Education platform interactive method, system and electronic device based on mobile Internet

Abstract

The present application discloses an interactive method, system and electronic device for an educational platform based on mobile Internet. After obtaining a first semantic feature vector, a second semantic feature vector and a third semantic feature vector from the questions to be assigned, the questions that the students have answered before and the portrait data of the students themselves through a semantic understanding model, the method of Bayesian estimation plus maximum conditional likelihood estimation is adopted to remove the differences in feature dimensions and information amounts of these feature vectors at the probability space level, thereby accurately estimating the similarity score, so that the matching result between the questions to be assigned and the students is more accurate.

Description

Technical Field

[0001] The present invention relates to the field of Internet education, and more specifically, to an educational platform interaction method, system and electronic device based on mobile Internet. Background Art

[0002] Internet education is a new form of education that combines Internet technology with the field of education as science and technology continues to develop. At present, Internet education is constantly promoting the innovation of future education models. For the online education industry, how to popularize education, how to use technology to integrate better educational resources and talent resources, and bring new development vitality and models to the education industry are the common expectations of the education industry.

[0003] However, the existing Internet education can only provide one-way education. For example, online courses recorded by teachers from famous schools and online teaching in tutoring classes can only be taught by excellent teachers through video. After the teachers teach, they usually need to assign homework to students to review the knowledge they have learned. However, due to the different learning abilities of each student and differences in age, academic year, and gender, the learning abilities of each student will be different. In this way, the teaching system cannot assign appropriate and matching homework according to the differences between students when assigning homework, which will prevent students from getting the best practice results, and thus lead to poor learning results.

[0004] Therefore, in order to assign matching homework questions based on students' personal circumstances, an interactive method of an educational platform based on mobile Internet is desired. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiments of the present application provide an interactive method, system and electronic device for an education platform based on mobile Internet, which obtains the first semantic feature vector, the second semantic feature vector and the third semantic feature vector through a semantic understanding model using the questions to be assigned, the questions that the students have answered before and the portrait data of the students themselves, and then uses the Bayesian estimation plus the maximum conditional likelihood estimation method to remove the differences in feature dimensions and information amounts of these feature vectors at the probability space level, thereby accurately estimating the similarity score, so that the matching result between the questions to be assigned and the students is more accurate.

[0006] According to one aspect of the present application, a method for interactive education platform based on mobile Internet is provided, which comprises:

[0007] Get the text data of the questions to be assigned and the questions that the students have previously answered;

[0008] Using a semantic understanding model for question understanding, respectively obtaining a first semantic feature vector and a second semantic feature vector from the text data of the question to be assigned and the text data of the question that the student has previously answered;

[0009] Acquire the portrait data of the student and obtain a third semantic feature vector corresponding to the portrait data through the semantic understanding model;

[0010] Adjusting the first semantic feature vector, the second semantic feature vector and the third semantic feature vector to the same length and uniformly mapping them into a probability space to obtain a first probabilistic semantic feature vector, a second probabilistic semantic feature vector and the third probabilistic semantic feature vector;

[0011] Taking the feature value of each position in the third probabilistic semantic feature vector as the prior probability, the feature value of each position in the first probabilistic semantic feature vector as the event probability and the feature value of each position in the second probabilistic semantic feature vector as the evidence probability, the posterior probability of each position in the third probabilistic semantic feature vector is calculated based on the Bayesian algorithm to obtain a Bayesian feature vector, wherein the posterior probability is equal to the prior probability multiplied by the event probability divided by the evidence probability;

[0012] Performing maximum conditional likelihood estimation on the Bayesian feature vector to obtain a classification feature vector composed of a plurality of maximum conditional natural estimation values, wherein the maximum conditional likelihood estimation is performed based on a cosine distance between feature values ​​at each position in the first probabilized semantic feature vector and the second probabilized semantic feature vector;

[0013] Passing the classification feature vector through a classifier to obtain a classification result, wherein the classification result is used to represent a matching result between the to-be-assigned topic and the student; and

[0014] Based on the classification results, questions are assigned to the students on a mobile Internet-based education platform.

[0015] According to another aspect of the present application, there is provided an educational platform interactive system based on mobile Internet, comprising:

[0016] A text data acquisition unit, used to acquire text data of questions to be assigned and questions that students have previously answered;

[0017] a feature vector generating unit, configured to obtain a first semantic feature vector and a second semantic feature vector respectively from the text data of the to-be-assigned question obtained by the text data obtaining unit and the text data of the question that the student has previously answered obtained by the text data obtaining unit, using a semantic understanding model for question understanding;

[0018] A portrait data feature vector generating unit, used for acquiring the portrait data of the student and obtaining a third semantic feature vector corresponding to the portrait data through the semantic understanding model;

[0019] A probability mapping unit, used for adjusting the first semantic feature vector obtained by the feature vector generating unit, the second semantic feature vector obtained by the feature vector generating unit, and the third semantic feature vector obtained by the portrait data feature vector generating unit to the same length and uniformly mapping them into a probability space to obtain a first probabilistic semantic feature vector, a second probabilistic semantic feature vector, and the third probabilistic semantic feature vector;

[0020] A Bayesian unit, configured to calculate the posterior probability of each position in the third probabilistic semantic feature vector obtained by the probability mapping unit based on a Bayesian algorithm, using the feature value of each position in the third probabilistic semantic feature vector obtained by the probability mapping unit as a prior probability, the feature value of each position in the first probabilistic semantic feature vector obtained by the probability mapping unit as an event probability, and the feature value of each position in the second probabilistic semantic feature vector obtained by the probability mapping unit as an evidence probability, to obtain a Bayesian feature vector, wherein the posterior probability is equal to the prior probability multiplied by the event probability divided by the evidence probability;

[0021] a maximum conditional likelihood estimation unit, configured to perform maximum conditional likelihood estimation on the Bayesian feature vector obtained by the Bayesian unit to obtain a classification feature vector composed of a plurality of maximum conditional natural estimation values, wherein the maximum conditional likelihood estimation is performed based on a cosine distance between feature values ​​at each position in the first probabilized semantic feature vector and the second probabilized semantic feature vector;

[0022] a classifier processing unit, configured to pass the classification feature vector obtained by the maximum conditional likelihood estimation unit through a classifier to obtain a classification result, wherein the classification result is used to represent a matching result between the to-be-assigned topic and the student; and

[0023] An allocating unit is used to allocate questions to the students on a mobile Internet-based education platform based on the classification result obtained by the classifier processing unit.

