A method for monitoring chat content

Abstract

The present disclosure provides a method for monitoring chat content. By extracting the fusion feature vectors between user chat contents and comparing the fusion feature vectors with the knowledge base feature vectors, it is determined whether the user's chat content is related to learning, and then the user's chat quality is evaluated. When the user's chat quality is less than a preset threshold, it is determined that the user's chat content has a poor correlation with learning, and it is determined that the user is chatting idly at this time, and a warning message is sent, realizing real-time monitoring of the user's learning state, improving the accuracy of monitoring the user's learning state, and being able to timely remind the user and the guardian, thereby improving the user's learning efficiency.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular, to a method for monitoring chat content. Background Art

[0002] In recent years, with the continuous advancement and development of computer technology and educational informatization, computer and artificial intelligence technologies have been gradually applied to various daily educational teaching activities. More and more intelligent devices have been gradually applied to teaching scenarios, and students can learn through intelligent devices to provide more convenient learning services for students. However, during the process of students using intelligent devices for learning, there may be situations of ineffective learning such as chatting with classmates, which is not conducive to students' learning.

[0003] Currently, in offline teaching scenarios, relevant information of students is usually obtained by collecting students' images or audio, and then it is determined whether students are learning. However, this monitoring method is not applicable to the teaching scenario of intelligent devices. When students chat using intelligent devices, the monitoring device cannot collect students' audio information, and the image information during chatting may not be very different from the image information during students' learning, so it is impossible to accurately identify the learning status of students. Therefore, a student chat monitoring scheme for intelligent teaching scenarios is needed. Summary of the Invention

[0004] The present disclosure provides a method for monitoring chat content.

[0005] According to a first aspect of the present disclosure, there is provided a method for monitoring chat content, the method including: in response to a chat start request and a chat termination instruction initiated by a first user to a second user, determining chat information between the first user and the second user, and a fusion feature vector of one or more chat contents included in the chat information; based on the fusion feature vector, using a preset recognition model, determining a knowledge base feature vector corresponding to the fusion feature vector in preset knowledge base information, so as to determine a first similarity between the chat content and the preset knowledge base information based on the fusion feature vector, the knowledge base feature vector, and the standard deviation of a preset feature space; determining chat contents with the first similarity greater than a preset similarity threshold as chat contents belonging to the knowledge base information, so as to determine a chat quality evaluation parameter of the first user based on the first number of chat contents belonging to the preset knowledge base information, the total number of chat contents, the first similarity, and the fusion feature vector; when the chat quality evaluation parameter is less than a preset chat warning threshold, sending a warning message to the guardian terminal of the first user, so that the guardian of the first user determines the learning status of the first user.

[0006] In some embodiments of the present disclosure, determining the fusion feature vector of one or more chat contents included in the chat information includes: using a preset feature editor to extract the semantic feature vector and / or formula feature vector in the chat content; performing weighted summation on the semantic feature vector and / or formula feature vector to determine the fusion feature vector of the chat content.

[0007] In some embodiments of the present disclosure, the method further includes: parsing the preset knowledge base information to obtain a training data set, where the training data set includes at least one of the following: knowledge base text segments, formula information with a syntax graph structure, and chart images with coordinate annotations; randomly replacing the nodes in the syntax graph structure to obtain replacement formula information, and randomly changing the positions of the coordinate annotations to obtain disrupted chart images; using the knowledge base text segments, formula information, and chart images as positive training samples, and using the replacement formula information and disrupted chart images as negative training samples to perform first training on the to-be-trained recognition model to obtain a first trained recognition model; performing synonym replacement and / or voice conversion on the knowledge base text segments to obtain replacement text segments; and randomly discarding the nodes in the syntax graph structure to obtain defective formula information; based on the replacement text segments, defective formula information, and disrupted chart images, performing second training on the first trained recognition model to obtain a preset recognition model.

[0008] In some embodiments of the present disclosure, determining the first similarity between the chat content and the preset knowledge base information based on the fusion feature vector, the knowledge base feature vector, and the standard deviation of the preset feature space includes: based on the fusion feature vector, the knowledge base feature vector, and the standard deviation of the preset feature space, combining the following formula 1 to determine the first similarity.

[0009] Formula 1,

[0010] where, represents the first similarity, represents the fusion feature vector, represents the knowledge base feature vector, represents the standard deviation of the preset feature space.

[0011] In some embodiments of the present disclosure, determining the chat quality evaluation parameter of the first user based on the number of the first messages of the chat content belonging to the preset knowledge base information, the total number of the chat content, the first similarity, and the fusion feature vector includes: determining a dependence parameter of the chat information and the preset knowledge base information based on the number of the first messages of the chat content belonging to the preset knowledge base information, the total number of the chat content, and the first similarity, where the dependence parameter is used to indicate the degree of association between the chat information and the preset knowledge base information; determining the semantic jump distance of adjacent chat content based on the fusion feature vector in the adjacent chat content, so as to determine the thinking transition parameter of the chat information based on the semantic jump distance and the preset time decay weight, where the thinking transition parameter is used to indicate the logical relationship between the chat contents; determining a dynamic penalty parameter based on the number of the first messages, the total number, and the threshold of the number of messages of the chat content that do not belong to the preset knowledge base content, where the dynamic penalty parameter is used to indicate the degree of irrelevance between the chat information and the preset knowledge base information; and performing weighted summation on the dependence parameter, the semantic jump distance, and the dynamic penalty parameter to obtain the chat quality evaluation parameter of the first user.

[0012] In some embodiments of the present disclosure, determining a dependence parameter of the chat information and the preset knowledge base information based on the number of the first messages of the chat content belonging to the preset knowledge base information, the total number of the chat content, and the first similarity includes: determining the dependence parameter in combination with the following formula 2 based on the number of the first messages of the chat content belonging to the preset knowledge base information, the total number of the chat content, and the first similarity:

[0013] Formula 2

[0014] where DI represents the dependence parameter, K represents the number of the first messages of the chat content belonging to the preset knowledge base information, N represents the total number of the chat content, represents the first similarity of the i-th chat content.

