Knowledge point identification method, device, equipment, storage medium and program product

Abstract

The present application relates to a knowledge point identification method, device, equipment, storage medium and program product. The method comprises: obtaining a target document, determining a first knowledge point set and a second knowledge point set respectively according to the file name and content information of the target document, the first knowledge point set and the second knowledge point set respectively including at least one knowledge point; determining a third knowledge point set according to the basic information of the uploading user of the target document, the third knowledge point set including at least one knowledge point; performing cluster analysis on the target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set, and taking the knowledge point whose similarity meets the preset requirements as the target knowledge point; the target knowledge point set includes the first knowledge point set, the second knowledge point set and the third knowledge point set. The use of this method can improve the accuracy of knowledge point annotation.

Description

Technical Field

[0001] The present application relates to the field of information processing technology, and in particular to a knowledge point recognition method, device, equipment, storage medium and program product. Background Art

[0002] In the field of education, knowledge point annotation is of great significance for the management and application of teaching resources. After teachers upload teaching resources and files, they can use knowledge point annotation to identify the knowledge points and course teaching information contained in the uploaded files.

[0003] In the prior art, knowledge point annotation relies on natural language processing (NLP) technology, such as part-of-speech tagging, syntactic analysis, and semantic analysis, to understand and mark the knowledge points in the uploaded files. However, there is a problem of low accuracy when relying on natural language processing methods to annotate knowledge points. Summary of the invention

[0004] Based on this, it is necessary to provide a knowledge point identification method, device, equipment, storage medium and program product that can provide accurate knowledge point annotation in response to the above technical problems.

[0005] In a first aspect, the present application provides a knowledge point identification method, comprising:

[0006] Acquire a target document, and determine a first knowledge point set and a second knowledge point set according to the file name and content information of the target document, wherein the first knowledge point set and the second knowledge point set respectively include at least one knowledge point;

[0007] Determine a third knowledge point set according to basic information of the user who uploaded the target document, where the third knowledge point set includes at least one knowledge point;

[0008] A cluster analysis is performed on the target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set, and knowledge points whose similarity meets preset requirements are used as target knowledge points; the target knowledge point set includes a first knowledge point set, a second knowledge point set and a third knowledge point set.

[0009] In one embodiment, cluster analysis is performed on the target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set, including:

[0010] A vector space is constructed according to each knowledge point in the target knowledge point set; the similarity between any two knowledge points in the vector space is calculated to generate a similarity matrix; the similarity of each knowledge point is determined according to the DBSCAN algorithm and the similarity matrix, wherein the knowledge points whose similarity meets the preset requirements are divided into the same cluster.

[0011] In one embodiment, determining the first knowledge point set and the second knowledge point set respectively according to the file name and content information of the target document includes:

[0012] The first knowledge point set is determined according to the file name of the target document and a preset dictionary; and the second knowledge point set is determined according to the content information of the target document and a preset recognition model.

[0013] In one embodiment, determining the second knowledge point set according to the content information of the target document and a preset recognition model includes:

[0014] The content information is segmented; the data obtained after the word segmentation is input into a bidirectional encoder for feature extraction to obtain multiple word vectors; the word vectors are input into a preset recognition model, and a second knowledge point set is determined according to the output of the recognition model, wherein the preset recognition model includes a bidirectional long short-term memory network, and / or a conditional random field.

[0015] In one embodiment, word segmentation processing is performed on the content information, including:

[0016] The content information is cleaned to remove noise data in the content information; the cleaned content information is input into a hidden Markov model, and the content information is segmented according to the hidden Markov model.

[0017] In one embodiment, the knowledge point identification method further includes:

[0018] Get the preset dictionary, which includes subjects, knowledge points and course catalog; update the preset dictionary according to the target knowledge points.

[0019] In a second aspect, the present application also provides a knowledge point recognition device, comprising:

[0020] A first determination module is used to obtain a target document, and determine a first knowledge point set and a second knowledge point set according to the file name and content information of the target document, wherein the first knowledge point set and the second knowledge point set each include at least one knowledge point;

[0021] A second determination module, configured to determine a third knowledge point set according to basic information of a user who uploaded the target document, wherein the third knowledge point set includes at least one knowledge point;

[0022] An identification module is used to perform cluster analysis on a target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set, and to use knowledge points whose similarity meets preset requirements as target knowledge points; the target knowledge point set includes the first knowledge point set, the second knowledge point set and the third knowledge point set.

[0023] In one of the embodiments, the recognition module is specifically used to construct a vector space based on each knowledge point in the target knowledge point set; calculate the similarity between any two knowledge points in the vector space to generate a similarity matrix; determine the similarity of each knowledge point based on the DBSCAN algorithm and the similarity matrix, wherein the knowledge points whose similarity meets the preset requirements are divided into the same cluster.

[0024] In one of the embodiments, the first determination module is specifically used to determine the first knowledge point set according to the file name of the target document and a preset dictionary; and determine the second knowledge point set according to the content information of the target document and a preset recognition model.

[0025] In one of the embodiments, the first determination module is specifically used to perform word segmentation processing on the content information; the data obtained after the word segmentation processing is input into a bidirectional encoder for feature extraction processing to obtain multiple word vectors; the word vectors are input into a preset recognition model, and the second knowledge point set is determined according to the output of the recognition model, and the preset recognition model includes a bidirectional long short-term memory network, and / or a conditional random field.