[0024] According to another aspect of the present application, an electronic device is provided, comprising: a processor; and a memory, in which computer program instructions are stored, and when the computer program instructions are executed by the processor, the processor executes the mobile Internet-based education platform interaction method as described above.

[0025] According to yet another aspect of the present application, a computer-readable medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the mobile Internet-based education platform interaction method as described above.

[0026] Compared with the prior art, the mobile Internet-based educational platform interactive method, system and electronic device provided by the present application, after obtaining the first semantic feature vector, the second semantic feature vector and the third semantic feature vector through the semantic understanding model of the questions to be assigned, the questions that the students have previously answered and the portrait data of the students themselves, adopts the method of Bayesian estimation plus maximum conditional likelihood estimation. It can remove the differences in feature dimensions and information amounts of these feature vectors at the probability space level, thereby accurately estimating the similarity score, so that the matching results between the questions to be assigned and the students are more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0028] Figure 1 is a flow chart of an educational platform interaction method based on mobile Internet according to an embodiment of the present application;

[0029] Figure 2 A schematic diagram of the system architecture of an educational platform interactive method based on mobile Internet according to an embodiment of the present application;

[0030] Figure 3 A flowchart of obtaining a first semantic feature vector and a second semantic feature vector from the text data of the to-be-assigned question and the text data of the question that the student has previously answered, respectively, using a semantic understanding model for question understanding in an interactive method for an education platform based on a mobile Internet according to an embodiment of the present application;

[0031] Figure 4 A flowchart of performing maximum conditional likelihood estimation on the Bayesian feature vector to obtain a classification feature vector composed of multiple maximum conditional natural estimation values ​​in an interactive method for an education platform based on a mobile Internet according to an embodiment of the present application;

[0032] Figure 5 A block diagram of an interactive system for an education platform based on mobile Internet according to an embodiment of the present application;

[0033] Figure 6A block diagram of a feature vector generation unit in an interactive education platform system based on mobile Internet according to an embodiment of the present application;

[0034] Figure 7 A block diagram of a maximum conditional likelihood estimation unit in an interactive education platform system based on mobile Internet according to an embodiment of the present application;

[0035] Figure 8 is a block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0036] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0037] Scenario Overview

[0038] As mentioned above, existing Internet education can only provide one-way education, such as online courses recorded by teachers from famous schools, online teaching of tutoring classes, etc. These can only allow excellent teachers to give video lectures. After the teachers teach, they usually need to assign homework to students to review the knowledge they have learned. However, due to the different learning abilities of each student and the differences in age, school year, and gender, the acceptance abilities of each student will be different. In this way, when assigning homework, the teaching system cannot assign appropriate and matching homework according to the differences between students, which will prevent students from getting the best practice effect, and thus lead to poor learning results. Therefore, in order to assign matching homework questions based on the individual situation of students, an interactive method of an education platform based on mobile Internet is expected.

[0039] Based on this, in order to reasonably assign homework questions to students, we need to consider not only the students' personal situations, but also the students' previous answers and the questions to be assigned. Therefore, it is necessary to express the features of this data appropriately and make reasonable matches based on the expressed features.

[0040] Specifically, a semantic understanding model for question understanding is used to obtain a first semantic feature vector and a second semantic feature vector from the text data of the question to be assigned and the question that the student has previously answered. The semantic understanding model is, for example, a Bert model superimposed with a bidirectional LSTM model. In addition, the student's own portrait data is also represented by a weighted word sequence, so it is also processed using the above semantic understanding model to obtain a third semantic feature vector.

[0041] Then, the first to third semantic feature vectors are first converted to the same length through linear transformation and mapped to the probability space with a Softmax-like function. Then, the feature values ​​of each position in the third semantic feature vector, the first semantic feature vector and the second semantic feature vector are used as the prior probability, event probability and evidence probability respectively, and the Bayesian feature vector is obtained according to the Bayesian algorithm.

[0042] Next, the maximum conditional likelihood estimation is performed based on the Bayesian feature vector, that is, the cosine distance between the feature values ​​at each position of the first and second semantic feature vectors is first calculated to obtain the angle vector The maximum conditional likelihood estimate is then calculated for each position of the Bayesian eigenvector:

[0043]

[0044] Among them, v i is the eigenvalue of each position of the Bayesian eigenvector V, Represents an angle vector The second norm of , α is a predetermined parameter, for example, a hyperparameter.

[0045] Then, the classification feature vector is obtained from the maximum conditional likelihood estimate of each position, and then through the classifier, the matching result between the assigned questions and the students can be obtained.

[0046] Based on this, the present application proposes an interactive method for an educational platform based on mobile Internet, which includes: obtaining text data of questions to be assigned and questions that students have answered before; using a semantic understanding model for question understanding to obtain a first semantic feature vector and a second semantic feature vector from the text data of the questions to be assigned and the text data of the questions that students have answered before; obtaining the student's portrait data and obtaining a third semantic feature vector corresponding to the portrait data through the semantic understanding model; adjusting the first semantic feature vector, the second semantic feature vector and the third semantic feature vector to the same length and uniformly mapping them to a probability space to obtain a first probabilistic semantic feature vector, a second probabilistic semantic feature vector and the third probabilistic semantic feature vector; using the feature values ​​of each position in the third probabilistic semantic feature vector as the prior probability, and mapping the feature values ​​of each position in the first probabilistic semantic feature vector as the prior probability. The characteristic value of the position is used as the event probability and the characteristic value of each position in the second probabilistic semantic feature vector is used as the evidence probability, and the posterior probability of each position in the third probabilistic semantic feature vector is calculated based on the Bayesian algorithm to obtain a Bayesian feature vector, wherein the posterior probability is equal to the prior probability multiplied by the event probability divided by the evidence probability; the maximum conditional likelihood estimation is performed on the Bayesian feature vector to obtain a classification feature vector composed of multiple maximum conditional natural estimation values, wherein the maximum conditional likelihood estimation is based on the cosine distance between the characteristic values ​​of each position in the first probabilistic semantic feature vector and the second probabilistic semantic feature vector; the classification feature vector is passed through a classifier to obtain a classification result, and the classification result is used to represent the matching result between the question to be assigned and the student; and, based on the classification result, the question is assigned to the student on an education platform based on mobile Internet.