[0015] In some embodiments of the present disclosure, determining the semantic jump distance of adjacent chat content based on the fusion feature vector in the adjacent chat content, so as to determine the thinking transition parameter of the chat information based on the semantic jump distance and the preset time decay weight includes: determining the semantic jump distance of adjacent chat content in combination with the following formula 3 based on the fusion feature vector in the adjacent chat content:

[0016] Formula 3

[0017] where represents the fusion feature vector of the j-th chat content, represents the j-th chat content and the (j - 1)-th chat content The semantic jump distance therebetween; based on the semantic jump distance and a preset time decay weight, in combination with the following formula 4, determine the thinking transition parameter of the chat message:

[0018] Formula 4

[0019] wherein, CLS represents the thinking transition parameter represents the mean value of the semantic jump distance in the chat message of represents the semantic jump distance in the chat message , represents the preset time decay weight.

[0020] In some embodiments of the present disclosure, determining the dynamic penalty parameter based on the first number, the total number, and the threshold number of the chat content that does not belong to the preset knowledge base content includes: determining the second number of the chat content that does not belong to the preset knowledge base in the chat message based on the first number and the total number; based on the second number, the total number, and the threshold number of the chat content that does not belong to the preset knowledge base content, in combination with the following formula 5, determine the dynamic penalty parameter:

[0021] Formula 5

[0022] wherein represents the dynamic penalty parameter represents the second number represents the threshold number.

[0023] In some embodiments of the present disclosure, the method further includes: determining the ratio between the learning duration corresponding to the chat content belonging to the preset knowledge base information and the chat duration corresponding to the chat message as the effective learning parameter; when the effective learning parameter is less than the preset learning efficiency threshold, determining the learning duration as the warning information.

[0024] In some embodiments of the present disclosure, the warning information further includes at least one of the following: the chat start request of the first user, the chat termination instruction of the first user, the chat start time corresponding to the chat start request, and the chat termination time corresponding to the chat termination instruction.

[0025] In summary, a chat content monitoring method proposed by the present disclosure includes: in response to a chat start request and a chat termination instruction initiated by a first user to a second user, determining the chat information between the first user and the second user, and the fusion feature vector of one or more chat contents included in the chat information; based on the fusion feature vector, using a preset recognition model to determine the knowledge base feature vector corresponding to the fusion feature vector in the preset knowledge base information, so as to determine the first similarity between the chat content and the preset knowledge base information based on the fusion feature vector, the knowledge base feature vector, and the standard deviation of the preset feature space; determining the chat content with the first similarity greater than the preset similarity threshold as the chat content belonging to the knowledge base information, so as to determine the chat quality evaluation parameter of the first user based on the first number of the chat content belonging to the preset knowledge base information, the total number of chat contents, the first similarity, and the fusion feature vector; when the chat quality evaluation parameter is less than the preset chat warning threshold, sending a warning message to the guardian terminal of the first user, so that the guardian of the first user can determine the learning state of the first user. The method of the present disclosure extracts the fusion feature vector between user chat contents and compares it with the knowledge base feature vector, thereby determining whether the user's chat content is related to learning, and then evaluating the user's chat quality. When the user's chat quality is less than the preset threshold, it is determined that the user's chat content has a poor correlation with learning, and it is determined that the user is chatting at this time, and a warning message is sent, realizing the real-time monitoring of the user's learning state, improving the accuracy of monitoring the user's learning state, and being able to timely remind the user and the guardian, improving the user's learning efficiency.

[0026] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The drawings are used to better understand the solution and do not constitute a limitation to the present disclosure. Among them:

[0028] Figure 1 It is a flowchart of a chat content monitoring method provided by an embodiment of the present disclosure;

[0029] Figure 2 It is a flowchart of another chat content monitoring method provided by an embodiment of the present disclosure;

[0030] Figure 3 It is a flowchart of another chat content monitoring method provided by an embodiment of the present disclosure;

[0031] Figure 4 It is a flowchart of another chat content monitoring method provided by an embodiment of the present disclosure;

[0032] Figure 5 A flowchart of another chat content monitoring method provided by an embodiment of the present disclosure;

[0033] Figure 6 A structural schematic diagram of a chat content monitoring device provided by an embodiment of the present disclosure;

[0034] Figure 7 A hardware structural schematic diagram of an electronic device provided by an embodiment of the present disclosure. Detailed implementation manners

[0035] The following describes exemplary embodiments of the present disclosure with reference to the accompanying drawings. Various details of the embodiments of the present disclosure are included to assist in understanding, and they should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for clarity and conciseness, descriptions of well-known functions and structures are omitted below.

[0036] In recent years, with the continuous advancement and development of computer technology and educational informatization, computer and artificial intelligence technologies have been gradually applied to various daily educational teaching activities. More and more intelligent devices have been gradually applied to teaching scenarios, and students can learn through intelligent devices to provide more convenient learning services for students. However, during the process of students using intelligent devices for learning, there may be situations of ineffective learning such as chatting with classmates, which is not conducive to students' learning.

[0037] Currently, in the offline teaching scenario, relevant information of students is usually obtained by collecting students' images or audio, and then it is determined whether students are learning. However, this monitoring method is not applicable to the intelligent device teaching scenario. When students use intelligent devices to chat, the monitoring device cannot collect students' audio information, and the image information during chatting may not be very different from the image information during students' learning, so it is impossible to accurately identify the learning state of students. Therefore, a student chat monitoring scheme for intelligent teaching scenarios is needed.

[0038] To solve the problems in the related art, a chat content monitoring method proposed by the present disclosure extracts the fusion feature vectors between user chat contents and compares the fusion feature vectors with the knowledge base feature vectors, thereby determining whether the user's chat content is related to learning, and then evaluating the chat quality of the user. When the chat quality of the user is less than a preset threshold, it is determined that the user's chat content has a poor correlation with learning, it is determined that the user is chatting at this time, and a warning message is sent, realizing real-time monitoring of the user's learning state, improving the accuracy of monitoring the user's learning state, and being able to timely remind the user and the guardian, improving the user's learning efficiency.

[0039] The following describes a method for monitoring chat content according to an embodiment of the present disclosure with reference to the accompanying drawings.

[0040] Figure 1 It is a schematic flowchart of a method for monitoring chat content provided by an embodiment of the present disclosure. As Figure 1 shown, the method includes:

[0041] Step 101: In response to a chat start request and a chat termination instruction initiated by a first user to a second user, determine the chat information between the first user and the second user, and the fusion feature vector of one or more chat contents included in the chat information.

[0042] In some embodiments, the first user may initiate a chat request to the second user through devices such as a smart terminal or a smart teaching system, but is not limited thereto. The present disclosure does not limit the manner in which the first user and the second user implement chatting.

[0043] In some embodiments, the first user is the user who initiates the chat request, and the second user is the user who accepts the chat request of the first user. In other words, the second user is the user who chats with the first user.

