Student score prediction method and device, equipment and storage medium

Abstract

The invention provides a student score prediction method and device, equipment and a storage medium, and relates to the technical field of big data information processing, historical course features and score vectors of historical course samples are obtained, and a score prediction result is obtained based on a cosine distance between the historical course features and target course features corresponding to a target course. Performing weighting processing on the score vector, determining historical score characteristics of the target object, training a preset score prediction model based on the historical score characteristics and click stream characteristics corresponding to the target course in a target time period, and generating a target score prediction model, and inputting a course to be predicted into the target score prediction model, and determining a predicted score of the target object. According to the method, the historical scores are combined with the current click stream data, the features are effectively extracted from the time sequence data, a more comprehensive view angle is provided for score prediction, and the accuracy and reliability of score prediction are improved.

Description

Technical Field

[0001] The present invention relates to the field of natural language processing technology, and in particular to a student performance prediction method, device, equipment and storage medium. Background Art

[0002] Due to the limitations of technology and tools, traditional learning behavior theory research can only explore behaviors that can be directly measured and observed. However, the study of implicit correlations in learning behavior has always lacked first-hand event data, and a systematic theoretical analysis system has not yet been formed. In recent years, with the construction of campus information systems, a large amount of student learning-related data has gradually accumulated, providing a data basis for learning behavior analysis and modeling under big data conditions, making data-driven learning behavior modeling and analysis work increasingly valued by the academic community.

[0003] Currently, student performance prediction methods mainly rely on a single type of data, such as historical performance or online learning behavior. These student performance prediction methods fail to fully utilize the diversity and complexity of students' learning processes. Although there are student performance prediction models for machine learning and deep learning, most of these models only consider one-sided features such as historical performance, which limits the generalization ability and accuracy of the prediction model, resulting in the prediction results of grades not achieving the expected results. Summary of the invention

[0004] The purpose of the present invention is to solve the problem that the prediction results of knowledge points based on current machine learning methods cannot achieve the expected effect. The present invention provides a student performance prediction method, device, equipment and storage medium.

[0005] The technical solution of the embodiment of the present application is implemented as follows: The first aspect of the embodiment of the present application provides a method for predicting student scores, comprising: Obtaining history course features and a score vector of a history course sample; the history course features are related to course content, course objectives, teaching methods, and evaluation methods; the score vector represents score information corresponding to the history course sample; Based on the cosine distance between the history course feature and the target course feature corresponding to the target course, weighting the score vector to determine the history score feature of the target object; Training a preset score prediction model based on the historical score features and the clickstream features corresponding to the target course in the target time period to generate a target score prediction model; The course to be predicted is input into the target performance prediction model to determine the predicted performance of the target object.

[0006] Optionally, the step of obtaining history course features and achievement vectors of history course samples includes: Performing word segmentation on the course description information of the history course sample, and obtaining word vectors from the words and sentences obtained by the word segmentation; Determine the TF-IDF value of each word vector by using the TF-IDF statistical method, and perform weighted summation on the word vectors based on the TF-IDF value to obtain the text features of the history course sample; The text feature is concatenated with the semester information feature of the history course sample to obtain the history course feature of the history course sample; the semester information feature represents the class schedule of the history course sample.

[0007] Optionally, the step of weighting the score vector based on the cosine distance between the history course feature and the target course feature corresponding to the target course to determine the history score feature of the target object includes: Determining the degree of association between each history course sample and the target course based on the cosine distance between the history course feature and the target course feature corresponding to the target course; Based on the degree of association, the score vectors corresponding to the various history course samples are weighted to determine the history score characteristics of the target object.

[0008] Optionally, the step of weighting the score vector based on the cosine distance between the history course feature and the target course feature corresponding to the target course to determine the history score feature of the target object includes: If the cosine distance between the first history course feature and the target course feature is greater than the cosine distance between the second history course feature and the target course feature, the weight of the score vector corresponding to the first history course is less than the weight of the score vector corresponding to the second history course.

[0009] Optionally, the training of a preset score prediction model based on the historical score features and the click stream features corresponding to the target course in the target time period to generate a target score prediction model includes: Inputting the historical score features and the clickstream features corresponding to the target course in the target time period into the preset score prediction model to obtain an output score; Determine the loss function corresponding to the preset score prediction model based on the output score and the expected score; The loss function is used in combination with a back-propagation algorithm to adjust the parameters of the preset performance prediction model until the loss function reaches a preset value, thereby obtaining a target performance prediction model.