[0047] After introducing the basic principles of the present application, various non-limiting embodiments of the present application will be described in detail with reference to the accompanying drawings.

[0048] Exemplary Methods

[0049] Figure 1 The figure shows a flow chart of the interactive method of the education platform based on mobile Internet. Figure 1As shown, according to the embodiment of the present application, the interactive method of the education platform based on the mobile Internet includes: S110, obtaining the text data of the questions to be assigned and the questions that the students have answered before; S120, using the semantic understanding model for question understanding to obtain the first semantic feature vector and the second semantic feature vector from the text data of the questions to be assigned and the text data of the questions that the students have answered before; S130, obtaining the portrait data of the student and obtaining the third semantic feature vector corresponding to the portrait data through the semantic understanding model; S140, adjusting the first semantic feature vector, the second semantic feature vector and the third semantic feature vector to the same length and uniformly mapping them to the probability space to obtain the first probabilistic semantic feature vector, the second probabilistic semantic feature vector and the third probabilistic semantic feature vector; S150, using the feature values ​​of each position in the third probabilistic semantic feature vector as the prior probability, the first probabilistic semantic feature vector and the third probabilistic semantic feature vector. The eigenvalues ​​of each position in the vector are used as event probabilities and the eigenvalues ​​of each position in the second probabilized semantic feature vector are used as evidence probabilities, and the posterior probabilities of each position in the third probabilized semantic feature vector are calculated based on the Bayesian algorithm to obtain a Bayesian feature vector, wherein the posterior probability is equal to the prior probability multiplied by the event probability divided by the evidence probability; S160, performing maximum conditional likelihood estimation on the Bayesian feature vector to obtain a classification feature vector composed of multiple maximum conditional natural estimation values, wherein the maximum conditional likelihood estimation is based on the cosine distance between the eigenvalues ​​of each position in the first probabilized semantic feature vector and the second probabilized semantic feature vector; S170, passing the classification feature vector through a classifier to obtain a classification result, and the classification result is used to represent the matching result between the question to be assigned and the student; and, S180, based on the classification result, assigning questions to the students on an education platform based on mobile Internet.

[0050] Figure 2 The figure shows a schematic diagram of the architecture of the interactive method of the education platform based on the mobile Internet according to the embodiment of the present application. Figure 2 As shown, in the network architecture of the mobile Internet-based educational platform interactive method, first, a semantic understanding model for question understanding is used (for example, Figure 2 SUM) are obtained from the text data of the subject to be assigned (for example, Figure 2 P1) and text data of questions that the student has previously answered (e.g., Figure 2 P2 as shown in FIG. 2 ) obtains a first semantic feature vector (for example, Figure 2 FV1 as shown in FIG) and a second semantic feature vector (e.g., Figure 2Then, the semantic understanding model is used to obtain the portrait data corresponding to the acquired student (for example, Figure 2 The third semantic feature vector (e.g., P3) of Figure 2 Then, the first semantic feature vector, the second semantic feature vector and the third semantic feature vector are adjusted to the same length and uniformly mapped into the probability space to obtain a first probabilistic semantic feature vector (for example, Figure 2 V1 as shown in FIG), a second probabilistic semantic feature vector (e.g., Figure 2 V2) and the third probabilistic semantic feature vector (e.g., Figure 2 V3 as shown in the figure); then, taking the feature value of each position in the third probabilistic semantic feature vector as the prior probability, the feature value of each position in the first probabilistic semantic feature vector as the event probability and the feature value of each position in the second probabilistic semantic feature vector as the evidence probability, the posterior probability of each position in the third probabilistic semantic feature vector is calculated based on the Bayesian algorithm to obtain a Bayesian feature vector (for example, Figure 2 Then, the Bayesian feature vector is subjected to maximum conditional likelihood estimation to obtain a classification feature vector composed of multiple maximum conditional natural estimation values ​​(for example, Figure 2 Then, the classification feature vector is passed through a classifier (e.g., Figure 2 The circle S shown in FIG) is used to obtain a classification result, wherein the classification result is used to represent the matching result between the topic to be assigned and the student; and, finally, based on the classification result, the topic is assigned to the student on an education platform based on the mobile Internet.

[0051] In step S110 and step S120, the text data of the questions to be assigned and the questions that the students have answered before are obtained; and the first semantic feature vector and the second semantic feature vector are obtained from the text data of the questions to be assigned and the text data of the questions that the students have answered before using the semantic understanding model for question understanding. As mentioned above, in order to reasonably assign homework questions to students, it is necessary not only to consider the personal situation of the students, but also to consider the students' previous answering situation and the situation of the questions to be assigned. Therefore, it is necessary to perform appropriate feature expression on the data in this regard and perform reasonable matching based on the expressed features. Therefore, in the technical solution of the present application, first, the text data of the questions to be assigned and the questions that the students have answered before are obtained from the cloud storage end of the education platform; then, the first semantic feature vector and the second semantic feature vector are obtained from the text data of the questions to be assigned and the questions that the students have answered before using the semantic understanding model for question understanding. That is, the text data of the questions to be assigned and the questions that the students have answered before are processed by the semantic understanding model to extract the semantic information in the text data, thereby obtaining the first semantic feature vector and the second semantic feature vector. It is worth mentioning that, here, the semantic understanding model can be a Bert model superimposed with a bidirectional LSTM model.