[0044] In some embodiments, the chat start request may be a learning help request initiated by the first user to the second user, that is, the first user sends a chat message to the second user to request the second user to help solve a learning problem; it may also be a learning material sharing request initiated by the first user to the second user, that is, the first user sends a message to the second user to request the second user to share learning materials for the first user, but is not limited thereto. The present disclosure does not limit the specific manner of the learning help request.

[0045] In some embodiments, the chat termination instruction may be an instruction triggered when the first user or the second user clicks the button to exit the chat interface, or may be an instruction triggered when the first user or the second user sends a message indicating that the problem has been solved / the learning material has been sent, etc., but is not limited thereto. The present disclosure does not limit the specific form of the chat termination instruction.

[0046] In some embodiments, the chat content included between the chat start request and the chat termination instruction may be determined as the chat information between the first user and the second user. In other words, the chat information is all the chat messages between the first user and the second user within a complete chat start and end cycle.

[0047] In some embodiments, one or more chat contents included in the chat information mean that there are multiple chat messages between the first user and the second user between the chat start request and the chat termination instruction, and one chat content is one chat message between the first user and the second user.

[0048] In some embodiments, a preset feature editor may be used to extract semantic feature vectors and / or formula feature vectors from chat content, and then by fusing the semantic feature vectors and / or formula feature vectors, a fused feature vector may be determined. For specific reference, see Figure 2 the embodiments shown, which will not be elaborated here.

[0049] In some embodiments, the fused feature vector may be used to indicate the information contained in each piece of chat content, such as the knowledge points, formulas, etc.

[0050] Step 102: Based on the fused feature vector, use a preset recognition model to determine the knowledge base feature vector corresponding to the fused feature vector in the preset knowledge base information, so as to determine the first similarity between the chat content and the preset knowledge base information based on the fused feature vector, the knowledge base feature vector, and the standard deviation of the preset feature space.

[0051] In some embodiments, the preset knowledge base information may include textbook texts, formula information, chart information, etc. The present disclosure does not limit this, and it should be understood that the preset knowledge base information may be increased or deleted as the samples such as textbooks are changed.

[0052] In some embodiments, the preset recognition model may extract relevant features from the preset knowledge base information to generate multiple knowledge base feature vectors, and then by comparing the fused feature vector with the multiple knowledge base feature vectors, determine the knowledge base feature vector that contains the same or similar content as the fused feature vector. Among them, a training method for a preset recognition model can be seen in Figure 3 the embodiments shown, which will not be elaborated here.

[0053] In some embodiments, the first similarity between the fused feature vector and the corresponding knowledge base feature vector may be determined using Formula 1 based on the fused feature vector, the knowledge base feature vector, and the standard deviation of the preset feature space, so as to judge the similarity between the chat content and the knowledge base information:

[0054] Formula 1

[0055] wherein, represents the first similarity, represents the fused feature vector, represents the knowledge base feature vector, represents the standard deviation of the preset feature space.

[0056] Step 103: Determine the chat content with the first similarity greater than the preset similarity threshold as the chat content belonging to the knowledge base information, so as to determine the chat quality evaluation parameter of the first user based on the first number of the chat content belonging to the preset knowledge base information, the total number of the chat content, the first similarity, and the fusion feature vector.

[0057] In some embodiments, when the first similarity is greater than the preset similarity threshold, it can be considered that the chat content corresponding to the first similarity is related to the preset knowledge base information, that is, the chat content is related to learning.

[0058] In some embodiments, the preset similarity threshold can be dynamic, and the value of the preset similarity threshold can be related to the training of the preset recognition model. Specifically, the value of the preset similarity threshold can be determined by the following formula:

[0059]

[0060] where represents the preset initial threshold, η represents the maximum threshold adjustment amplitude, k represents the adjustment rate of the model, t represents the current training round of the model, and t0 represents the starting round when the model triggers adjustment.

[0061] In some embodiments, the dependence parameter, the thinking transition parameter, and the dynamic penalty parameter of the first user can be determined based on the first number of the chat content belonging to the preset knowledge base information, the total number of the chat content, the first similarity, and the fusion feature vector, so as to perform a weighted sum on the dependence parameter, the semantic jump distance, and the dynamic penalty parameter to obtain the chat quality evaluation parameter of the first user.

[0062] In other words, in some embodiments, the degree of association between the chat information between the first user and the second user and the preset knowledge base information can be determined according to the first number of the chat content, the total number of the chat content, the first similarity, and the fusion feature vector, so as to determine whether the chat information between the first user and the second user is related to learning.

[0063] In other words, the chat quality evaluation parameter can indicate whether the chat information between the first user and the second user is related to learning.

[0064] Specifically, in some embodiments, the dependence parameter of the chat information and the preset knowledge base information can be determined based on the first number of the chat content belonging to the preset knowledge base information, the total number of the chat content, and the first similarity, and the dependence parameter is used to indicate the degree of association between the chat information and the preset knowledge base information.

[0065] Specifically, in some embodiments, the semantic jump distance of adjacent chat content may be determined based on the fusion feature vectors in the adjacent chat content, so as to determine the thinking transition parameter of the chat information based on the semantic jump distance and the preset time decay weight, and the thinking transition parameter is used to indicate the logical relationship between chat contents.

[0066] Specifically, in some embodiments, the dynamic penalty parameter may be determined based on the first number, the total number, and the threshold number of chat contents that do not belong to the preset knowledge base content, and the dynamic penalty parameter is used to indicate the degree of irrelevance between the chat information and the preset knowledge base information.

[0067] Step 104: When the chat quality evaluation parameter is less than the preset chat warning threshold, send a warning message to the guardian terminal of the first user, so that the guardian of the first user can determine the learning status of the first user.

[0068] In some embodiments, the larger the chat quality evaluation parameter is, the more relevant the chat information between the first user and the second user is to the preset knowledge base information, that is, the more relevant the chat information between the first user and the second user is to learning.

[0069] In some embodiments, when the chat quality evaluation parameter is less than the preset chat warning threshold, it indicates that the relevance between the chat information between the first user and the second user and the preset knowledge base information is poor, that is, the chat information between the first user and the second user may be irrelevant to learning. At this time, a warning message may be sent to the guardian terminal of the first user, so that the guardian of the first user can determine the learning status of the first user.