[0010] Optionally, the training of a preset score prediction model based on the historical score features and the clickstream features corresponding to the target course in the target time period to generate a target score prediction model further includes: Performing fuzzy clustering on the history course samples, and dividing the history course samples into different categories by similarity; The preset score prediction model is trained based on the historical score features corresponding to the history course samples of different categories and the click stream features corresponding to the target course in the target time period to generate a target score prediction model.

[0011] Optionally, inputting the course to be predicted into the target grade prediction model to determine the predicted grade of the target object includes: Determining the target subject's mastery of the knowledge points of each history course sample according to the history performance characteristics; determining an adjustment parameter based on the mastery level; If the mastery level is greater than a first preset value, the sum of the output score of the target score prediction model and the adjustment parameter is determined as the predicted score; if the mastery level is less than a second preset value, the difference between the output score of the target score prediction model and the adjustment parameter is determined as the predicted score; if the mastery level is between the first preset value and the second preset value, the output score of the target score prediction model is determined as the predicted score.

[0012] The second aspect of the embodiment of the present application provides a student performance prediction device, including: an acquisition module, a weighting module, a training module and a determination module, wherein: The acquisition module is configured to acquire history course characteristics and a score vector of the history course sample; the history course characteristics are related to course content, course objectives, teaching methods and evaluation methods; the score vector represents the score information corresponding to the history course sample; The weighting module is configured to perform weighted processing on the score vector based on the cosine distance between the history course feature and the target course feature corresponding to the target course, so as to determine the history score feature of the target object; The training module is configured to train a preset score prediction model based on the historical score features and the clickstream features corresponding to the target course in the target time period to generate a target score prediction model; The determination module is configured to input the course to be predicted into the target grade prediction model to determine the predicted grade of the target object.

[0013] A third aspect of an embodiment of the present application provides an electronic device, comprising a processor and a memory; the memory stores a computer program, wherein the computer program, when executed by the processor, implements the student performance prediction method described in the first aspect.

[0014] A fourth aspect of an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0015] Compared with the prior art, the technical solution provided by this application has the following beneficial effects: The present invention provides a student performance prediction method, device, equipment and storage medium, which obtains the history course features and performance vector of the history course sample, performs weighted processing on the performance vector based on the cosine distance between the history course features and the target course features corresponding to the target course, determines the history performance features of the target object, trains a preset performance prediction model based on the history performance features and the click stream features corresponding to the target course in the target time period, generates a target performance prediction model, inputs the course to be predicted into the target performance prediction model, and determines the predicted performance of the target object. The present invention combines the history performance with the current click stream data, effectively extracts features from the time series data, and uses more sufficient data to complete the training of the performance prediction model, thereby providing a more comprehensive perspective for performance prediction, improving the generalization ability and prediction performance of the performance prediction model, and thus improving the accuracy and reliability of performance prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flow chart of a method for predicting student scores provided in an embodiment of the present application; Figure 2 A schematic diagram of the structure of a performance prediction device provided in an embodiment of the present application; Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] Below, embodiments of the present application will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present application. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present application. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present application. The terms used herein are only for describing specific embodiments and are not intended to limit the present application. The terms "include", "comprising", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components. All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification, and should not be interpreted in an idealized or overly rigid manner.

[0018] Some block diagrams and / or flow charts are shown in the accompanying drawings. It should be understood that some blocks or combinations thereof in the block diagrams and / or flow charts may be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that these instructions, when executed by the processor, may create a device for implementing the functions / operations described in these block diagrams and / or flow charts.

[0019] In some embodiments, see Figure 1 , Figure 1 A flow chart of a method for predicting student scores provided in an embodiment of the present application; the method for predicting student scores provided in an embodiment of the present application includes: S110, obtaining history course characteristics and score vectors of history course samples; the history course characteristics are related to course content, course objectives, teaching methods and evaluation methods; the score vector represents the score information corresponding to the history course sample.

[0020] In this embodiment, the course description information of the history course sample is obtained. The course description information can be formally expressed as ,in Represents the course description information of a sample history course. The course description information of the target course is represented. The corresponding course features are obtained by extracting features from the course description information. Exemplarily, the history course features may be the history course name, history course theme, history course content, history course duration, etc., which are not enumerated. The performance vector may be obtained based on the historical performance data of the target object and the corresponding time series information. Here, the historical performance data may include classroom test scores, examination scores, and other assessment scores.