[0026] In one embodiment, the first determination module is specifically used to clean the content information and remove noise data in the content information; input the cleaned content information into the hidden Markov model, and perform word segmentation processing on the content information according to the hidden Markov model.

[0027] In one of the embodiments, the recognition module is further used to obtain a preset dictionary, which includes subjects, knowledge points and course catalogs; and update the preset dictionary according to the target knowledge points.

[0028] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, any of the methods described in the first aspect is implemented.

[0029] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements any of the methods described in the first aspect above.

[0030] In a fifth aspect, the present application further provides a computer program product, including a computer program, which, when executed by a processor, implements any of the methods described in the first aspect above.

[0031] The above-mentioned knowledge point identification method, device, equipment, storage medium and program product obtain the target document, and determine the first knowledge point set and the second knowledge point set respectively according to the file name and content information of the target document, and the first knowledge point set and the second knowledge point set respectively include at least one knowledge point; then, determine the third knowledge point set according to the basic information of the uploading user of the target document, and the third knowledge point set includes at least one knowledge point; finally, cluster analysis is performed on the target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set, and the knowledge point whose similarity meets the preset requirements is used as the target knowledge point; the target knowledge point set includes the first knowledge point set, the second knowledge point set and the third knowledge point set. By determining multiple knowledge point sets from the analysis of different levels such as the content information, file name and basic information of the uploading user of the target document, and then determining the target knowledge point set from the multiple knowledge point sets based on cluster analysis, the accuracy of knowledge point annotation can be effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0033] Figure 1 An application environment diagram of a knowledge point identification method in an embodiment;

[0034] Figure 2 is a flow chart of a knowledge point identification method in one embodiment;

[0035] Figure 3 A schematic diagram of a flow chart of a step of determining similarity of each knowledge point in a target knowledge point set in one embodiment;

[0036] Figure 4 A schematic diagram of a flow chart of the steps of determining a first knowledge point set and a second knowledge point set in one embodiment;

[0037] Figure 5 A schematic diagram of a flow chart of a step of determining a second knowledge point set in another embodiment;

[0038] Figure 6 A schematic diagram of a flow chart of a word segmentation process for content information in one embodiment;

[0039] Figure 7 is a flow chart of a knowledge point identification method in another embodiment;

[0040] Figure 8is a flow chart of a knowledge point identification method in another embodiment;

[0041] Fig. 9 is a structural block diagram of a knowledge point identification device in one embodiment;

[0042] Fig.10 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0044] The knowledge point recognition method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The knowledge point recognition method can be applied to the server 104, and can also be implemented through the interaction between the terminal 102 and the server 104. For example, the uploading user uploads the target document to the server 104 through the terminal 102, and the server 104 obtains the target document, and determines the first knowledge point set and the second knowledge point set according to the file name and content information of the target document, respectively. The first knowledge point set and the second knowledge point set each include at least one knowledge point, and then determines the third knowledge point set according to the basic information of the uploading user of the target document, and the third knowledge point set includes at least one knowledge point, and then performs cluster analysis on the target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set, and the knowledge point whose similarity meets the preset requirements is used as the target knowledge point, and the target knowledge point set includes the first knowledge point set, the second knowledge point set and the third knowledge point set. The terminal 102 may be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart speakers, smart TVs, smart air conditioners, smart car devices, projection devices, etc. Portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. Head-mounted devices may be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. The server 104 may be an independent physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides cloud computing services.

[0045] In an exemplary embodiment, Figure 2As shown, a knowledge point recognition method is provided, which is applied to Figure 1 The server 104 in the example is used as an example to illustrate, including the following steps 201 to 203. Among them:

[0046] S201, obtaining a target document, and determining a first knowledge point set and a second knowledge point set according to the file name and content information of the target document.

[0047] The first knowledge point set and the second knowledge point set each include at least one knowledge point.

[0048] Optionally, the target document may be uploaded to the server 104 by the teacher via the terminal 102 , and the target document is stored in a storage medium of the server 104 . The target document may be a teaching resource, such as courseware, test papers, or lesson plans.

[0049] Optionally, the target document may be obtained immediately when the teacher uploads it, or may be obtained from a storage medium according to a scheduled task. The embodiment of the present application does not limit the triggering timing for obtaining the target document.

[0050] Optionally, the first knowledge point set may be determined according to the file name of the target document, and the second knowledge point set may be determined according to the content information of the target document.

[0051] Optionally, the target document's file name may be named according to preset rules, so the target document's file name may be divided according to the preset rules to obtain a first knowledge point set, or the target document's file name may be matched with a preset dictionary and the matching result may be used as the first knowledge point set.

[0052] Optionally, when determining the second set of knowledge points based on the content information of the target document, the content information in the target document can be read and then input into a statistical machine learning model or a deep learning model to determine the second set of knowledge points based on the output of the statistical machine learning model or the deep learning model; or, information such as titles, subtitles, and lists in the content information can be obtained and then matched with the information such as titles, subtitles, and lists through a preset dictionary, and the matching results can be used as the second set of knowledge points.

[0053] S202: Determine a third knowledge point set according to basic information of the user who uploaded the target document.

[0054] Among them, the third knowledge point set includes at least one knowledge point.

[0055] It is understandable that the uploading user may be a teacher. When uploading the target document through the terminal, the uploading user needs to log in. After the uploading user's account authority is verified, the uploading user can successfully upload the target document.