[0070] It should be understood that in some alternative embodiments, a warning message may also be sent to the guardian terminal of the second user who chats with the first user, so that the guardian of the second user can determine the learning status of the second user. A warning reminder may also be sent to the first user and / or the second user to prompt the first user and / or the second user to change the learning status and reduce the number of chats unrelated to learning.

[0071] In some embodiments, the warning message may include: the effective learning duration of the first user and / or the second user, where the effective learning duration may be the learning duration corresponding to the chat content belonging to the preset knowledge base information.

[0072] In some alternative embodiments, the ratio between the learning duration corresponding to the chat content belonging to the preset knowledge base information and the chat duration corresponding to the chat information may also be determined as the effective learning parameter; when the effective learning parameter is less than the preset learning efficiency threshold, the learning duration is determined as the warning message.

[0073] In some embodiments, the warning information further includes at least one of the following: a chat start request of the first user, a chat termination instruction of the first user, a chat start time corresponding to the chat start request, and a chat termination time corresponding to the chat termination instruction.

[0074] It should be understood that the above warning information is only an example, and the warning information may further include other information, such as the learning subjects of the first user, the number of chat start requests initiated by the first user within a preset time period, etc., which can be used to indicate the user's learning behavior or learning habits.

[0075] In summary, the present disclosure provides a chat content monitoring method, including: in response to a chat start request and a chat termination instruction initiated by a first user to a second user, determining chat information between the first user and the second user, and a fusion feature vector of one or more chat contents included in the chat information; based on the fusion feature vector, using a preset recognition model, determining a knowledge base feature vector corresponding to the fusion feature vector in the preset knowledge base information, so as to determine a first similarity between the chat content and the preset knowledge base information based on the fusion feature vector, the knowledge base feature vector, and the standard deviation of the preset feature space; determining the chat content with the first similarity greater than the preset similarity threshold as the chat content belonging to the knowledge base information, so as to determine a chat quality evaluation parameter of the first user based on the first number of the chat content belonging to the preset knowledge base information, the total number of the chat contents, the first similarity, and the fusion feature vector; when the chat quality evaluation parameter is less than the preset chat warning threshold, sending warning information to the guardian terminal of the first user, so that the guardian of the first user can determine the learning state of the first user. The method of the present disclosure extracts the fusion feature vector between user chat contents and compares it with the knowledge base feature vector, thereby determining whether the user's chat content is related to learning, further evaluating the user's chat quality, and when the user's chat quality is less than the preset threshold, determining that the user's chat content has a poor correlation with learning, determining that the user is chatting at this time, and sending a warning information, realizing real-time monitoring of the user's learning state, improving the accuracy of monitoring the user's learning state, and being able to timely remind the user and the guardian, improving the user's learning efficiency.

[0076] As a possible implementation, as Figure 2 shown in the flowchart of another chat content monitoring method, on the basis of the above embodiments, further explaining the determination of the knowledge base feature vector corresponding to the fusion feature vector in the preset knowledge base information by using the preset recognition model, including the following steps:

[0077] Step 201: Using a preset feature editor, extracting a semantic feature vector and / or a formula feature vector from the chat content.

[0078] In some embodiments, the text in the chat content can be subjected to knowledge distillation through a preset feature editor, so as to extract the semantic feature vector in the chat content.

[0079] In some embodiments, the formulas involved in the chat content can be subjected to structure parsing through a preset feature editor (for example, parsing the operation symbols, variable values, etc. used in the formulas), so as to extract the formula feature vector in the chat content.

[0080] In some embodiments, the preset feature editor is, for example, a Bidirectional Encoder Representation from Transformers (BERT), or a variant feature editor of BERT, etc., and the present disclosure does not limit this.

[0081] It should be understood that when extracting the feature vector from the chat content, only the semantic feature vector can be extracted, or only the formula feature vector can be extracted, or both the semantic feature vector and the formula feature vector can be extracted simultaneously.

[0082] For example, when the chat content only contains text information, only the semantic feature vector can be extracted; when the chat content only contains formula information, only the formula feature vector can be extracted; when the chat content contains both text information and formula information, both the semantic feature vector and the formula feature vector can be extracted simultaneously.

[0083] Exemplarily, when the chat content is: The Pythagorean theorem formula is , then the semantic feature vector can be extracted according to "Pythagorean theorem", and the formula feature vector can be extracted according to " ".

[0084] Step 202: Perform weighted summation on the semantic feature vector and / or the formula feature vector to determine the fused feature vector of the chat content.

[0085] In some embodiments, the semantic feature vector and the formula feature vector can be subjected to weighted summation to achieve feature fusion, so as to determine the fused feature vector of the chat content, where the weights corresponding to the semantic feature vector and the formula feature vector are not limited.

[0086] For example, with the formula , the fused feature vector is determined, where represents the fused feature vector, represents the semantic feature vector, and represents the formula feature vector.

[0087] In some embodiments, when only semantic feature vectors are extracted from the chat content, the semantic feature vectors can be directly determined as the fused feature vectors; when only formula feature vectors are extracted from the chat content, the formula feature vectors can be directly determined as the fused feature vectors.

[0088] In summary, the present disclosure proposes a chat content monitoring method, including: using a preset feature editor to extract semantic feature vectors and / or formula feature vectors from the chat content; performing weighted summation on the semantic feature vectors and / or formula feature vectors to determine the fused feature vectors of the chat content. By extracting semantic feature vectors and / or formula feature vectors from the chat content and generating fused feature vectors, the present disclosure improves the accuracy of feature extraction determination, and further improves the accuracy of determining whether the chat content belongs to the knowledge base information.

[0089] As a possible implementation, as Figure 3 shown in the flowchart of another chat content monitoring method, on the basis of the above embodiments, a method for obtaining a preset recognition model is shown, including the following steps:

[0090] Step 301: Parse the preset knowledge base information to obtain a training data set.

[0091] In some embodiments, according to information such as chapters in the textbook, the text information in the preset knowledge base can be parsed into knowledge base text segments. Further, the text segment can be a tag sequence with specific hierarchical markings.

[0092] In some embodiments, the formulas in the preset knowledge base can be parsed into an encoding format (such as LaTeX format), and then formula information with a specific graph structure can be generated according to the syntax corresponding to the encoding. The node set in the graph structure can be operators, operands, special symbols, etc. in the formula, and the edge set in the graph structure can be the dependency relationship between the corresponding encoding syntaxes in the formula.

[0093] In some embodiments, coordinate annotation can be performed on the charts in the preset knowledge base, so as to parse the image into a chart image with coordinate annotations.

