teaching resources in online English teaching system based on SVM is proposed. According to learners’ demands for teaching resources, the tasks and objectives of online English teaching are determined, and the content performance characteristics of teaching resources are given through the digital integration of teaching resources in the online English teaching system. Based on learners’ internal psychological activity process and cognitive rules, a feature extraction model of teaching resources is constructed to extract the features of English teaching resources. According to the optimal classification plane of support vector machine, the nonlinear classification problem of teaching resource features is transformed into a quadratic optimization problem, and the Gaussian kernel function is selected as the kernel function of support vector machine to classify the features of English teaching resources. By calculating the weighted vector of English teaching resources, English teaching resources are cleaned, and the continuous sliding window distance of English teaching resources attribute compression is given by using attribute compression. Combined with the spatial trajectory function of quantitative coding, the characteristics of English teaching resources are quantified and coded. Combined with the design of teaching resource sharing algorithm, the automatic classification and sharing of teaching resources in online English teaching system is realized. The experimental results show that the proposed method can improve the utilization rate of teaching resources, reduce the sharing delay and improve the classification and sharing performance of teaching resources, no matter with or without manual intervention. Therefore, it shows that this method can improve the classification and sharing effectiveness of English teaching resources.

In the process of data association mining using traditional methods, data mining is affected by incomplete data mining, resulting in large errors in data mining. A data association mining method based on machine learning model for vocational college students’ employment education is proposed. Analyze the internal and external factors affecting students’ employment, calculate the difference series formed by two data under a certain time series, and determine the spatial correlation of the series. The confidence is compared with the expected confidence, and the interest is obtained. According to the association rules, the machine learning model of educational data association rules is constructed. Build a data gain evaluation function, calculate the frequency of word segmentation feature set in education data, and calculate the similarity between any two education data. Through the similarity calculation results of association oriented data, the implicit low rank characteristics of the data are obtained, and the association education data clustering is realized. Preprocess the associated data, and design the process of clustering data association mining under the machine learning model. The experimental results show that the mining data type of this method is consistent with the experimental data type, and the minimum mining error is 0.05, which shows that this method can obtain good mining results.

A personalized course content delivery method based on machine learning for online teaching of English translation is studied. The difficulties and challenges faced in English education are introduced, pointing out that there are differences and diversity in levels, interests and learning styles, etc.; machine learning algorithms are used to deeply analyze data in students’ personal files, including their learning history records, course grades and personal information, etc., to discover students’ learning levels, learning interests and learning characteristics, etc. Given a training sample of course standards, an interest model, solving the similarity among users using the similarity of clouds, acquiring the nearest neighbors of target users, and generating the pushed course contents and learning materials. The experimental results show that the results of personalized course content pushing for online teaching of English translation meet the real needs and have good application effects.

In order to accurately distinguish the wrong language signals and word signals in the process of online Japanese teaching, and realize the accurate detection of oral pronunciation errors, this paper studies the detection methods of oral pronunciation errors in the process of online Japanese teaching. According to the phoneme characteristics, the key audio features are extracted, and then the quality of Japanese pronunciation is evaluated by solving the linear prediction coefficient. On this basis, the corpus and pronunciation test set are constructed, and the oral signals are combined to determine the value range of detection interpolation, and the design of oral error pronunciation detection method in the process of online Japanese teaching is completed. The experimental results show that the accuracy of error language signal and word signal is more than 90% under the above method, which is in line with the practical application requirements of accurate detection of incorrect pronunciation in spoken Japanese.

In order to solve the problem of low accuracy and detection rate of video keyframe extraction in existing methods, a compressed domain based keyframe extraction algorithm for sports teaching videos is proposed. Based on the logical structure of physical education teaching videos, convolutional autoencoders are used to extract video features; Introduce HSV space and use non-uniform quantization of HSV space to achieve shot segmentation in sports teaching videos, and restore lost key frames; On the basis of video preprocessing, key frame extraction of physical education teaching videos is achieved through compressed domain. The experimental results show that the accuracy of keyframe extraction and detection rate of this algorithm are high, indicating that this algorithm has high application value.

In order to solve the problem of limited number of online people in the teaching of new media art majors, this research focuses on the Interaction design method of multi person VR distance education, and is committed to creating a more authentic multi person VR interactive education environment. Firstly, by improving the “dominant subject” teaching structure of new media, we aim to enhance teaching effectiveness and participation. By selecting key application technologies, determine the real-time connection status of the teaching project editor, and then define the connection conditions of the education editor for teaching new media art majors. Secondly, a multi person teaching scenario was designed, and with the help of VR remote education editor components, response objects at all levels of terminals were retrieved to construct an interactive scenario model. This design can achieve a multi person VR remote education interaction method for teaching new media art majors, providing students with a more realistic and immersive teaching experience. Finally, the effectiveness of the above method was demonstrated through experiments. The experimental results show that the application of the above methods can significantly increase the number of remote online people in the teaching task of the new media art specialty, and meet the practical application needs of creating a real multi person VR interactive education environment.