[0056] Optionally, basic information of the uploading user may be obtained based on the uploading user's account. The basic information may include the school where the uploading user is studying, such as elementary school, junior high school, high school or university, and may also include the subjects, grades and class information taught by the uploading user.

[0057] Optionally, the third knowledge point set can be determined based on the basic information of the uploading user. For example, the third knowledge point set can be determined based on the subject, age, and / or school of the uploading user. For example, if the subject is mathematics, then mathematics-related knowledge points can be obtained as the third knowledge point set. Or, if the subject is mathematics and the age is first grade, then first-grade mathematics-related knowledge points can be used as the third knowledge point set. The embodiment of the present application does not limit the method for determining the third knowledge point set.

[0058] S203, performing cluster analysis on the target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set, and taking the knowledge points whose similarity meets preset requirements as target knowledge points.

[0059] Among them, the target knowledge point set includes a first knowledge point set, a second knowledge point set and a third knowledge point set.

[0060] Optionally, a target knowledge point set can be obtained by summarizing the knowledge points in the first knowledge point set, the second knowledge point set, and the third knowledge point set. It can be understood that there may be duplicate knowledge points in the first knowledge point set, the second knowledge point set, and the third knowledge point set. When determining the target knowledge point set, the knowledge points can be deduplicated or the duplicate knowledge points can be retained. The embodiment of the present application does not limit the method for determining the target knowledge point set.

[0061] Optionally, clustering analysis of the target knowledge point set may include a hierarchical clustering method, a partition clustering method or a density clustering method. By performing cluster analysis on the target knowledge point set, at least one cluster can be obtained. It can be understood that the similarity of knowledge points in the same cluster is higher than that in not the same cluster.

[0062] In one possible implementation, knowledge points whose similarity meets preset requirements are taken as target knowledge points. By determining the number of knowledge points in each cluster, the cluster with the largest number of knowledge points can be taken as the target cluster, and the knowledge points included in the target cluster can be taken as the target knowledge points.

[0063] In another possible implementation, knowledge points whose similarity meets preset requirements are taken as target knowledge points. Alternatively, a cluster with more than one knowledge point in the cluster is taken as the target cluster, and then the knowledge points included in the target cluster are taken as the target knowledge points.

[0064] In another possible implementation, when the number of knowledge points in the cluster is one, the knowledge point can be sent to the uploading user so that the uploading user can determine whether the knowledge point is the target knowledge point based on the content information of the target document. If a confirmation instruction is received, the knowledge point can also be used as the target knowledge point.

[0065] It can be understood that the target knowledge point includes at least one knowledge point, and the target knowledge point can be a knowledge point with relatively high similarity in the first knowledge point set, the second knowledge point set and the third knowledge point set. For example, if a knowledge point is considered to be a target knowledge point, then the knowledge point or a knowledge point that is relatively similar to the knowledge point exists in at least two knowledge point sets.

[0066] Optionally, the hierarchical clustering method can construct a hierarchical structure of clusters based on the similarity between clusters, which can be divided into agglomerative hierarchical clustering and divisive hierarchical clustering. The agglomerative hierarchical clustering can start with each knowledge point in the target knowledge point set as a separate cluster, and then continuously merge clusters with high similarity. The divisive hierarchical clustering can start with all knowledge points in the target knowledge point set in one cluster, and gradually split into different clusters. When clustering is performed based on the hierarchical clustering method, the similarity between knowledge points can be determined by a similarity measurement method, such as through Euclidean distance, Manhattan distance or cosine similarity, and then a hierarchical structure is constructed until a preset stopping condition is reached, for example, the number of clusters reaches a preset number or the distance between clusters is greater than a preset threshold.

[0067] Optionally, the partitioning clustering method can be K-means clustering, which divides the target knowledge point set into a pre-specified number of clusters so that the data points within each cluster have high similarity, while the data points between different clusters have low similarity. When clustering is performed based on the partitioning clustering method, a specified number of clusters can be randomly initialized, and then each knowledge point is assigned to the cluster where the cluster center closest to it is located. After all knowledge points are assigned, the center of each cluster is recalculated, and the process of knowledge point assignment and center calculation is repeated until the cluster center no longer changes or the preset number of iterations is reached.

[0068] Optionally, the density clustering method may be based on the density of knowledge points for clustering. If the density of knowledge points in a region exceeds a preset threshold, the region may be considered as a cluster.

[0069] The above-mentioned knowledge point identification method obtains the target document, and determines the first knowledge point set and the second knowledge point set respectively according to the file name and content information of the target document, and the first knowledge point set and the second knowledge point set respectively include at least one knowledge point; then, the third knowledge point set is determined according to the basic information of the uploading user of the target document, and the third knowledge point set includes at least one knowledge point; finally, cluster analysis is performed on the target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set, and the knowledge point whose similarity meets the preset requirements is used as the target knowledge point; the target knowledge point set includes the first knowledge point set, the second knowledge point set and the third knowledge point set. By determining multiple knowledge point sets from the analysis of different levels such as the content information of the target document, the file name and the basic information of the uploading user, and then determining the target knowledge point set from the multiple knowledge point sets based on cluster analysis, the accuracy of knowledge point annotation can be effectively improved.

[0070] In an exemplary embodiment, Figure 3 As shown, optionally, cluster analysis is performed on the target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set, including the following steps 301 to 303. Among them:

[0071] S301, constructing a vector space according to each knowledge point in the target knowledge point set.