[0021] In one example, the course content, course objectives, teaching methods, and evaluation methods are converted into quantifiable forms for data analysis. Specifically, the course content can be coded (such as using numbers or labels) to represent different historical periods and themes. Course objectives can be evaluated by a scale to assess students' mastery of knowledge, expressed in scores or grades. Teaching methods can be recorded by recording the teaching methods used in each course and their proportions (such as 40% for lectures, 30% for discussions, etc.). Evaluation methods include recording the weight or score of each evaluation method.

[0022] In some embodiments, S110, obtaining history course features and achievement vectors of history course samples includes: Perform word segmentation on the course description information of the history course sample, and obtain word vectors from the words and sentences obtained by word segmentation; The TF-IDF statistical method is used to determine the TF-IDF value of each word vector, and the word vectors are weighted and summed based on the TF-IDF value to obtain the text features of the history course sample; The text features are concatenated with the semester information features of the history course samples to obtain the history course features of the history course samples; the semester information features represent the class schedule of the history course samples.

[0023] TF (term frequency) refers to the frequency of a word appearing in the course description information, and IDF (inverse document frequency) is a measure of the general importance of a word. The TF-IDF value of each word vector can be used to determine the importance of the word vector, and then weighted processing can be performed. The higher the importance, the greater the weight; the lower the importance, the smaller the weight. The semester information feature represents the scheduling time of the history course sample, that is, the teaching semester in which the history course sample is scheduled. Taking the eight semesters in four years of university as an example, the semester information can be initialized as an 8-dimensional one-hot encoding , where each bit represents a specific semester. If the course is offered in that semester, the corresponding position is 1, otherwise it is 0.

[0024] In a specific example, Word2Vec is used to extract semantic features of history course samples. , then calculate the TF-IDF value of each word as the weight, and finally the word vector of all words Perform weighted summation to obtain the text features of the history course sample, and then add the semester features The historical course features are obtained by splicing. The calculation formula is as follows: ; in represents the number of words in the course description information of the i-th history course sample, Represents the word vector of the jth word in the course description information of the i-th history course sample, represents the TF-IDF value of the jth word in the course description information of the i-th history course sample, Represents a splicing operation, Represents the semester information features corresponding to the i-th history course sample.

[0025] S120, based on the cosine distance between the history course feature and the target course feature corresponding to the target course, weighted processing is performed on the score vector to determine the history score feature of the target object.

[0026] Cosine distance measures the difference in direction between two vectors, not their size. Here, the cosine similarity formula can be used to calculate the similarity between the characteristics of the history course and the characteristics of the target course. The greater the similarity, the closer the corresponding history course sample is to the content of the target course, and the greater the impact on the student's performance in the current target course. The calculation principle of cosine similarity CS (Cosine Similarity) can be expressed as: ; in, and It should be noted that Softmax can also be used to normalize the similarity results and finally convert them into a scalar of [0, 1] to more intuitively show the connection between the characteristics of the history course and the characteristics of the target course.

[0027] In some embodiments, S120, based on the cosine distance between the history course feature and the target course feature corresponding to the target course, weighting the score vector to determine the history score feature of the target object includes: Determine the degree of association between each history course sample and the target course based on the cosine distance between the history course characteristics and the target course characteristics corresponding to the target course; Based on the degree of association, the score vectors corresponding to each history course sample are weighted to determine the history score characteristics of the target object.

[0028] For example, the historical scores are initialized as a 100-dimensional one-hot vector , and transform the one-hot vector into a dense vector through a fully connected layer. This vector not only retains the original grade information, but also extracts the features in the grade data, namely the grade vector. Based on the cosine distance between each history course sample and the target course, the weighted summation with the corresponding grade vector is performed, and finally the student's history grade features can be obtained. .

[0029] ; Where n is the number of history courses, Represents the initial one-hot vector of the historical scores of the i-th subject, represents the weight matrix, represents the bias matrix.

[0030] In some embodiments, S120, based on the cosine distance between the history course feature and the target course feature corresponding to the target course, weighting the score vector to determine the history score feature of the target object includes: If the cosine distance between the first history course feature and the target course feature is greater than the cosine distance between the second history course feature and the target course feature, the weight of the score vector corresponding to the first history course is less than the weight of the score vector corresponding to the second history course.