[0052] Specifically, in an embodiment of the present application, a semantic understanding model for question understanding is used to obtain a first semantic feature vector and a second semantic feature vector from the text data of the question to be assigned and the text data of the question that the student has previously answered, respectively, including: first, the text data of the question to be assigned and the text data of the question that the student has previously answered are respectively subjected to word segmentation processing; then, the text data of the question to be assigned after word segmentation processing and the text data of the question that the student has previously answered after word segmentation processing are respectively passed through a word embedding model to obtain a first word vector sequence and a second word vector sequence; then, the first word vector sequence and the second word vector sequence are respectively passed through a Bert model to obtain a first word feature vector sequence and a second word feature vector sequence; and, finally, the first word feature vector sequence and the second word feature vector sequence are respectively passed through a bidirectional LSTM model to obtain the first semantic feature vector and the second semantic feature vector. It should be understood that because text data has a certain order, therefore, when extracting features from text data, it is also necessary to map this order to a high-dimensional space to achieve semantic coherence, thereby obtaining a semantic feature vector.

[0053] Figure 3The figure shows a flow chart of obtaining a first semantic feature vector and a second semantic feature vector from the text data of the to-be-assigned question and the text data of the question that the student has previously answered, respectively, using a semantic understanding model for question understanding in an interactive method for an education platform based on mobile Internet according to an embodiment of the present application. Figure 3 As shown, in an embodiment of the present application, a semantic understanding model for question understanding is used to obtain a first semantic feature vector and a second semantic feature vector from the text data of the question to be assigned and the text data of the question that the student has previously answered, respectively, including: S210, performing word segmentation processing on the text data of the question to be assigned and the text data of the question that the student has previously answered, respectively; S220, passing the text data of the question to be assigned after word segmentation processing and the text data of the question that the student has previously answered after word segmentation processing, respectively, through a word embedding model to obtain a first word vector sequence and a second word vector sequence; S230, passing the first word vector sequence and the second word vector sequence, respectively, through a Bert model to obtain a first word feature vector sequence and a second word feature vector sequence; and, S240, passing the first word feature vector sequence and the second word feature vector sequence, respectively, through a bidirectional LSTM model to obtain the first semantic feature vector and the second semantic feature vector.

[0054] In step S130, the portrait data of the student is obtained and the third semantic feature vector corresponding to the portrait data is obtained through the semantic understanding model. It should be understood that the portrait data of the student itself is also represented by a weighted word sequence. Therefore, in the technical solution of the present application, similarly, the above-mentioned semantic understanding model is also used for processing to extract the semantic information in the text data, thereby obtaining the third semantic feature vector. It is worth mentioning that here, the semantic understanding model can be a Bert model superimposed with a bidirectional LSTM model.

[0055] In step S140, the first semantic feature vector, the second semantic feature vector and the third semantic vector are adjusted to the same length and uniformly mapped to the probability space to obtain the first probabilistic semantic feature vector, the second probabilistic semantic feature vector and the third probabilistic semantic feature vector. That is, first, the first to the third semantic feature vectors are first converted to the same length through a linear transformation and mapped to the probability space with a Softmax-like function. It should be understood that if the similarity score is estimated only based on the distance between the first semantic feature vector and the second semantic feature vector, then when the score is fused with the third semantic feature vector, there will be huge differences in feature dimension and information amount, thereby affecting the fusion accuracy.

[0056] Accordingly, in a specific example of the present application, the process of adjusting the first semantic feature vector, the second semantic feature vector and the third semantic vector to the same length and uniformly mapping them into the probability space to obtain the first probabilistic semantic feature vector, the second probabilistic semantic feature vector and the third probabilistic semantic feature vector includes: first, adjusting the first semantic feature vector, the second semantic feature vector and the third semantic vector to the same length through a linear transformation to facilitate subsequent calculations and improve the accuracy of fusion; then, the first semantic feature vector, the second semantic feature vector and the third semantic vector adjusted to the same length are input into a Softmax-like function to map the feature values ​​of each position in the first semantic feature vector, the second semantic feature vector and the third semantic vector to a probability interval of 0 to 1 to obtain the first probabilistic semantic feature vector, the second probabilistic semantic feature vector and the third probabilistic semantic feature vector, wherein the Softmax-like function is y=exp(xi) / ∑ i exp(xi). It should be understood that by processing in this way, the dimensional influence and differences between data features can be eliminated, which is convenient for measurement and subsequent calculations.

[0057] In particular, in another specific example of the present application, the process of adjusting the first semantic feature vector, the second semantic feature vector and the third semantic vector to the same length and uniformly mapping them to the probability space to obtain the first probabilistic semantic feature vector, the second probabilistic semantic feature vector and the third probabilistic semantic feature vector includes: first, adjusting the first semantic feature vector, the second semantic feature vector and the third semantic feature vector to the same length through linear transformation to facilitate subsequent calculations and improve the accuracy of fusion; then, the first semantic feature vector, the second semantic feature vector and the third semantic vector adjusted to the same length are input into the Sigmoid function to map the feature values ​​of each position in the first semantic feature vector, the second semantic feature vector and the third semantic vector to a probability interval of 0 to 1 to obtain the first probabilistic semantic feature vector, the second probabilistic semantic feature vector and the third probabilistic semantic feature vector. It should be understood that by processing in this way, the differences in feature dimensions and information volume of these feature vectors can be removed at the probability space level, which is convenient for subsequent calculations.

[0058] In step S150, the feature values ​​of each position in the third probabilized semantic feature vector are used as prior probabilities, the feature values ​​of each position in the first probabilized semantic feature vector are used as event probabilities, and the feature values ​​of each position in the second probabilized semantic feature vector are used as evidence probabilities. The posterior probabilities of each position in the third probabilized semantic feature vector are calculated based on the Bayesian algorithm to obtain a Bayesian feature vector, wherein the posterior probability is equal to the prior probability multiplied by the event probability divided by the evidence probability. It should be understood that in order to reduce the huge differences in feature dimensions and information volume between the first and second semantic feature vectors when they are fused with the third semantic feature vector, so as to improve the fusion accuracy. Therefore, in the technical solution of the present application, the method of Bayesian estimation plus maximum conditional likelihood estimation is adopted. That is, first, the feature values ​​of each position in the third semantic feature vector, the first semantic feature vector and the second semantic feature vector are used as prior probabilities, event probabilities and evidence probabilities, respectively, and the Bayesian feature vector is obtained according to the Bayesian algorithm.