[0072] Optionally, since each knowledge point in the target knowledge point set is text data, features can be extracted based on lexical and syntactic analysis to construct a vector. For example, a bag-of-words model or term frequency-inverse document frequency (TF-IDF) methods can be used to extract features of each knowledge point, and the features are arranged in a certain order to form a vector in a fixed format.

[0073] Optionally, the vectors corresponding to the knowledge points are integrated together to form a vector space, mapping the vocabulary corresponding to the knowledge point to a point in a high-dimensional space.

[0074] S302, calculating the similarity between any two knowledge points in the vector space and generating a similarity matrix.

[0075] Optionally, the similarity between any two knowledge points in the vector space may be calculated according to a preset similarity measurement algorithm, and each element in the similarity matrix represents the degree of similarity between the ith knowledge point and the jth knowledge point.

[0076] Optionally, the similarity measurement algorithm may include Euclidean distance, Manhattan distance, cosine similarity, etc.

[0077] For example, on a two-dimensional plane, when the similarity between any two knowledge points in a vector space is determined by a cosine similarity algorithm, vector A is ( , ), vector B is ( ) as an example, the following formula can be used:

[0078]

[0079] in, is the similarity, and its value range is between -1 and 1. When the value is close to 1, it means that the angle between the two vectors is small, that is, their directions are very close, so the similarity is high; conversely, when the value is close to -1, it means that the angle between the two vectors is large, that is, their directions are opposite, so the similarity is low. When the cosine similarity is 0, it means that the angle between the two vectors is 90 degrees, that is, they are orthogonal to each other. In this case, the two vectors are completely unrelated in direction, so it can be considered that their similarity is low or not similar.

[0080] S303, determining the similarity of each knowledge point according to the DBSCAN algorithm and the similarity matrix.

[0081] Among them, knowledge points whose similarity meets the preset requirements are divided into the same cluster.

[0082] Optionally, the similarity matrix can be analyzed according to the DBSCAN algorithm, where the DBSCAN algorithm is a density clustering method that can divide clusters based on the density of data points. Unlike traditional clustering algorithms based on distance or center points, the DBSCAN algorithm can find clusters of any shape and can identify noise points. DBSCAN determines the core objects and boundary objects based on the preset neighborhood radius and the minimum number of samples, and marks objects with density reachability as the same cluster. Based on the output results of DBSCAN, the similarity between knowledge points can be evaluated. If two knowledge points are classified into the same cluster, they are considered to have a high similarity; otherwise, their similarity is low.

[0083] Optionally, the core object can be determined based on the similarity matrix. Based on the preset neighborhood radius and minimum number of samples, the similarity matrix can be used to determine whether an object is a core object. If the number of objects in the neighborhood of an object (which objects are in the neighborhood determined by the similarity matrix) is not less than the minimum number of samples, then the object is a core object.

[0084] Optionally, at the beginning of clustering, after a core object is determined, all objects in its neighborhood, including other core objects and boundary objects, can be found according to the similarity matrix, thereby dividing the initial range of a cluster. By continuously checking the neighborhoods of the core objects in these neighborhoods, the boundaries of the cluster are gradually expanded. In this process, the similarity matrix continues to play a role, helping to determine which objects should be included in the cluster and which objects should be excluded.

[0085] Optionally, the similarity matrix can also be used as a reference for evaluating the quality of clustering. After clustering is completed, the similarity between knowledge points in different clusters and the similarity between knowledge points in different clusters can be observed to determine whether the clustering is reasonable. For example, if the similarity between knowledge points in a cluster is low, or the similarity between knowledge points in different clusters is high, it may mean that the clustering effect is not ideal, and it is necessary to adjust the algorithm parameters or reconsider the feature representation of the data.

[0086] The above method constructs a vector space according to each knowledge point in the target knowledge point set; calculates the similarity between any two knowledge points in the vector space to generate a similarity matrix; determines the similarity of each knowledge point according to the DBSCAN algorithm and the similarity matrix, wherein the knowledge points whose similarity meets the preset requirements are divided into the same cluster, and clustering analysis can be performed based on the DBSCAN algorithm. The number of clusters can be automatically determined according to the density of the knowledge points, and there is no need to preset the number of clusters. At the same time, noise points can be identified. When processing noisy data, the noise points can be effectively separated without causing too much interference to the clustering results, which can improve the accuracy of knowledge point recognition.

[0087] In an exemplary embodiment, Figure 4 As shown, optionally, determining the first knowledge point set and the second knowledge point set respectively according to the file name and content information of the target document includes the following steps 401 to 402. Among them:

[0088] S401, determining a first knowledge point set according to the file name of the target document and a preset dictionary.

[0089] Optionally, the file name of the target document can be extracted through the pdfkit library of the python language, or the file name of the target document can be obtained through front-end parsing when the target document is uploaded. The embodiment of the present application does not limit the method of obtaining the file name of the target document.

[0090] Optionally, the preset dictionary may include subjects, knowledge points and courses. It can be understood that when determining the first knowledge point set, the preset dictionary is up-to-date and accurate, which can improve the accuracy of the first knowledge point set.

[0091] Optionally, the file name of the target document may be matched with a preset dictionary according to a forward maximum matching method, a reverse maximum matching method, a bidirectional maximum matching method, or the like to determine the first knowledge point set.