[0031] In this embodiment, the smaller the cosine distance between the history course sample and the target course, the greater the correlation between the two, and the corresponding weight is greater. It should be noted that the similarity here is cosine similarity. Cosine similarity is the cosine value of the angle between two vectors. The range of this value is from -1 to 1. When the directions of the two vectors are exactly the same (i.e. the angle is 0 degrees), the cosine similarity is 1, indicating the maximum similarity. When the directions of the two vectors are completely opposite (i.e. the angle is 180 degrees), the cosine similarity is -1, indicating the maximum dissimilarity or difference. When the two vectors are orthogonal (i.e. the angle is 90 degrees), the cosine similarity is 0, indicating that there is no linear relationship between them. Cosine distance refers to a metric opposite to cosine similarity, i.e. 1 minus cosine similarity.

[0032] S130, training a preset score prediction model based on historical score features and clickstream features corresponding to a target course in a target time period to generate a target score prediction model.

[0033] The clickstream features corresponding to the target course in the target time period represent the click behavior of the target object when watching the corresponding video of the target course. The higher the clickstream frequency, the higher the activity of the target object.

[0034] In one example, the video clickstream sequence for each chapter of the target course Input into Transformer to convert the discrete click stream sequence into a continuous vector and obtain the embedding features of all chapters of the target course The calculation process is shown in the following formula.

[0035] ; in, represents the click stream sequence of the i-th chapter, Represents the embedding features of the click stream sequence of the i-th chapter through the Transformer.

[0036] By feeding the embedding features of all chapters into CNN, the video clickstream data can be analyzed to identify local patterns in student behavior. Through the operation of the convolutional layer, the embedding of each chapter is extracted. Then the pooling layer is used for further processing to reduce the number of parameters of the model and reduce the risk of overfitting, thereby improving the generalization ability of the model. After that, the extracted features are converted into a clickstream feature through the fully connected layer. This feature not only includes all the students' click behaviors, but also integrates the time sequence of these behaviors, providing a solid foundation for subsequent student score prediction. The calculation process can be expressed as: .

[0037] In some embodiments, S130, training a preset score prediction model based on historical score features and clickstream features corresponding to a target course in a target time period to generate a target score prediction model includes: Input the historical score features and the clickstream features corresponding to the target course in the target time period into the preset score prediction model to obtain the output score; Determine the loss function corresponding to the preset score prediction model based on the output score and the expected score; The loss function is used in combination with the back propagation algorithm to adjust the parameters of the preset performance prediction model until the loss function reaches the preset value and the target performance prediction model is obtained.

[0038] In this embodiment, the historical performance features and clickstream features can be integrated to complete the training of the preset performance prediction model. The calculation process is shown in the following formula.

[0039] ; in, represents the predicted output, represents the weight matrix, represents the bias matrix.

[0040] Here, the preset score prediction model can be adjusted by using the loss function and the back propagation algorithm through the output score, so that the error between the output score of the preset score prediction model and the expected score becomes smaller and smaller. It should be noted that the data used for model training are all past data of the target object. When the preset score prediction model is trained using the data of the history course sample, the preset score prediction model can be trained together with the click stream characteristics of the time period in which the history course sample is located, so as to further improve the preset score prediction model and thus improve the final prediction accuracy.

[0041] In an example, the loss function can be expressed as: ; in, represents the number of samples in the batch, and Represent the output score and expected score of the i-th sample respectively.

[0042] In some embodiments, S130, training a preset score prediction model based on historical score features and clickstream features corresponding to a target course in a target time period to generate a target score prediction model further includes: Perform fuzzy clustering on history course samples and divide them into different categories by similarity; Based on the historical score features corresponding to different categories of history course samples and the click stream features corresponding to the target courses in the target time period, the preset score prediction models are trained respectively to generate a target score prediction model.

[0043] In this embodiment, fuzzy clustering is performed on history course samples to classify the history course samples, and then a preset score prediction model is trained based on these categories to generate a target score prediction model, thereby ensuring the consistency of the training samples and improving the accuracy and effectiveness of the training.

[0044] S140, inputting the course to be predicted into the target grade prediction model to determine the predicted grade of the target object.

[0045] After the loss function of the target prediction model reaches a preset value, the target performance prediction model is obtained, thereby determining the predicted performance of the target object.