[0094] In other words, the preset knowledge base information can be parsed, and the parsing result can be determined as the training data set. In other words, the training data set can include at least one of knowledge base text segments, formula information with a syntax graph structure, and chart images with coordinate annotations.

[0095] Step 302: Randomly replace the nodes in the syntax graph structure to obtain replacement formula information, and randomly change the positions of the coordinate annotations to obtain a disrupted chart image.

[0096] In some embodiments, to improve the recognition accuracy of a preset recognition model, the data in the training dataset can be divided into positive samples and negative samples, so that the recognition model to be trained can learn better correct sample information during training and improve the model learning efficiency.

[0097] In some embodiments, nodes in the syntax graph structure can be randomly replaced to replace the parsed formula information and generate incorrect formula information, that is, replace the formula information.

[0098] In some embodiments, the position of the coordinate annotation can be mechanically changed to generate an incorrect chart image, that is, disrupt the chart image.

[0099] Among them, the coordinate annotation of the chart can be changed through the following formula:

[0100]

[0101] Among them, represents disrupting the chart image, I represents the chart image, and Δ represents the displacement of the coordinate annotation.

[0102] Step 303: Use the knowledge base text segment, formula information, and chart image as positive training samples, and use the replaced formula information and disrupted chart image as negative training samples to perform the first training on the recognition model to be trained to obtain the first trained recognition model.

[0103] In some embodiments, the determined knowledge base text segment, formula information, and chart image (that is, the training dataset obtained by parsing the preset knowledge base information) are used as positive training samples, and the replaced formula information and disrupted chart image are used as negative training samples to perform the first training on the recognition model to be trained, so that the model to be trained can accelerate learning the correct preset knowledge base information.

[0104] In some embodiments, the loss function used during the first training can be as follows, but is not limited thereto:

[0105]

[0106] Among them, represents the loss function, a represents a preset anchor sample as a reference, p represents a positive training sample, and n represents a negative training sample.

[0107] Step 304: Perform synonym replacement and / or voice conversion on the knowledge base text segment to obtain a replaced text segment; and randomly discard nodes in the syntax graph structure to obtain defective formula information.

[0108] In some embodiments, to enhance the recognition ability of the preset recognition model for the preset knowledge base information and non-preset knowledge base information, the training data set can be fine-tuned.

[0109] In some embodiments, synonym replacement and / or voice conversion are performed on the knowledge base text segments to provide more rich semantic training samples.

[0110] Among them, the formula used for synonym replacement can be as follows:

[0111]

[0112] Among them, represents synonym replacement, x is the word to be replaced, w represents the synonym of x, and cos represents the similarity between x and w.

[0113] In some instances, nodes in the syntax graph structure can be randomly discarded to obtain defective formula information, thereby providing more rich formula information.

[0114] Among them, random discarding of nodes in the syntax graph structure can be achieved through the following formula:

[0115]

[0116] Among them, represents defective formula information, G represents formula information, DropNode represents the random discard function, and p represents the preset parameter in this function.

[0117] Step 305: Based on the replacement text segment, defective formula information, and scrambled chart image, perform second training on the first training recognition model to obtain the preset recognition model.

[0118] In some embodiments, the first training recognition model can be second-trained according to the replacement text segment, defective formula information, and scrambled chart image to improve the recognition ability of the preset recognition model obtained after training.

[0119] In some embodiments, the second training can include multiple stages of training. For example, training is performed separately in three stages: primary, intermediate, and advanced. Among them, the sample complexity can gradually increase by stage, so as to improve the accuracy of the preset recognition model.

[0120] In some embodiments, when performing the first training and / or the second training, the following formula can be used to perform convolutional iterative update on the defective formula information, replacement formula information, formula information, etc., so as to improve the recognition accuracy of the preset recognition model:

[0121]

[0122] Among them, represents the feature vector of node v in the k-th convolutional layer of the model in the graph structure, represents the neighbor set of node v, represents the normalization coefficient of the edge (v, u) between node v and node u in the graph structure. represents the learnable weight matrix of the k-th layer, represents the activation function.

[0123] In some embodiments, when performing the first training and / or the second training, the replacement text segment, the knowledge base text segment, etc. can be encoded through the following formula for the convenience of model learning:

[0124]

[0125] Among them, PE represents the encoding result, c represents the chapter number indicating the chapter position of the text segment (i.e., the replacement text segment and / or the knowledge base text segment) in the textbook, l represents the hierarchical depth of the text segment (for example, if the text segment belongs to a chapter, l is 1; if the text segment belongs to a section, l is 2; if the text segment belongs to a subsection, l is 3), pos represents the position of the character in the paragraph, A represents the hidden layer dimension of the model, and i represents the dimension index. The term is used to achieve the decaying attention of the model to the deep chapter structure. The deeper the hierarchy (such as a subsection), the smaller the fluctuation amplitude of the position encoding.

[0126] In summary, a chat content monitoring method proposed by the present disclosure divides training samples into positive and negative samples, and by partially modifying or replacing the samples, improves the recognition ability of the preset recognition model obtained using the training samples, and further improves the accuracy of determining whether the student's chat content is related to the knowledge base information.

[0127] As a possible implementation, as Figure 4 shown in the flowchart of another chat content monitoring method, on the basis of the above embodiments, step 103 is further explained, including the following steps:

[0128] Step 401, based on the first number of chat contents belonging to the preset knowledge base information, the total number of chat contents, and the first similarity, determine the dependence parameter of the chat information on the preset knowledge base information.

[0129] In some embodiments, based on the first number of chat contents belonging to the preset knowledge base information, the total number of chat contents, and the first similarity, the dependence parameter can be determined in combination with the following formula 2:

[0130] Formula 2,

[0131] Wherein, DI represents a dependency parameter, K represents the first number of chat contents belonging to the preset knowledge base information, and N represents the total number of chat contents. represents the first similarity of the i-th chat content.

[0132] In some embodiments, the dependency parameter is used to indicate the degree of association between the chat information and the preset knowledge base information.

[0133] Step 402: Based on the fused feature vectors in adjacent chat contents, determine the semantic jump distance of the adjacent chat contents, so as to determine the thinking transition parameter of the chat information based on the semantic jump distance and the preset time decay weight.