[0092] S402: Determine a second knowledge point set according to content information of the target document and a preset recognition model.

[0093] In one possible implementation, the file format of the target document can be converted into the target format, and then the content information of the target document can be extracted through a preset parsing method. For example, the file format of the target document can be converted into a PDF version, and then the PDF parsing library PyPDF2 can be used to read the text content of the target document.

[0094] In another possible implementation manner, a corresponding parsing method may be determined according to the file format of the target document, and content information of the target document may be extracted according to the corresponding parsing method.

[0095] Optionally, the embodiments of the present application support conversion and processing of multiple file formats and have strong flexibility and adaptability.

[0096] Optionally, the preset recognition model can be implemented based on machine learning, deep learning or a pre-trained model. For example, the recognition model can be determined using a hidden Markov chain and a conditional random field, or the content information can be input into a convolutional neural network or a recurrent neural network or a bidirectional encoder to encode and represent the text token sequence, and then a fully connected layer is used to classify each token in the sequence, and finally Softmax or a conditional random field is used to make the final label judgment.

[0097] By determining the first knowledge point set according to the target document's file name and a preset dictionary, and determining the second knowledge point set according to the target document's content information and a preset recognition model, the first knowledge point set and the second knowledge point set can be determined to facilitate subsequent determination of the target knowledge points.

[0098] In an exemplary embodiment, Figure 5 As shown, optionally, determining the second knowledge point set according to the content information of the target document and a preset recognition model includes the following steps 501 to 503. Among them:

[0099] S501, performing word segmentation processing on content information.

[0100] Optionally, the content information may be segmented to divide the continuous content information into meaningful word units according to certain rules.

[0101] Exemplarily, the Chinese word segmentation tool Jieba can be used to segment sentences in the content information into words or phrases to implement word segmentation processing. Jieba word segmentation uses the precise mode, which attempts to segment sentences most accurately and is suitable for text analysis. It uses a method based on the Hidden Markov Model (HMM).

[0102] In one possible implementation, Figure 6 As shown, the word segmentation process for content information includes the following steps 601 to 602. Among them:

[0103] S601, cleaning the content information to remove noise data in the content information.

[0104] Optionally, the content information may be cleaned to remove noise data such as HTML tags and special characters in the content information. The content information may be cleaned using the Python program NLTK library.

[0105] S602: Input the cleaned content information into a hidden Markov model, and perform word segmentation processing on the content information according to the hidden Markov model.

[0106] Optionally, special tags, such as [CLS] and [SEP], may be added to the content information during word segmentation. [CLS] may be added to the beginning of a sentence in the content information, and [SEP] is used to separate different sentences, so that the boundary of each sentence can be better understood during word segmentation.

[0107] Optionally, a hidden Markov model is a statistical model used to describe a Markov process with unknown parameters. The hidden Markov model regards the word segmentation problem as a probabilistic model with hidden states (word boundary conditions, such as the beginning, middle, and end of a word) and observation states (characters or words in the text). By training on a large amount of corpus, the transition probability and emission probability between the hidden state and the observation state are learned, and then the most likely word segmentation result in the sentence is determined based on these probabilities, where the transition probability is the probability of transferring from one hidden state to another hidden state, and the emission probability is the probability of generating a certain observation value (Chinese character) in a certain hidden state.

[0108] Optionally, when performing word segmentation processing on content information according to the hidden Markov model, the optimal state sequence in the hidden Markov, that is, the most likely word segmentation method, can be solved based on the Viterbi algorithm, wherein the core formula in the Viterbi algorithm is as follows:

[0109]

[0110]

[0111] in, represents the maximum probability of state i at time t, is the state transition probability, which indicates the probability of transitioning from state j to state i. is observed in state i The probability of emission is is the maximum probability of reaching the state at time t The previous state j.

[0112] S502, input the data obtained after word segmentation processing into a bidirectional encoder for feature extraction processing to obtain multiple word vectors.

[0113] Optionally, the bidirectional encoder can be a pre-trained language model based on the Transformer architecture (Bidirectional Encoder Representations from Transformers, BERT).

[0114] Optionally, after the input data enters the bidirectional encoder, it will be encoded in multiple Transformer layers in sequence. In each Transformer layer, the input token representation is transformed and updated mainly through the Multi-Head Attention mechanism and the Feed-Forward Network. The Multi-Head Attention mechanism enables each token to pay attention to other related tokens in the input sequence, thereby capturing rich semantic relationships. The token can be the data obtained by word segmentation processing.

[0115] Optionally, after multiple layers of encoding by the bidirectional encoder, the hidden state corresponding to each token, i.e., the word vector, will be output in the end. You can choose to use the last layer of hidden state, the first layer of hidden state, or the hidden state of a specific layer as the word vector, where the last layer of hidden state is the last layer of word vector obtained after processing by the bidirectional encoder, which contains the richest and most comprehensive semantic information after being fully encoded by the bidirectional encoder; the first layer of hidden state reflects the most initial feature representation of the input data, which may retain more features about the surface of the data; the hidden state of a specific layer can be the hidden state of a certain middle layer as the word vector, or a fusion strategy can be used to perform weighted evaluation, splicing, and other operations on the hidden states of multiple layers to obtain the word vector.