[0059] Specifically, in an embodiment of the present application, the process of taking the feature value of each position in the third probabilistic semantic feature vector as the prior probability, the feature value of each position in the first probabilistic semantic feature vector as the event probability and the feature value of each position in the second probabilistic semantic feature vector as the evidence probability, and calculating the posterior probability of each position in the third probabilistic semantic feature vector based on the Bayesian algorithm to obtain a Bayesian feature vector includes: calculating the Bayesian probability of each position in the third probabilistic semantic feature vector according to the following formula to obtain a Bayesian feature vector; the formula is: Pi = Ai*Bi / Ci, wherein Ai represents the prior probability being the feature value of each position in the third probabilistic semantic feature vector, Bi represents the event probability being the feature value of each position in the first probabilistic semantic feature vector, and Ci represents the evidence probability being the feature value of each position in the second probabilistic semantic feature vector.

[0060] In step S160, the maximum conditional likelihood estimation is performed on the Bayesian feature vector to obtain a classification feature vector composed of multiple maximum conditional natural estimation values, wherein the maximum conditional likelihood estimation is performed based on the cosine distance between the feature values ​​at each position in the first probabilistic semantic feature vector and the second probabilistic semantic feature vector. It should be understood that in order to reduce the differences in feature dimension and amount of information during fusion, the maximum conditional likelihood estimation is used after the Bayesian estimation to improve the accuracy of the fusion. That is, the cosine distance between the feature values ​​at each position of the first and second semantic feature vectors is first calculated to obtain the angle vector The maximum conditional likelihood estimate is then calculated for each position of the Bayesian feature vector.

[0061] Specifically, in an embodiment of the present application, the process of performing maximum conditional likelihood estimation on the Bayesian feature vector to obtain a classification feature vector composed of multiple maximum conditional natural estimation values ​​includes: first, calculating the cosine distance between the feature values ​​at each position in the first probabilistic semantic feature vector and the second probabilistic semantic feature vector to obtain an angle vector composed of multiple cosine distances. Then, based on the angle vector, the maximum conditional likelihood estimate corresponding to the eigenvalues ​​at each position in the Bayesian feature vector is calculated to obtain the classification feature vector. Accordingly, in a specific example, based on the angle vector, the maximum conditional likelihood estimate corresponding to the eigenvalues ​​at each position in the Bayesian feature vector is calculated using the following formula to obtain the classification feature vector; wherein the formula is:

[0062]

[0063] Among them, v i is the eigenvalue of each position of the Bayesian eigenvector V, Represents an angle vector The second norm of , α is a predetermined parameter, for example, can be used as a hyperparameter.

[0064] Figure 4 The figure shows a flow chart of performing maximum conditional likelihood estimation on the Bayesian feature vector in the mobile Internet-based education platform interaction method according to an embodiment of the present application to obtain a classification feature vector composed of multiple maximum conditional natural estimation values. Figure 4 As shown, in an embodiment of the present application, the maximum conditional likelihood estimation is performed on the Bayesian feature vector to obtain a classification feature vector composed of multiple maximum conditional natural estimation values, including: S310, calculating the cosine distance between the feature values ​​at each position in the first probabilistic semantic feature vector and the second probabilistic semantic feature vector to obtain an angle vector composed of multiple cosine distances; and, S320, based on the angle vector, calculating the maximum conditional likelihood estimation value corresponding to the feature value at each position in the Bayesian feature vector to obtain the classification feature vector.

[0065] In step S170 and step S180, the classification feature vector is passed through a classifier to obtain a classification result, and the classification result is used to represent the matching result between the topic to be assigned and the student, and based on the classification result, the topic is assigned to the student on the mobile Internet-based education platform. That is, in a specific example, first, the classification feature vector is input into the Softmax classification function of the classifier to obtain a first probability that the classification feature vector belongs to a match between the topic to be assigned and the student and a second probability that the classification feature vector belongs to a mismatch between the topic to be assigned and the student. Then, based on the comparison between the first probability and the second probability, the classification result is determined. Specifically, when the first probability is greater than the second probability, the classification result is a match between the topic to be assigned and the student; when the first probability is less than the second probability, the classification result is a mismatch between the topic to be assigned and the student. Finally, in response to the classification result being a match between the topic to be assigned and the student, the topic to be classified is assigned to the student through the mobile Internet-based education platform.

[0066] In summary, the interactive method of the education platform based on mobile Internet in the embodiment of the present application is explained. After obtaining the first semantic feature vector, the second semantic feature vector and the third semantic feature vector through the semantic understanding model of the questions to be assigned, the questions that the students have previously answered and the portrait data of the students themselves, the Bayesian estimation plus the maximum conditional likelihood estimation method is adopted. The differences in feature dimensions and information amounts of these feature vectors can be removed at the probability space level, thereby accurately estimating the similarity score, so that the matching results between the questions to be assigned and the students are more accurate.