[0046] In some embodiments, S140, inputting the course to be predicted into the target grade prediction model to determine the predicted grade of the target object includes: Determine the target subject's mastery of each history course sample's knowledge point based on the history performance characteristics; Determine adjustment parameters based on mastery level; If the mastery level is greater than the first preset value, the sum of the output score of the target score prediction model and the adjustment parameter is determined as the predicted score; if the mastery level is less than the second preset value, the difference between the output score of the target score prediction model and the adjustment parameter is determined as the predicted score; if the mastery level is between the first preset value and the second preset value, the output score of the target score prediction model is determined as the predicted score.

[0047] Taking student grades as an example, in a class or grade, students' grades tend to be normally distributed, that is, there are fewer grades in the low and high segments, and more grades are distributed in the middle segment. Therefore, here we adjust the parameters to compensate for the predicted grades to improve the accuracy of the predicted grades.

[0048] In an optional embodiment, in order to verify the effectiveness of the student score prediction method of the present application, the prediction results of the present application method are compared with those of several other common models, including Transformer, CNN, LSTM, and LSTM+CNN. As shown in Table 1, it can be seen from the table that the present application method has taken a lead over other methods in the data set.

[0049] Taking ACC as an example, the absolute value of the performance improvement of the method proposed in this application compared with the four methods of Transformer, CNN, LSTM, and LSTM+CNN is 24.38%, 3.45%, 3.71%, and 4.39%, respectively, which proves the effectiveness of the method in capturing the relationship between students' historical performance records and MOOC click stream data. Specifically, historical performance is converted into a feature vector representing students' past abilities, and MOOC click stream data is converted into a feature vector representing students' learning attitudes and learning habits towards the current course. Through this fusion strategy, the performance of the model can be significantly improved, thereby more accurately predicting students' performance.

[0050]

[0051] Table 1 The significance test is a method used in statistics to determine whether the difference between two sets of data is statistically significant. It can test whether the results produced by the current method are different from those produced by other methods, and verify whether the difference is accidental. Here, the non-parametric test in the significance test is used to verify the significant difference between the student score prediction results of the current method and the prediction results of the four models of Transformer, CNN, LSTM, and LSTM+CNN.

[0052] In an optional embodiment, 15 rounds of experiments are performed on all methods to obtain 15 rounds of different results. A non-parametric test is used to compare the median difference between the experimental group (method of the present application) and the control group (other methods). The test results show that the difference between the two groups is statistically significant, as reflected in the p value, as shown in Table 2, which is lower than the traditional significance level of 0.05. This shows that the experimental group is significantly improved compared with the control group in indicators such as AUC, ACC, and RMSE, so it can be confirmed that the student performance prediction method proposed in this application has a statistically significant effect. Among them, ***≤0.0001.

[0053]

[0054] Table 2 The embodiments of the present invention combine historical scores with current clickstream data, effectively extract data features from time series data, and use more sufficient data to complete the training of the score prediction model, thereby providing a more comprehensive perspective for score prediction, improving the generalization ability and prediction performance of the score prediction model, and thus improving the accuracy and reliability of score prediction.

[0055] In some embodiments, see Figure 2 , Figure 2 A schematic diagram of the structure of a student performance prediction device provided in an embodiment of the present application; the performance prediction device 200 provided in an embodiment of the present application includes: an acquisition module 210, a weighting module 220, a training module 230 and a determination module 240, wherein: The acquisition module 210 is configured to acquire history course characteristics and a score vector of the history course sample; the history course characteristics are related to the course content, course objectives, teaching methods and evaluation methods; the score vector represents the score information corresponding to the history course sample; A weighting module 220 is configured to perform weighted processing on the score vector based on the cosine distance between the history course feature and the target course feature corresponding to the target course, and determine the history score feature of the target object; A training module 230 is configured to train a preset score prediction model based on historical score features and clickstream features corresponding to a target course in a target time period to generate a target score prediction model; The determination module 240 is configured to input the course to be predicted into the target grade prediction model to determine the predicted grade of the target object.

[0056] In some embodiments, the acquisition module 210 is specifically configured as follows: Perform word segmentation on the course description information of the history course sample, and obtain word vectors from the words and sentences obtained by word segmentation; The TF-IDF statistical method is used to determine the TF-IDF value of each word vector, and the word vectors are weighted and summed based on the TF-IDF value to obtain the text features of the history course sample; The text features are concatenated with the semester information features of the history course samples to obtain the history course features of the history course samples; the semester information features represent the class schedule of the history course samples.