[0116] Optionally, when training the bidirectional encoder, annotators will mark the locations of knowledge points in sentences in the training dataset as a basis for supervised learning, construct the training dataset, and check the automated results to correct errors and resolve complex situations that the model cannot handle.

[0117] S503, input the word vector into a preset recognition model, and determine the second knowledge point set according to the output of the recognition model.

[0118] Among them, the preset recognition model includes a bidirectional long short-term memory network and / or a conditional random field.

[0119] Optionally, the preset recognition model can be a bidirectional long-short-term memory network (Bidirectional Long-Short-Term Memory, BiLSTM). The recognition model further processes word vectors based on its own modeling ability of sequential information and mines deeper sequence features. This combination allows the rich semantic information obtained by BERT to be further refined and expanded in BiLSTM based on a sequential perspective. BiLSTM can use the high-quality word vectors output by BERT to better capture the sequential association and local semantic changes of words in the text, making up for the relatively less intuitive shortcomings of BERT in sequential information modeling; and BERT's powerful semantic understanding ability also provides BiLSTM with more valuable input, reducing BiLSTM's dependence on a large number of manual feature engineering. The two work together to improve the overall representation ability of the text. At this time, there is Softmax in the bidirectional long-short-term memory network to determine the final label of the word vector and determine the second knowledge point set.

[0120] Optionally, the recognition model may also be a conditional random field (CRF). The conditional random field is a probabilistic graphical model used to label or analyze sequence data. The word vector may be input into the CRF to obtain a second set of knowledge points.

[0121] Optionally, the recognition model can also be a bidirectional long-term memory network and a conditional random field, that is, the word vector is input into the bidirectional long-term memory network to obtain deeper sequence features, and then the output of the bidirectional long-term memory network is used as the input of the conditional random field. Finally, the second knowledge point set is obtained, and CRF is added after the bidirectional long-term memory network for the sequence labeling task, which can ensure the consistency and rationality of the annotation and recognition of the second knowledge point set.

[0122] Optionally, during the training of the recognition model, a loss function and an optimization item may be used to train the recognition model.

[0123] In an exemplary embodiment, Figure 7 As shown, the knowledge point recognition method further includes the following steps 701 to 702. Among them:

[0124] S701, obtaining a preset dictionary.

[0125] Among them, the preset dictionary includes subjects, knowledge points and course catalogues.

[0126] Optionally, the preset dictionary may be summarized in advance by the uploading user and saved in a database or stored as a configuration file.

[0127] It is understandable that the preset dictionary can be continuously updated and can provide a basic reference for knowledge point annotation.

[0128] S702, updating the preset dictionary according to the target knowledge point.

[0129] Optionally, the preset dictionary can be updated according to the target knowledge points. The preset dictionary can be compared with the target knowledge points, and the target knowledge points not in the preset dictionary can be added to the preset dictionary, that is, the preset dictionary can be updated according to the difference between the target knowledge points and the preset dictionary, so that the preset dictionary is more accurate. In this way, it can be more accurate when the first knowledge point set is determined subsequently. At the same time, the preset dictionary can be continuously updated and improved according to new text data to achieve continuous learning and optimization, which helps to maintain the timeliness and accuracy of the dictionary and improve the long-term performance of knowledge point recognition.

[0130] As an optional implementation, Figure 8 As shown, the knowledge point identification method provided in the embodiment of the present application may include the following specific steps:

[0131] S801, obtaining a target document.

[0132] S802: Determine a first knowledge point set according to the file name of the target document and a preset dictionary.

[0133] S803: Clean the content information of the target document to remove noise data in the content information.

[0134] S804: Input the cleaned content information into a hidden Markov model, and perform word segmentation processing on the content information according to the hidden Markov model.

[0135] S805, input the data obtained after word segmentation processing into a bidirectional encoder for feature extraction processing to obtain multiple word vectors.

[0136] S806, input the word vector into a preset recognition model, and determine the second knowledge point set according to the output of the recognition model.

[0137] Among them, the preset recognition model includes a bidirectional long short-term memory network and / or a conditional random field.

[0138] S807: Determine a third knowledge point set according to the basic information of the user who uploaded the target document.

[0139] Among them, the first knowledge point set, the second knowledge point set and the third knowledge point set each include at least one knowledge point.

[0140] S808, constructing a vector space according to each knowledge point in the target knowledge point set.

[0141] Among them, the target knowledge point set includes a first knowledge point set, a second knowledge point set and a third knowledge point set.

[0142] S809, calculating the similarity between any two knowledge points in the vector space and generating a similarity matrix.

[0143] S810, determining the similarity of each knowledge point according to the DBSCAN algorithm and the similarity matrix.

[0144] Among them, knowledge points whose similarity meets the preset requirements are divided into the same cluster.

[0145] S811, taking the knowledge points whose similarity meets the preset requirements as target knowledge points.

[0146] S812, obtaining a preset dictionary.

[0147] Among them, the preset dictionary includes subjects, knowledge points and course catalogues.

[0148] S813, updating the preset dictionary according to the target knowledge point.

[0149] Optionally, the knowledge point identification method in the embodiment of the present application is automated from file uploading to dictionary updating, greatly reducing manual intervention and operating costs. Teachers only need to upload files, and the system can automatically complete all subsequent steps. Moreover, through automated knowledge point mining and annotation technology, teachers can more conveniently obtain and organize knowledge point information in teaching content, thereby carrying out teaching design and implementation in a more targeted manner, which helps to improve teaching quality and students' learning outcomes.