[0134] In some embodiments, based on the fused feature vectors in adjacent chat contents, combined with the following formula 3, determine the semantic jump distance of the adjacent chat contents:

[0135] Formula 3

[0136] Wherein, represents the fused feature vector of the j-th chat content, represents the j-th chat content and the (j - 1)-th chat content the semantic jump distance therebetween;

[0137] Furthermore, based on the semantic jump distance and the preset time decay weight, combined with the following formula 4, determine the thinking transition parameter of the chat information:

[0138] Formula 4

[0139] Wherein, CLS represents the thinking transition parameter, represents the mean value of the semantic jump distance in the chat information , represents the semantic jump distance in the chat information , represents the preset time decay weight.

[0140] In some embodiments, the thinking transition parameter is used to indicate the logical relationship between chat contents

[0141] Step 403: Based on the first number, the total number, and the threshold of the number of chat contents that do not belong to the preset knowledge base content, determine the dynamic penalty parameter.

[0142] In some embodiments, the second number of chat contents that do not belong to the preset knowledge base in the chat information can be determined by taking the difference based on the first number and the total number; furthermore, based on the second number, the total number, and the threshold of the number of chat contents that do not belong to the preset knowledge base content, combined with the following formula 5, determine the dynamic penalty parameter:

[0143] Formula 5,

[0144] wherein, represents the dynamic penalty parameter, represents the second number, represents the number threshold.

[0145] In some embodiments, the dynamic penalty parameter is used to indicate the degree of irrelevance between the chat information and the preset knowledge base information.

[0146] Step 404, perform a weighted sum on the dependency parameter, the semantic jump distance, and the dynamic penalty parameter to obtain the chat quality evaluation parameter of the first user.

[0147] In some instances, by performing a weighted sum on the dependency parameter, the semantic jump distance, and the dynamic penalty parameter, the weighted sum result is determined as the chat quality evaluation parameter of the first user.

[0148] In summary, a chat content monitoring method proposed by the present disclosure determines the dependency parameter, the semantic jump distance, and the dynamic penalty parameter, and determines the chat quality evaluation parameter according to the dependency parameter, the semantic jump distance, and the dynamic penalty parameter, so as to evaluate whether the chat content of the user is related to the preset learning knowledge base from multiple perspectives, improve the objectivity and accuracy of the chat quality evaluation parameter, and reduce the probability of false triggering of warning information.

[0149] The following is an exemplary description of a chat content monitoring method proposed by the present disclosure. As Figure 5 shown, it may include the following steps:

[0150] 1. Perform multimodal extraction on the chat information input by the user to obtain a semantic feature vector and / or a formula feature vector.

[0151] 2. Perform knowledge distillation on the knowledge base features (i.e., the above-mentioned preset knowledge base information) and cache them to improve the rate of comparing the chat information and the preset knowledge base information.

[0152] 3. Perform multimodal extraction and cross-modal fusion on the cached information to obtain a knowledge base feature vector.

[0153] 4. Perform cross-modal fusion on the semantic feature vector and / or the formula feature vector to obtain a fusion feature vector.

[0154] 5. Compare the similarity between the knowledge base feature vector and the fusion feature vector to dynamically determine whether the chat content is within the time domain knowledge base information.

[0155] Corresponding to the above-mentioned method for monitoring chat content, the present invention also provides a device for monitoring chat content. Since the device embodiments of the present invention correspond to the above-mentioned method embodiments, details not disclosed in the device embodiments may be referred to the above-mentioned method embodiments, and will not be elaborated herein.

[0156] Figure 6 As shown in the structural schematic diagram of a device for monitoring chat content provided by an embodiment of the present disclosure, Figure 6 the device includes:

[0157] A first determination unit 610, configured to determine chat information between a first user and a second user, and a fusion feature vector of one or more pieces of chat content included in the chat information, in response to a chat start request and a chat termination instruction initiated by the first user to the second user;

[0158] A second determination unit 620, configured to determine, based on the fusion feature vector and using a preset recognition model, a knowledge base feature vector corresponding to the fusion feature vector in preset knowledge base information, so as to determine a first similarity between the chat content and the preset knowledge base information based on the fusion feature vector, the knowledge base feature vector, and the standard deviation of a preset feature space;

[0159] A third determination unit 630, configured to determine chat content with a first similarity greater than a preset similarity threshold as chat content belonging to the knowledge base information, so as to determine a chat quality evaluation parameter of the first user based on the first number of pieces of chat content belonging to the preset knowledge base information, the total number of pieces of chat content, the first similarity, and the fusion feature vector;

[0160] An early warning unit 640, configured to send an early warning message to a guardian terminal of the first user when the chat quality evaluation parameter is less than a preset chat early warning threshold, so that the guardian of the first user can determine the learning state of the first user.

[0161] In some embodiments of the present disclosure, the first determination unit 610 is further configured to: extract a semantic feature vector and / or a formula feature vector from the chat content using a preset feature editor; perform weighted summation on the semantic feature vector and / or the formula feature vector to determine a fusion feature vector of the chat content.

[0162] In some embodiments of the present disclosure, the first determination unit 610 is further configured to: parse the preset knowledge base information to obtain a training data set, where the training data set includes at least one of the following: knowledge base text segments, formula information with a syntax graph structure, and chart images with coordinate annotations; randomly replace nodes in the syntax graph structure to obtain replacement formula information, and randomly change the positions of the coordinate annotations to obtain disrupted chart images; use the knowledge base text segments, formula information, and chart images as positive training samples, and use the replacement formula information and disrupted chart images as negative training samples to perform first training on the to-be-trained recognition model to obtain a first trained recognition model; perform synonym replacement and / or voice conversion on the knowledge base text segments to obtain replacement text segments; and randomly discard nodes in the syntax graph structure to obtain defective formula information; based on the replacement text segments, defective formula information, and disrupted chart images, perform second training on the first trained recognition model to obtain a preset recognition model.

[0163] In some embodiments of the present disclosure, the second determination unit 620 is further configured to: determine a first similarity based on the fusion feature vector, the knowledge base feature vector, and the standard deviation of the preset feature space, in combination with the following formula 1,

[0164] Formula 1,

[0165] where, represents the first similarity, represents the fusion feature vector, represents the knowledge base feature vector, represents the standard deviation of the preset feature space.