[0067] Exemplary Systems

[0068] Figure 5 FIG. 1 is a block diagram of an interactive education platform based on mobile Internet according to an embodiment of the present application. Figure 5As shown, according to the embodiment of the present application, the interactive system 600 of the education platform based on the mobile Internet includes: a text data acquisition unit 610, which is used to acquire the text data of the questions to be assigned and the questions that the students have answered before; a feature vector generation unit 620, which is used to use the semantic understanding model for question understanding to obtain the first semantic feature vector and the second semantic feature vector from the text data of the questions to be assigned obtained by the text data acquisition unit 610 and the text data of the questions that the students have answered before obtained by the text data acquisition unit 610; a portrait data feature vector generation unit 630, which is used to acquire the portrait data of the student and generate the first semantic feature vector and the second semantic feature vector from the text data of the questions to be assigned obtained by the text data acquisition unit 610 and the text data of the questions that the students have answered before obtained by the text data acquisition unit 610; and a portrait data feature vector generation unit 630, which is used to acquire the portrait data of the student and generate the first semantic feature vector and the second semantic feature vector from the text data of the questions to be assigned obtained by the text data acquisition unit 610 and the text data of the questions that the students have answered before obtained by the text data acquisition unit 610. model to obtain a third semantic feature vector corresponding to the portrait data; a probability mapping unit 640, which is used to adjust the first semantic feature vector obtained by the feature vector generation unit 620, the second semantic feature vector obtained by the feature vector generation unit 620 and the third semantic vector obtained by the portrait data feature vector generation unit 630 to the same length and uniformly map them into a probability space to obtain a first probabilistic semantic feature vector, a second probabilistic semantic feature vector and the third probabilistic semantic feature vector; a Bayesian unit 650, which is used to use the feature values ​​of each position in the third probabilistic semantic feature vector obtained by the probability mapping unit 640 as the prior probability, the feature value of each position in the first probabilistic semantic feature vector obtained by the probability mapping unit 640 as the event probability and the feature value of each position in the second probabilistic semantic feature vector obtained by the probability mapping unit 640 as the evidence probability, and the posterior probability of each position in the third probabilistic semantic feature vector obtained by the probability mapping unit is calculated based on the Bayesian algorithm to obtain a Bayesian feature vector, wherein the posterior probability is equal to the prior probability multiplied by the event probability divided by the evidence probability; a maximum conditional likelihood estimation unit 660 is used to perform maximum conditional likelihood on the Bayesian feature vector obtained by the Bayesian unit 650 A classification feature vector composed of multiple maximum conditional natural estimation values ​​is obtained by naturally estimating the maximum conditional likelihood, wherein the maximum conditional likelihood estimation is performed based on the cosine distance between the feature values ​​at each position in the first probabilistic semantic feature vector and the second probabilistic semantic feature vector; a classifier processing unit 670 is used to pass the classification feature vector obtained by the maximum conditional likelihood estimation unit 660 through a classifier to obtain a classification result, and the classification result is used to represent the matching result between the question to be assigned and the student; and an assignment unit 680 is used to assign questions to the student on an education platform based on the mobile Internet based on the classification result obtained by the classifier processing unit 670.

[0069] In one example, in the above-mentioned mobile Internet-based education platform interactive system 600, as Figure 6As shown, the feature vector generating unit 620 includes: a word segmentation subunit 621, which is used to perform word segmentation processing on the text data of the to-be-assigned question and the text data of the question that the student has answered before, respectively; a word embedding model subunit 622, which is used to pass the text data of the to-be-assigned question obtained by the word segmentation subunit 621 after the word segmentation processing and the text data of the question that the student has answered before, respectively, through a word embedding model to obtain a first word vector sequence and a second word vector sequence; a Bert model subunit 623, which is used to pass the first word vector sequence obtained by the word embedding model subunit 622 and the second word vector sequence obtained by the word embedding model subunit 622 through a Bert model to obtain a first word feature vector sequence and a second word feature vector sequence; and an LSTM model subunit 624, which is used to pass the first word feature vector sequence obtained by the Bert model subunit 623 and the second word feature vector sequence obtained by the Bert model subunit 623 through a bidirectional LSTM model to obtain the first semantic feature vector and the second semantic feature vector.

[0070] In one example, in the above-mentioned mobile Internet-based educational platform interactive system 600, the probability mapping unit 640 is further used to: adjust the first semantic feature vector, the second semantic feature vector and the third semantic vector to the same length through linear transformation; and, input the first semantic feature vector, the second semantic feature vector and the third semantic vector adjusted to the same length into a Softmax-like function to map the feature values ​​of each position in the first semantic feature vector, the second semantic feature vector and the third semantic vector to a probability interval of 0 to 1, so as to obtain the first probabilistic semantic feature vector, the second probabilistic semantic feature vector and the third probabilistic semantic feature vector, wherein the Softmax-like function is y=exp(xi) / ∑ i exp(xi).

[0071] In one example, in the above-mentioned mobile Internet-based education platform interactive system 600, the probability mapping unit 640 is further used to: adjust the first semantic feature vector, the second semantic feature vector and the third semantic vector to the same length through linear transformation; and, input the first semantic feature vector, the second semantic feature vector and the third semantic vector adjusted to the same length into a Sigmoid function to map the feature values ​​of each position in the first semantic feature vector, the second semantic feature vector and the third semantic vector to a probability interval of 0 to 1, so as to obtain the first probabilistic semantic feature vector, the second probabilistic semantic feature vector and the third probabilistic semantic feature vector.

[0072] In one example, in the above-mentioned mobile Internet-based education platform interactive system 600, as Figure 7 As shown, the maximum conditional likelihood estimation unit 660 includes: a cosine distance calculation subunit 661, used to calculate the cosine distance between the feature values ​​at each position in the first probabilistic semantic feature vector and the second probabilistic semantic feature vector to obtain an angle vector composed of multiple cosine distances; a classification feature vector generation subunit 662, used to calculate the maximum conditional likelihood estimation value corresponding to the feature value at each position in the Bayesian feature vector based on the angle vector obtained by the cosine distance calculation subunit 661 to obtain the classification feature vector.

[0073] In one example, in the above-mentioned mobile Internet-based education platform interactive system 600, the classification feature vector generating subunit 662 is further used to: based on the angle vector, calculate the maximum conditional likelihood estimate corresponding to the feature value at each position in the Bayesian feature vector using the following formula to obtain the classification feature vector; wherein the formula is:

[0074]

[0075] Among them, v i is the eigenvalue of each position of the Bayesian eigenvector V, Represents an angle vector The second norm of , α is a predetermined parameter.