[0150] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0151] Based on the same inventive concept, the embodiment of the present application also provides a knowledge point identification device for implementing the above-mentioned knowledge point identification method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in one or more knowledge point identification device embodiments provided below can refer to the limitations of the knowledge point identification method above, and will not be repeated here.

[0152] In an exemplary embodiment, Fig. 9 As shown, a knowledge point recognition device 900 is provided, comprising: a first determination module 901, a second determination module 902 and a recognition module 903, wherein:

[0153] A first determination module 901 is used to obtain a target document, and determine a first knowledge point set and a second knowledge point set according to the file name and content information of the target document, wherein the first knowledge point set and the second knowledge point set each include at least one knowledge point;

[0154] A second determining module 902 is used to determine a third knowledge point set according to basic information of the user who uploaded the target document, wherein the third knowledge point set includes at least one knowledge point;

[0155] Identification module 903 is used to perform cluster analysis on the target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set, and take the knowledge points whose similarity meets preset requirements as target knowledge points; the target knowledge point set includes the first knowledge point set, the second knowledge point set and the third knowledge point set.

[0156] In an embodiment of an embodiment, the identification module 903 is specifically used to construct a vector space based on each knowledge point in the target knowledge point set; calculate the similarity between any two knowledge points in the vector space to generate a similarity matrix; determine the similarity of each knowledge point based on the DBSCAN algorithm and the similarity matrix, wherein the knowledge points whose similarity meets the preset requirements are divided into the same cluster.

[0157] In one embodiment, the first determination module 901 is specifically used to determine a first set of knowledge points according to the file name of the target document and a preset dictionary; and determine a second set of knowledge points according to content information of the target document and a preset recognition model.

[0158] In one embodiment, the first determination module 901 is specifically used to perform word segmentation processing on content information; input the data obtained after the word segmentation processing into a bidirectional encoder for feature extraction processing to obtain multiple word vectors; input the word vectors into a preset recognition model, and determine the second knowledge point set according to the output of the recognition model, and the preset recognition model includes a bidirectional long short-term memory network, and / or a conditional random field.

[0159] In one embodiment, the first determination module 901 is specifically used to clean the content information to remove noise data in the content information; input the cleaned content information into the hidden Markov model, and perform word segmentation processing on the content information according to the hidden Markov model.

[0160] In one embodiment, the identification module 903 is further used to obtain a preset dictionary, which includes subjects, knowledge points and course catalogs; and update the preset dictionary according to the target knowledge points.

[0161] Each module in the above-mentioned knowledge point recognition device can be implemented in whole or in part by software, hardware and their combination. Each module can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to each module above.

[0162] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Fig.10As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data related to the knowledge point recognition method. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a knowledge point recognition method is implemented.

[0163] Those skilled in the art will understand that Fig.10 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0164] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0165] Obtain a target document, and determine a first knowledge point set and a second knowledge point set according to the file name and content information of the target document, respectively, wherein the first knowledge point set and the second knowledge point set respectively include at least one knowledge point; determine a third knowledge point set according to basic information of the uploading user of the target document, wherein the third knowledge point set includes at least one knowledge point; perform cluster analysis on the target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set, and use the knowledge points whose similarity meets preset requirements as target knowledge points; the target knowledge point set includes the first knowledge point set, the second knowledge point set, and the third knowledge point set.

[0166] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0167] A vector space is constructed according to each knowledge point in the target knowledge point set; the similarity between any two knowledge points in the vector space is calculated to generate a similarity matrix; the similarity of each knowledge point is determined according to the DBSCAN algorithm and the similarity matrix, wherein the knowledge points whose similarity meets the preset requirements are divided into the same cluster.

[0168] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0169] The first knowledge point set is determined according to the file name of the target document and a preset dictionary; and the second knowledge point set is determined according to the content information of the target document and a preset recognition model.

[0170] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0171] The content information is segmented; the data obtained after the word segmentation is input into a bidirectional encoder for feature extraction to obtain multiple word vectors; the word vectors are input into a preset recognition model, and a second knowledge point set is determined according to the output of the recognition model, wherein the preset recognition model includes a bidirectional long short-term memory network, and / or a conditional random field.

[0172] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0173] The content information is cleaned to remove noise data in the content information; the cleaned content information is input into a hidden Markov model, and the content information is segmented according to the hidden Markov model.

[0174] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0175] Get the preset dictionary, which includes subjects, knowledge points and course catalog; update the preset dictionary according to the target knowledge points.

[0176] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0177] Obtain a target document, and determine a first knowledge point set and a second knowledge point set according to the file name and content information of the target document, respectively, wherein the first knowledge point set and the second knowledge point set respectively include at least one knowledge point; determine a third knowledge point set according to basic information of the uploading user of the target document, wherein the third knowledge point set includes at least one knowledge point; perform cluster analysis on the target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set, and use the knowledge points whose similarity meets preset requirements as target knowledge points; the target knowledge point set includes the first knowledge point set, the second knowledge point set, and the third knowledge point set.

[0178] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0179] A vector space is constructed according to each knowledge point in the target knowledge point set; the similarity between any two knowledge points in the vector space is calculated to generate a similarity matrix; the similarity of each knowledge point is determined according to the DBSCAN algorithm and the similarity matrix, wherein the knowledge points whose similarity meets the preset requirements are divided into the same cluster.