[0057] In some embodiments, the weighting module 220 is specifically configured as follows: Determine the degree of association between each history course sample and the target course based on the cosine distance between the history course characteristics and the target course characteristics corresponding to the target course; Based on the degree of association, the score vectors corresponding to each history course sample are weighted to determine the history score characteristics of the target object.

[0058] In some embodiments, the weighting module 220 is specifically configured as follows: If the cosine distance between the first history course feature and the target course feature is greater than the cosine distance between the second history course feature and the target course feature, the weight of the score vector corresponding to the first history course is less than the weight of the score vector corresponding to the second history course.

[0059] In some embodiments, the training module 230 is specifically configured as follows: Input the historical score features and the clickstream features corresponding to the target course in the target time period into the preset score prediction model to obtain the output score; Determine the loss function corresponding to the preset score prediction model based on the output score and the expected score; The loss function is used in combination with the back propagation algorithm to adjust the parameters of the preset performance prediction model until the loss function reaches the preset value and the target performance prediction model is obtained.

[0060] In some embodiments, the training module 230 is further configured to: Perform fuzzy clustering on history course samples and divide them into different categories by similarity; Based on the historical score features corresponding to different categories of history course samples and the click stream features corresponding to the target courses in the target time period, the preset score prediction models are trained respectively to generate a target score prediction model.

[0061] In this embodiment, fuzzy clustering is performed on history course samples to classify the history course samples, and then a preset score prediction model is trained based on these categories to generate a target score prediction model, thereby ensuring the consistency of the training samples and improving the accuracy and effectiveness of the training.

[0062] In some embodiments, the determination module 240 is further configured to: Determine the target subject's mastery of each history course sample's knowledge point based on the history performance characteristics; Determine adjustment parameters based on mastery level; If the mastery level is greater than the first preset value, the sum of the output score of the target score prediction model and the adjustment parameter is determined as the predicted score; if the mastery level is less than the second preset value, the difference between the output score of the target score prediction model and the adjustment parameter is determined as the predicted score; if the mastery level is between the first preset value and the second preset value, the output score of the target score prediction model is determined as the predicted score.

[0063] The student performance prediction device provided in the embodiment of the present application is based on the above-mentioned student performance prediction method, which will not be described again here to avoid repetition.

[0064] It should be noted that the student score prediction device provided in the embodiment of the present application and the student score prediction method provided in the embodiment of the present application are based on the same application concept. Therefore, the specific implementation of this embodiment can refer to the implementation of the aforementioned student score prediction method, and the repeated parts will not be repeated.

[0065] In some embodiments, see Figure 3 , Figure 3 The electronic device 300 provided in the embodiment of the present application includes a processor 310 and a memory 320; the memory 320 stores a computer program, wherein the computer program implements the above-mentioned student score prediction method when executed by the processor. Specifically, the processor 310 may include, for example, a general-purpose microprocessor, an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 310 may also include an onboard memory for cache purposes. The processor 310 may be a single processing unit or multiple processing units for executing different actions of the method flow according to the embodiment of the present application. The memory 320 may be any medium capable of containing, storing, conveying, propagating or transmitting instructions. For example, the memory 320 may include, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared or semiconductor system, device, component or propagation medium. Specific examples of the memory 320 include: a magnetic storage device, such as a magnetic tape or a hard disk (HDD); an optical storage device, such as a compact disk (CD-ROM); a random access memory (RAM) or a flash memory; and / or a wired / wireless communication link. The present application also provides a computer-readable medium on which a computer program is stored, and when the program is executed by a processor, the above-mentioned student score prediction method is implemented. The computer-readable medium may be included in the device / apparatus / system described in the above-mentioned embodiment; or it may exist independently without being assembled into the device / apparatus / system. The above-mentioned computer-readable medium carries one or more programs, and when the above-mentioned one or more programs are executed, the method according to the embodiment of the present application is implemented. According to an embodiment of the present application, a computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium or any combination of the two. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, wherein a computer-readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wired, optical cable, radio frequency signal, etc., or any suitable combination of the above. Those skilled in the art will appreciate that the features described in the various embodiments and / or claims of the present application may be combined and / or combined in a variety of ways, even if such combinations or combinations are not explicitly described in the present application. In particular, without departing from the spirit and teachings of the present application, the features described in the various embodiments and / or claims of the present application may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present application. Therefore, the scope of the present application should not be limited to the above-described embodiments, but should be determined not only by the appended claims, but also by the equivalents of the appended claims.