[0076] In one example, in the above-mentioned mobile Internet-based education platform interactive system 600, the classifier processing unit 670 is further used to: input the classification feature vector into the Softmax classification function of the classifier to obtain a first probability that the classification feature vector belongs to a match between the to-be-assigned question and the student and a second probability that the classification feature vector belongs to a mismatch between the to-be-assigned question and the student; and, based on a comparison between the first probability and the second probability, determine the classification result.

[0077] Here, those skilled in the art can understand that the specific functions and operations of each unit and module in the above-mentioned mobile Internet-based education platform interactive system 600 have been referred to above. Figures 1 to 4 The description of the interactive method of the mobile Internet-based educational platform has been introduced in detail, and therefore, its repeated description will be omitted.

[0078] As described above, the mobile Internet-based education platform interactive system 600 according to the embodiment of the present application can be implemented in various terminal devices, such as a server of the mobile Internet-based education platform interactive algorithm. In one example, the mobile Internet-based education platform interactive system 600 according to the embodiment of the present application can be integrated into the terminal device as a software module and / or a hardware module. For example, the mobile Internet-based education platform interactive system 600 can be a software module in the operating system of the terminal device, or it can be an application developed for the terminal device; of course, the mobile Internet-based education platform interactive system 600 can also be one of the many hardware modules of the terminal device.

[0079] Alternatively, in another example, the mobile Internet-based education platform interactive system 600 and the terminal device may also be separate devices, and the mobile Internet-based education platform interactive system 600 may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0080] Exemplary Electronic Devices

[0081] Below, reference Figure 8 To describe the electronic device according to the embodiment of the present application. Figure 8 As shown, the electronic device 10 includes one or more processors 11 and a memory 12. The processor 11 may be a central processing unit (CPU) or other processing units having data processing capability and / or instruction execution capability, and may control other components in the electronic device 10 to perform desired functions.

[0082] The memory 12 may include one or more computer program products, and the computer program product may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 11 may run the program instructions to implement the functions of the mobile Internet-based education platform interactive method of each embodiment of the present application described above and / or other desired functions. Various contents such as Bayesian feature vectors, classification feature vectors, etc. may also be stored in the computer-readable storage medium.

[0083] In one example, the electronic device 10 may further include: an input system 13 and an output system 14, and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0084] The input system 13 may include, for example, a keyboard, a mouse, and the like.

[0085] The output system 14 can output various information to the outside, including classification results, etc. The output system 14 can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, and the like.

[0086] Of course, to simplify, Figure 8 Only some of the components related to the present application in the electronic device 10 are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application situations, the electronic device 10 may also include any other appropriate components.

[0087] Exemplary computer program products and computer-readable storage media

[0088] In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps in the functions of the mobile Internet-based education platform interaction method according to various embodiments of the present application described in the above-mentioned "Exemplary Method" section of this specification.

[0089] The computer program product may be written in any combination of one or more programming languages ​​to write program codes for performing the operations of the embodiments of the present application, including object-oriented programming languages, such as Java, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0090] In addition, an embodiment of the present application may also be a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the steps of the mobile Internet-based educational platform interaction method described in the above "Exemplary Method" section of this specification.

[0091] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, system or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable 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.

[0092] The basic principles of the present application are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present application are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present application. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, not for limitation, and the above details do not limit the present application to being implemented by adopting the above specific details.

[0093] The block diagrams of the devices, systems, equipment, and systems involved in this application are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, systems, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.

[0094] It should also be noted that in the system, device and method of the present application, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0095] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0096] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. An interactive method for an education platform based on mobile Internet, characterized in that: include: Get the text data of the questions to be assigned and the questions that the students have previously answered; Using a semantic understanding model for question understanding, respectively obtaining a first semantic feature vector and a second semantic feature vector from the text data of the question to be assigned and the text data of the question that the student has previously answered; Acquire the portrait data of the student and obtain a third semantic feature vector corresponding to the portrait data through the semantic understanding model; Adjusting the first semantic feature vector, the second semantic feature vector and the third semantic feature vector to have the same length and uniformly mapping them into a probability space to obtain a first probabilistic semantic feature vector, a second probabilistic semantic feature vector and a third probabilistic semantic feature vector; Taking the feature value of each position in the third probabilistic semantic feature vector as the prior probability, the feature value of each position in the first probabilistic semantic feature vector as the event probability and the feature value of each position in the second probabilistic semantic feature vector as the evidence probability, the posterior probability of each position in the third probabilistic semantic feature vector is calculated based on the Bayesian algorithm to obtain a Bayesian feature vector, wherein the posterior probability is equal to the prior probability multiplied by the event probability divided by the evidence probability; Performing maximum conditional likelihood estimation on the Bayesian feature vector to obtain a classification feature vector composed of a plurality of maximum conditional natural estimation values, wherein the maximum conditional likelihood estimation is performed based on a cosine distance between feature values ​​at each position in the first probabilized semantic feature vector and the second probabilized semantic feature vector; Passing the classification feature vector through a classifier to obtain a classification result, wherein the classification result is used to represent a matching result between the to-be-assigned topic and the student; and Based on the classification results, questions are assigned to the students on a mobile Internet-based education platform.

2. The interactive method for an education platform based on mobile Internet according to claim 1, wherein: Using a semantic understanding model for question understanding, respectively obtaining a first semantic feature vector and a second semantic feature vector from the text data of the question to be assigned and the text data of the question that the student has previously answered, including: Performing word segmentation processing on the text data of the questions to be assigned and the text data of the questions that the student has previously answered respectively; Pass the text data of the to-be-assigned questions after word segmentation processing and the text data of the questions that the student has previously answered after word segmentation processing through a word embedding model to obtain a first word vector sequence and a second word vector sequence; Passing the first word vector sequence and the second word vector sequence through the Bert model respectively to obtain a first word feature vector sequence and a second word feature vector sequence; and The first word feature vector sequence and the second word feature vector sequence are respectively passed through a bidirectional LSTM model to obtain the first semantic feature vector and the second semantic feature vector.