[0180] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0181] The first knowledge point set is determined according to the file name of the target document and a preset dictionary; and the second knowledge point set is determined according to the content information of the target document and a preset recognition model.

[0182] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0183] The content information is segmented; the data obtained after the word segmentation is input into a bidirectional encoder for feature extraction to obtain multiple word vectors; the word vectors are input into a preset recognition model, and a second knowledge point set is determined according to the output of the recognition model, wherein the preset recognition model includes a bidirectional long short-term memory network, and / or a conditional random field.

[0184] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0185] The content information is cleaned to remove noise data in the content information; the cleaned content information is input into a hidden Markov model, and the content information is segmented according to the hidden Markov model.

[0186] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0187] Get the preset dictionary, which includes subjects, knowledge points and course catalog; update the preset dictionary according to the target knowledge points.

[0188] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0189] Obtain a target document, and determine a first knowledge point set and a second knowledge point set according to the file name and content information of the target document, respectively, wherein the first knowledge point set and the second knowledge point set respectively include at least one knowledge point; determine a third knowledge point set according to basic information of the uploading user of the target document, wherein the third knowledge point set includes at least one knowledge point; perform cluster analysis on the target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set, and use the knowledge points whose similarity meets preset requirements as target knowledge points; the target knowledge point set includes the first knowledge point set, the second knowledge point set, and the third knowledge point set.

[0190] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0191] A vector space is constructed according to each knowledge point in the target knowledge point set; the similarity between any two knowledge points in the vector space is calculated to generate a similarity matrix; the similarity of each knowledge point is determined according to the DBSCAN algorithm and the similarity matrix, wherein the knowledge points whose similarity meets the preset requirements are divided into the same cluster.

[0192] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0193] The first knowledge point set is determined according to the file name of the target document and a preset dictionary; and the second knowledge point set is determined according to the content information of the target document and a preset recognition model.

[0194] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0195] The content information is segmented; the data obtained after the word segmentation is input into a bidirectional encoder for feature extraction to obtain multiple word vectors; the word vectors are input into a preset recognition model, and a second knowledge point set is determined according to the output of the recognition model, wherein the preset recognition model includes a bidirectional long short-term memory network, and / or a conditional random field.

[0196] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0197] The content information is cleaned to remove noise data in the content information; the cleaned content information is input into a hidden Markov model, and the content information is segmented according to the hidden Markov model.

[0198] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0199] Get the preset dictionary, which includes subjects, knowledge points and course catalog; update the preset dictionary according to the target knowledge points.

[0200] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0201] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0202] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A knowledge point recognition method, characterized in that: The method comprises: Acquire a target document, and determine a first knowledge point set and a second knowledge point set according to the file name and content information of the target document, wherein the first knowledge point set and the second knowledge point set each include at least one knowledge point; Determine a third knowledge point set according to basic information of the user who uploaded the target document, wherein the third knowledge point set includes at least one knowledge point; A cluster analysis is performed on the target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set, and knowledge points whose similarity meets preset requirements are used as target knowledge points; the target knowledge point set includes the first knowledge point set, the second knowledge point set and the third knowledge point set.

2. The method according to claim 1, characterized in that The cluster analysis of the target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set includes: Constructing a vector space according to each knowledge point in the target knowledge point set; Calculating the similarity between any two knowledge points in the vector space to generate a similarity matrix; The similarity of each of the knowledge points is determined according to the DBSCAN algorithm and the similarity matrix, wherein the knowledge points whose similarity meets preset requirements are divided into the same cluster.

3. The method according to claim 1, characterized in that The determining the first knowledge point set and the second knowledge point set respectively according to the file name and content information of the target document comprises: Determining the first knowledge point set according to the file name of the target document and a preset dictionary; The second knowledge point set is determined according to the content information of the target document and a preset recognition model.

4. The method according to claim 3, characterized in that The determining the second knowledge point set according to the content information of the target document and a preset recognition model includes: Performing word segmentation processing on the content information; The data obtained after word segmentation is input into a bidirectional encoder for feature extraction to obtain multiple word vectors; The word vector is input into a preset recognition model, and the second knowledge point set is determined according to the output of the recognition model. The preset recognition model includes a bidirectional long short-term memory network and / or a conditional random field.

5. The method according to claim 4, characterized in that The word segmentation processing of the content information includes: Cleaning the content information to remove noise data in the content information; The cleaned content information is input into a hidden Markov model, and word segmentation processing is performed on the content information according to the hidden Markov model.

6. The method according to claim 1, characterized in that The method further comprises: Obtaining a preset dictionary, wherein the preset dictionary includes subjects, knowledge points and course catalogues; The preset dictionary is updated according to the target knowledge point.

7. A knowledge point recognition device, characterized in that: The device comprises: A first determination module is used to obtain a target document, and determine a first knowledge point set and a second knowledge point set according to the file name and content information of the target document, wherein the first knowledge point set and the second knowledge point set each include at least one knowledge point; A second determination module, configured to determine a third knowledge point set according to basic information of a user who uploaded the target document, wherein the third knowledge point set includes at least one knowledge point; An identification module is used to perform cluster analysis on a target knowledge point set to determine the similarity of each knowledge point in the target knowledge point set, and to use knowledge points whose similarity meets preset requirements as target knowledge points; the target knowledge point set includes the first knowledge point set, the second knowledge point set and the third knowledge point set.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images