Claims

1. A method for predicting student performance, characterized in that: include: Obtaining history course features and a score vector of a history course sample; the history course features are related to course content, course objectives, teaching methods, and evaluation methods; the score vector represents score information corresponding to the history course sample; Based on the cosine distance between the history course feature and the target course feature corresponding to the target course, weighting the score vector to determine the history score feature of the target object; Training a preset score prediction model based on the historical score features and the clickstream features corresponding to the target course in the target time period to generate a target score prediction model; The course to be predicted is input into the target performance prediction model to determine the predicted performance of the target object.

2. The student performance prediction method according to claim 1, characterized in that: The step of obtaining the history course features and score vector of the history course sample includes: Performing word segmentation on the course description information of the history course sample, and obtaining word vectors from the words and sentences obtained by the word segmentation; Determine the TF-IDF value of each word vector by using the TF-IDF statistical method, and perform weighted summation on the word vectors based on the TF-IDF value to obtain the text features of the history course sample; The text feature is concatenated with the semester information feature of the history course sample to obtain the history course feature of the history course sample; the semester information feature represents the class schedule of the history course sample.

3. The student performance prediction method according to claim 1, characterized in that: The step of weighting the score vector based on the cosine distance between the history course feature and the target course feature corresponding to the target course to determine the history score feature of the target object includes: Determining the degree of association between each history course sample and the target course based on the cosine distance between the history course feature and the target course feature corresponding to the target course; Based on the degree of association, the score vectors corresponding to the various history course samples are weighted to determine the history score characteristics of the target object.

4. The student performance prediction method according to claim 3, characterized in that: The step of weighting the score vector based on the cosine distance between the history course feature and the target course feature corresponding to the target course to determine the history score feature of the target object includes: If the cosine distance between the first history course feature and the target course feature is greater than the cosine distance between the second history course feature and the target course feature, the weight of the score vector corresponding to the first history course is less than the weight of the score vector corresponding to the second history course.

5. The student performance prediction method according to claim 1, characterized in that: The step of training a preset score prediction model based on the historical score features and the click stream features corresponding to the target course in the target time period to generate a target score prediction model includes: Inputting the historical score features and the clickstream features corresponding to the target course in the target time period into the preset score prediction model to obtain an output score; Determine the loss function corresponding to the preset score prediction model based on the output score and the expected score; The loss function is used in combination with a back-propagation algorithm to adjust the parameters of the preset performance prediction model until the loss function reaches a preset value, thereby obtaining a target performance prediction model.

6. The student performance prediction method according to claim 1, characterized in that: The training of a preset score prediction model based on the historical score features and the click stream features corresponding to the target course in the target time period to generate a target score prediction model also includes: Performing fuzzy clustering on the history course samples, and dividing the history course samples into different categories by similarity; The preset score prediction model is trained based on the historical score features corresponding to the history course samples of different categories and the click stream features corresponding to the target course in the target time period to generate a target score prediction model.

7. The student performance prediction method according to claim 1, characterized in that: The step of inputting the course to be predicted into the target grade prediction model to determine the predicted grade of the target object includes: Determining the target subject's mastery of the knowledge points of each history course sample according to the history performance characteristics; determining an adjustment parameter based on the mastery level; If the mastery level is greater than a first preset value, the sum of the output score of the target score prediction model and the adjustment parameter is determined as the predicted score; if the mastery level is less than a second preset value, the difference between the output score of the target score prediction model and the adjustment parameter is determined as the predicted score; if the mastery level is between the first preset value and the second preset value, the output score of the target score prediction model is determined as the predicted score.

8. A student performance prediction device, characterized in that: include: An acquisition module, a weighting module, a training module and a determination module, wherein: The acquisition module is configured to acquire history course characteristics and a score vector of the history course sample; the history course characteristics are related to course content, course objectives, teaching methods and evaluation methods; the score vector represents the score information corresponding to the history course sample; The weighting module is configured to perform weighted processing on the score vector based on the cosine distance between the history course feature and the target course feature corresponding to the target course, so as to determine the history score feature of the target object; The training module is configured to train a preset score prediction model based on the historical score features and the clickstream features corresponding to the target course in the target time period to generate a target score prediction model; The determination module is configured to input the course to be predicted into the target grade prediction model to determine the predicted grade of the target object.

9. An electronic device comprising a processor and a memory; the memory stores a computer program, wherein: When the computer program is executed by the processor, the computer program implements the student performance prediction method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, wherein when the program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images