[0166] In some embodiments of the present disclosure, the third determination unit 630 is further configured to: determine a dependence parameter of the chat information and the preset knowledge base information based on the first number of the chat content belonging to the preset knowledge base information, the total number of the chat content, and the first similarity, where the dependence parameter is used to indicate the degree of association between the chat information and the preset knowledge base information; determine the semantic jump distance of adjacent chat content based on the fusion feature vectors in the adjacent chat content, so as to determine the thinking transition parameter of the chat information based on the semantic jump distance and the preset time decay weight, where the thinking transition parameter is used to indicate the logical relationship between the chat contents; determine a dynamic penalty parameter based on the first number, the total number, and the number threshold of the chat content that does not belong to the preset knowledge base content, where the dynamic penalty parameter is used to indicate the degree of irrelevance between the chat information and the preset knowledge base information; perform weighted summation on the dependence parameter, the semantic jump distance, and the dynamic penalty parameter to obtain the chat quality evaluation parameter of the first user.

[0167] In some embodiments of the present disclosure, the third determination unit 630 is further configured to: based on the number of the first pieces of chat content belonging to the preset knowledge base information, the total number of chat content, and the first similarity, and in combination with the following formula 2, determine a dependency parameter:

[0168] Formula 2

[0169] where DI represents the dependency parameter, K represents the number of the first pieces of chat content belonging to the preset knowledge base information, N represents the total number of chat content, represents the first similarity of the i-th piece of chat content.

[0170] In some embodiments of the present disclosure, the third determination unit 630 is further configured to: based on the number of the first pieces, the total number, and the threshold number of pieces of chat content that do not belong to the preset knowledge base content, determine a dynamic penalty parameter, including: based on the number of the first pieces and the total number, determine the number of the second pieces of chat content that do not belong to the preset knowledge base in the chat information; based on the number of the second pieces, the total number, and the threshold number of pieces of chat content that do not belong to the preset knowledge base content, and in combination with the following formula 5, determine the dynamic penalty parameter:

[0171] Formula 5

[0172] where represents the dynamic penalty parameter, represents the number of the second pieces, represents the threshold number of pieces.

[0173] In some embodiments of the present disclosure, the warning determination unit 640 is further configured to: determine the ratio between the learning duration corresponding to the chat content belonging to the preset knowledge base information and the chat duration corresponding to the chat information as an effective learning parameter; when the effective learning parameter is less than a preset learning efficiency threshold, determine the learning duration as a warning message.

[0174] In some embodiments of the present disclosure, the warning message further includes at least one of the following: a chat start request of the first user, a chat termination instruction of the first user, a chat start time corresponding to the chat start request, and a chat termination time corresponding to the chat termination instruction.

[0175] It should be noted that the foregoing explanations of the method embodiments also apply to the apparatus in this embodiment, with the same principle, and will not be further limited in this embodiment.

[0176] Based on the method as Figures 1 to 5 shown above, correspondingly, this embodiment further provides a computer program product, including a computer program, where the computer program, when executed by a processor, implements the method as Figures 1 to 5 shown above.

[0177] Based on the above method as Figures 1 to 5 shown, correspondingly, this embodiment also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method as Figures 1 to 5 shown is implemented.

[0178] Based on such an understanding, the technical solution of this application can be embodied in the form of a software product. The software product can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods in various implementation scenarios of this application.

[0179] As Figure 7 shown, the following is a schematic diagram of the hardware structure of an electronic device according to the present invention, including:

[0180] At least one processor 701; and,

[0181] A memory 702 communicatively connected to at least one of the processors 701; wherein,

[0182] The memory 702 stores instructions executable by at least one of the processors. The instructions are executed by at least one of the processors so that at least one of the processors can execute a chat content monitoring method as described above.

[0183] Figure 7 Taking one processor 701 as an example in

[0184] The electronic device may further include: an input device 703 and a display device 704.

[0185] The processor 701, the memory 702, the input device 703 and the display device 704 may be connected through a bus or other means. In the figure, the connection through a bus is taken as an example.

[0186] The memory 702, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules, such as the program instructions / modules corresponding to the review content generation method in the embodiments of this application. For example, Figures 1 to 5 the method flow shown. The processor 701 executes various functional applications and data processing by running the non-volatile software programs, instructions, and modules stored in the memory 702, that is, implements a chat content monitoring method in the above embodiments.

[0187] The memory 702 may include a program storage area and a data storage area. The program storage area may store an operating system and application programs required for at least one function. The data storage area may store data created according to the use of the review content generation method, etc. In addition, the memory 702 may include high-speed random access memory and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid-state storage devices. In some embodiments, the memory 702 may optionally include a memory remotely provided with respect to the processor 701, and these remote memories may be connected to the device executing the review content generation method through a network. Examples of the above network include but are not limited to the Internet, intranet, local area network, mobile communication network, and combinations thereof.

[0188] The input device 703 may receive input user clicks and generate signal inputs related to user settings and function controls of the review content generation method. The display device 704 may include display devices such as a display screen.

[0189] When the one or more modules are stored in the memory 702 and run by the one or more processors 701, a chat content monitoring method in any of the above method embodiments is executed.

[0190] Optionally, the above physical device may further include a user interface, a network interface, a camera, a radio frequency (RF) circuit, sensors, an audio circuit, a WI-FI module, etc. The user interface may include a display screen (Display) and an input unit such as a keyboard (Keyboard), etc. Optionally, the user interface may further include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a WI-FI interface), etc.

[0191] Those skilled in the art can understand that the above physical device structure provided in this embodiment does not limit the physical device, and it may include more or fewer components, or combine certain components, or have different component arrangements.

[0192] The storage medium may further include an operating system and a network communication module. The operating system is a program for managing the hardware and software resources of the above physical device, and supports the operation of information processing programs and other software and / or programs. The network communication module is used to implement communication between components inside the storage medium and communication between other hardware and software in the information processing physical device.

[0193] Through the description of the above embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus a necessary general hardware platform, or can be implemented by hardware.

[0194] By applying the solution of this embodiment, compared with the current existing technologies, this solution extracts the fusion feature vectors between the user's chat contents, and uses the fusion feature vectors to compare with the knowledge base feature vectors, so as to judge whether the user's chat contents are related to learning, and then evaluate the user's chat quality. When the user's chat quality is less than the preset threshold, it is determined that the user's chat contents have a poor relevance to learning, and it is determined that the user is chatting idly at this time, and a warning message is sent, realizing the real-time monitoring of the user's learning state, improving the accuracy of monitoring the user's learning state, and being able to timely remind the user and the guardian, thereby improving the user's learning efficiency.

[0195] It should be noted that in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising a..." does not exclude the existence of additional identical elements in the process, method, article or device comprising the element.