3. The interactive method of the education platform based on mobile Internet according to claim 1, wherein: The first semantic feature vector, the second semantic feature vector, and the third semantic feature vector are adjusted to have the same length and uniformly mapped to a probability space to obtain a first probabilistic semantic feature vector, a second probabilistic semantic feature vector, and the third probabilistic semantic feature vector, including: adjusting the first semantic feature vector, the second semantic feature vector, and the third semantic feature vector to have the same length through linear transformation; and The first semantic feature vector, the second semantic feature vector and the third semantic feature vector, which are respectively adjusted to the same length, are input into a Softmax-like function to map the feature values ​​of each position in the first semantic feature vector, the second semantic feature vector and the third semantic feature vector to a probability interval of 0 to 1, so as to obtain the first probabilistic semantic feature vector, the second probabilistic semantic feature vector and the third probabilistic semantic feature vector, wherein the Softmax-like function is y=exp(xi) / ∑ i exp(xi).

4. The interactive method for an education platform based on mobile Internet according to claim 1, wherein: The first semantic feature vector, the second semantic feature vector, and the third semantic feature vector are adjusted to have the same length and uniformly mapped to a probability space to obtain a first probabilistic semantic feature vector, a second probabilistic semantic feature vector, and the third probabilistic semantic feature vector, including: adjusting the first semantic feature vector, the second semantic feature vector, and the third semantic feature vector to have the same length through linear transformation; and The first semantic feature vector, the second semantic feature vector and the third semantic feature vector, which are respectively adjusted to the same length, are input into the Sigmoid function to map the feature values ​​at each position in the first semantic feature vector, the second semantic feature vector and the third semantic feature vector to a probability interval of 0 to 1, so as to obtain the first probabilistic semantic feature vector, the second probabilistic semantic feature vector and the third probabilistic semantic feature vector.

5. The interactive method of the education platform based on mobile Internet according to claim 1, wherein: Performing maximum conditional likelihood estimation on the Bayesian feature vector to obtain a classification feature vector composed of multiple maximum conditional natural estimation values, including: Calculating the cosine distance between the feature values ​​at each position in the first probabilistic semantic feature vector and the second probabilistic semantic feature vector to obtain an angle vector composed of a plurality of cosine distances; Based on the angle vector, the maximum conditional likelihood estimate corresponding to the feature value at each position in the Bayesian feature vector is calculated to obtain the classification feature vector.

6. The interactive method of the education platform based on mobile Internet according to claim 5, wherein: Based on the angle vector, calculating the maximum conditional likelihood estimate corresponding to the eigenvalues ​​at each position in the Bayesian eigenvector to obtain the classification eigenvector, including: Based on the angle vector, the maximum conditional likelihood estimate corresponding to the eigenvalue at each position in the Bayesian eigenvector is calculated using the following formula to obtain the classification eigenvector; Wherein, the formula is: ， in, is the Bayesian feature vector The eigenvalues ​​at each position of Represents an angle vector The second norm of is a predetermined parameter.

7. The interactive method of the education platform based on mobile Internet according to claim 1, wherein: Passing the classification feature vector through a classifier to obtain a classification result includes: Inputting the classification feature vector into the Softmax classification function of the classifier to obtain a first probability that the classification feature vector belongs to a match between the to-be-assigned topic and the student and a second probability that the classification feature vector belongs to a mismatch between the to-be-assigned topic and the student; and The classification result is determined based on a comparison between the first probability and the second probability.

8. The mobile Internet-based educational platform interactive method according to claim 7, wherein: Based on the classification result, assigning questions to the students on a mobile Internet-based education platform includes: In response to the classification result being a match between the topic to be assigned and the student, the topic to be classified is assigned to the student via the mobile Internet-based education platform.

9. An interactive education platform system based on mobile Internet, characterized in that: include: A text data acquisition unit, used to acquire text data of questions to be assigned and questions that students have previously answered; a feature vector generating unit, configured to obtain a first semantic feature vector and a second semantic feature vector respectively from the text data of the to-be-assigned question obtained by the text data obtaining unit and the text data of the question that the student has previously answered obtained by the text data obtaining unit, using a semantic understanding model for question understanding; A portrait data feature vector generating unit, used for acquiring the portrait data of the student and obtaining a third semantic feature vector corresponding to the portrait data through the semantic understanding model; A probability mapping unit, used for adjusting the first semantic feature vector obtained by the feature vector generating unit, the second semantic feature vector obtained by the feature vector generating unit, and the third semantic feature vector obtained by the portrait data feature vector generating unit to the same length and uniformly mapping them into a probability space to obtain a first probabilistic semantic feature vector, a second probabilistic semantic feature vector, and a third probabilistic semantic feature vector; A Bayesian unit, configured to calculate the posterior probability of each position in the third probabilistic semantic feature vector obtained by the probability mapping unit based on a Bayesian algorithm, using the feature value of each position in the third probabilistic semantic feature vector obtained by the probability mapping unit as a prior probability, the feature value of each position in the first probabilistic semantic feature vector obtained by the probability mapping unit as an event probability, and the feature value of each position in the second probabilistic semantic feature vector obtained by the probability mapping unit as an evidence probability, to obtain a Bayesian feature vector, wherein the posterior probability is equal to the prior probability multiplied by the event probability divided by the evidence probability; a maximum conditional likelihood estimation unit, configured to perform maximum conditional likelihood estimation on the Bayesian feature vector obtained by the Bayesian unit to obtain a classification feature vector composed of a plurality of maximum conditional natural estimation values, wherein the maximum conditional likelihood estimation is performed based on a cosine distance between feature values ​​at each position in the first probabilized semantic feature vector and the second probabilized semantic feature vector; a classifier processing unit, configured to pass the classification feature vector obtained by the maximum conditional likelihood estimation unit through a classifier to obtain a classification result, wherein the classification result is used to represent a matching result between the to-be-assigned topic and the student; and An allocating unit is used to allocate questions to the students on a mobile Internet-based education platform based on the classification result obtained by the classifier processing unit.

10. An electronic device, comprising: processor; as well as A memory, in which computer program instructions are stored, and when the computer program instructions are executed by the processor, the processor executes the mobile Internet-based education platform interactive method as described in any one of claims 1-8.

Images