[0196] The above are only specific embodiments of the present application, enabling those skilled in the art to understand or implement the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments described herein, but rather will be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A chat content monitoring method, characterized in that: The method comprises: In response to a chat start request and a chat termination instruction initiated by a first user to a second user, determining chat information between the first user and the second user, and a fusion feature vector of one or more chat contents contained in the chat information; Based on the fused feature vector, using a preset recognition model, determining a knowledge base feature vector in the preset knowledge base information corresponding to the fused feature vector, so as to determine a first similarity between the chat content and the preset knowledge base information based on the fused feature vector, the knowledge base feature vector, and a standard deviation of a preset feature space; Determine the chat content whose first similarity is greater than a preset similarity threshold as the chat content belonging to the knowledge base information, and determine the chat quality evaluation parameter of the first user based on the first number of chat contents belonging to the preset knowledge base information, the total number of chat contents, the first similarity, and the fused feature vector; When the chat quality evaluation parameter is less than a preset chat warning threshold, sending a warning message to a guardian terminal of the first user, so that the guardian of the first user can determine the learning status of the first user; Wherein, the method further comprises: Parsing the preset knowledge base information to obtain a training data set, wherein the training data set includes at least one of the following: a knowledge base text segment, formula information with a syntax diagram structure, and a chart image with coordinate annotations; Randomly replacing nodes in the syntax graph structure to obtain replacement formula information, and randomly changing the positions of the coordinate annotations to obtain a disturbed graph image; Using the knowledge base text segment, the formula information and the chart image as positive training samples, and using the replacement formula information and the disturbed chart image as negative training samples, performing a first training on the recognition model to be trained to obtain a first training recognition model; Perform synonym replacement and / or voice conversion on the knowledge base text segment to obtain a replacement text segment; and randomly discard nodes in the grammar graph structure to obtain missing formula information; Based on the replacement text segment, the missing formula information and the disturbed chart image, performing a second training on the first training recognition model to obtain the preset recognition model; The determining of the chat quality evaluation parameter of the first user based on the first number of chat contents belonging to the preset knowledge base information, the total number of chat contents, the first similarity, and the fused feature vector includes: Determine, based on the first number of chat contents belonging to the preset knowledge base information, the total number of chat contents, and the first similarity, a dependency parameter between the chat information and the preset knowledge base information, wherein the dependency parameter is used to indicate a degree of association between the chat information and the preset knowledge base information; Determine the semantic jump distance of the adjacent chat contents based on the fused feature vectors in the adjacent chat contents, and determine the thought transition parameter of the chat information based on the semantic jump distance and a preset time decay weight, wherein the thought transition parameter is used to indicate the logical relationship between the chat contents; Determine a dynamic penalty parameter based on the first number of messages, the total number of messages, and a threshold number of messages of the chat content that do not belong to the preset knowledge base content, wherein the dynamic penalty parameter is used to indicate the degree of irrelevance of the chat information to the preset knowledge base information; A weighted sum is performed on the dependency parameter, the semantic jump distance, and the dynamic penalty parameter to obtain a chat quality evaluation parameter of the first user.

2. The method according to claim 1, characterized in that: Determining a fusion feature vector of one or more chat contents contained in the chat information includes: Using a preset feature editor, extracting semantic feature vectors and / or formula feature vectors from the chat content; A weighted sum is performed on the semantic feature vectors and / or formula feature vectors to determine a fusion feature vector of the chat content.

3. The method according to claim 1, characterized in that: The determining, based on the fused feature vector, the knowledge base feature vector, and the standard deviation of the preset feature space, the first similarity between the chat content and the preset knowledge base information comprises: Based on the fused feature vector, the knowledge base feature vector, and the standard deviation of the preset feature space, the first similarity is determined in combination with the following formula 1: Formula 1, in, represents the first similarity, represents the fused feature vector, represents the knowledge base feature vector, Represents the standard deviation of the preset feature space.

4. The method according to claim 1, characterized in that: The determining of the dependency parameter between the chat information and the preset knowledge base information based on the first number of chat contents belonging to the preset knowledge base information, the total number of chat contents, and the first similarity includes: The dependency parameter is determined based on the first number of chat contents belonging to the preset knowledge base information, the total number of chat contents, and the first similarity in combination with the following formula 2: Formula 2, Wherein, DI represents the dependency parameter, K represents the first number of chat contents belonging to the preset knowledge base information, and N represents the total number of chat contents. Indicates the first similarity of the i-th chat content.

5. The method according to claim 1, characterized in that: The determining of the semantic jump distance of the adjacent chat contents based on the fused feature vectors in the adjacent chat contents, and determining the thought transition parameter of the chat information based on the semantic jump distance and a preset time decay weight includes: Based on the fused feature vectors in the adjacent chat contents, combined with the following formula 3, the semantic jump distance of the adjacent chat contents is determined: Formula 3, in, represents the fused feature vector of the jth chat content, Indicates the jth chat content Chat content with Article j-1 The semantic jump distance between them; Based on the semantic jump distance and the preset time decay weight, combined with the following formula 4, the thought transition parameter of the chat information is determined: Formula 4, Wherein, CLS represents the thought transition parameter, Indicates the semantic jump distance in the chat message The mean of Indicates the semantic jump distance in the chat message The standard deviation of represents the preset time decay weight, K represents the first number of chat contents belonging to the preset knowledge base information, and N represents the total number of chat contents.

6. The method according to claim 1, characterized in that: The determining of the dynamic penalty parameter based on the first number of chats, the total number of chats, and the number threshold of chats that do not belong to the preset knowledge base content includes: Based on the first number and the total number, determining a second number of chat contents in the chat information that do not belong to the preset knowledge base; Based on the second number of chats, the total number of chats, and the number threshold of chats that do not belong to the preset knowledge base content, the dynamic penalty parameter is determined in combination with the following formula 5: Formula 5, in, represents the dynamic penalty parameter, the represents the second number, the represents the threshold number of chats, and N represents the total number of chat contents.

7. The method according to claim 1, characterized in that The method further comprises: Determine the ratio between the learning time corresponding to the chat content belonging to the preset knowledge base information and the chat time corresponding to the chat information as an effective learning parameter; When the effective learning parameter is less than a preset learning efficiency threshold, the learning duration is determined as warning information.

8. The method according to claim 1, characterized in that The warning information further includes at least one of the following: a chat start request of the first user, a chat termination instruction of the first user, a chat start time corresponding to the chat start request, and a chat termination time corresponding to the chat termination instruction.

Images