Course recommendation method and device based on intelligent terminal, intelligent terminal and computer readable storage medium

Abstract

The embodiment of the invention relates to the technical field of informatization teaching, in particular to a course recommendation method and device based on an intelligent terminal and a computer readable storage medium. The method comprises the following steps: processing a student data set according to a natural nearest neighbor algorithm to obtain a data relation set, constructing an undirected weighted graph according to the data relation set, determining the composition and number of each connected sub-graph of the undirected weighted graph, and then mutually merging and reducing the connected sub-graphs to a preset clustering number k; constructing a similarity matrix through a Gaussian kernel function, and clustering the similarity matrix through an NJW spectral clustering algorithm to obtain a clustering result; and generating a learning label set of the students according to the clustering result, and generating a course recommendation sequence according to the learning label set.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of information-based teaching technology, and more particularly to a course recommendation method and device based on an intelligent terminal, and a computer-readable storage medium. Background Art

[0002] The traditional course search and screening process may consume a lot of time and energy, while intelligent course resource recommendation can automatically screen out suitable learning resources based on the user's needs and background, helping users find suitable learning materials more quickly; intelligent course resource recommendation can provide personalized learning content based on the user's interests, learning style and ability level, helping users better discover and develop their own potential. Intelligent course resource recommendation analyzes the user's learning needs and difficulties through data analysis and learning algorithms, and provides corresponding solutions to improve learning effects and efficiency.

[0003] Patent document CN117575745B - Personalized recommendation method for course teaching resources based on AI big data (hereinafter referred to as Patent Document 1) discloses obtaining the user attention level of each cluster based on the browsing time and average number of clicks of the course resource window corresponding to the data point in each cluster and the user attention level of each cluster; obtaining discrete data points based on the user attention level of each cluster; obtaining the user's keyword recommendation sequence based on several preferred keywords of the course resource window corresponding to the discrete data points, and recommending teaching resources to the user based on the user's keyword recommendation sequence.

[0004] In Patent Document 1, course recommendations are made to users based solely on their course resource browsing data. However, since it is impossible to determine the user's concentration when browsing course resources and the recommendation data comes from a single source, it is easy for the recommended courses to fail to meet the user's learning and interest requirements, resulting in poor course completion. Summary of the Invention

[0005] In view of the above problems, the embodiments of the present invention provide a course recommendation method, device and computer-readable storage medium based on a smart terminal, which are used to solve the problem that in the prior art, courses are recommended to users only based on the users' course resource browsing data. Since it is impossible to judge the user's concentration when browsing course resources and the recommendation data source is single, it is easy for the recommended courses to fail to meet the users' learning and interest requirements, thereby resulting in poor course completion.

[0006] According to one aspect of an embodiment of the present invention, a course recommendation method based on an intelligent terminal is provided, the method comprising:

[0007] Get the student dataset;

[0008] The student dataset is processed according to the natural nearest neighbor algorithm to obtain a data relationship set, an undirected weighted graph is constructed based on the data relationship set, and the composition and number of each connected subgraph of the undirected weighted graph are determined, and then the connected subgraphs are merged to reduce the number of clusters to a preset number k;

[0009] The similarity matrix is ​​constructed by Gaussian kernel function, and the similarity matrix is ​​clustered by NJW spectral clustering algorithm to obtain clustering results;

[0010] Generating a learning tag set of students according to the clustering results, and generating a course recommendation sequence from the learning tag set;

[0011] The completion probability of each course in the course recommendation sequence is calculated using a preset pine and cypress model, and the course recommendation sequence and the completion probability of each course are pushed to the user; wherein the pine and cypress distribution parameters of the preset pine and cypress model are any one or more of the student's academic performance, the completion rate of courses in the same field, and the student's interest level;

[0012] The student data set includes student basic data and student behavior data.

[0013] In an optional manner, the student basic data includes at least one or more of name, gender, age, major, class, contact information, and place of origin, and the student behavioral data includes at least one or more of interests, career planning, basic course learning status, professional course learning status, and interest course learning status.

[0014] In an optional manner, constructing an undirected weighted graph according to the data relationship set specifically includes:

[0015] The data relationship set NaN1 is taken as the edge relationship set, and each data set of the student data set is taken as the vertex relationship set. The undirected weighted graph is: NSG = (V, NaN1), wherein the undirected weighted graph is composed of many connected subgraphs, and there are one or more edges between any two points in the same connected subgraph that directly or indirectly connect the two points.

[0016] In an optional manner, determining the composition and number of connected subgraphs of the phase-free weighted graph and then reducing the number of connected subgraphs to a preset k by merging the connected subgraphs may specifically include:

[0017] Obtaining the number of connected subgraphs according to the undirected weighted graph, and comparing the number of connected subgraphs with a preset number of clusters;

[0018] If the number of connected subgraphs is greater than the preset number of clusters, a relationship edge is inserted between two preset vertices, and the undirected weighted graph and the number of connected subgraphs are updated until the number of connected subgraphs is equal to the preset number of clusters.

[0019] In an optional manner, the similarity matrix is ​​constructed by using the Gaussian kernel function and is calculated using the following formula:

[0020]

[0021] Among them, max{d ih :(i,h)∈NaN1} is the longest edge in the same connected subgraph.

[0022] In an optional manner, clustering the similarity matrix using the NJW spectral clustering algorithm to obtain a clustering result specifically includes:

[0023] Calculating a degree matrix through a similarity matrix, and obtaining a Laplace matrix according to the degree matrix and the similarity matrix;

[0024] Compute the first k largest eigenvectors of the Laplacian matrix {z1,z2,…,z k}, build matrix Z and normalize matrix Z to obtain matrix Y;

[0025] All data points y in the i-th row of matrix Y are transformed into i Clustering results {A1, A2, ..., A k}.

[0026] In an optional manner, generating a learning tag set of students according to the clustering result, and generating a course recommendation sequence from the learning tag set includes:

[0027] Extracting common labels from each cluster of the clustering result as learning labels for students, and generating a learning label set;

[0028] For the learning label items in the learning label set, the Apriori algorithm is used to determine the frequent item sets in the learning label items. Each frequent item set k generates 2*(2k-1) association rules. Each association rule is sorted according to the confidence level to obtain a course recommendation sequence.

[0029] In an optional manner, after obtaining the course recommendation sequence, the completion probability of each course in the course recommendation sequence is calculated through a preset pine and cypress model; wherein the pine and cypress distribution parameters of the preset pine and cypress model are any one or more of the student's academic performance, the completion rate of courses in the same field, and the student's interest level.

[0030] According to another aspect of an embodiment of the present invention, there is provided an intelligent course recommendation device, the device comprising:

[0031] Data acquisition module, used to obtain student data sets;

[0032] The clustering calculation module is used to process the student data set according to the natural nearest neighbor algorithm to obtain a data relationship set, construct an undirected weighted graph based on the data relationship set, and determine the composition and number of each connected subgraph of the undirected weighted graph, and then reduce the connected subgraphs to a preset number of clusters k by merging them; construct a similarity matrix using the Gaussian kernel function, and cluster the similarity matrix using the NJW spectral clustering algorithm to obtain the clustering results

[0033] A course recommendation module, configured to generate a set of learning labels for students based on the clustering results, and generate a course recommendation sequence from the set of learning labels;

[0034] The data push module is used to calculate the completion probability of each course in the course recommendation sequence through the preset Pine Cypress model, and push the course recommendation sequence and the corresponding completion probability to the user; wherein, the Pine Cypress distribution parameters of the preset Pine Cypress model are any one or more of the student's academic performance, the completion rate of courses in the same field, and the student's interest level.

[0035] According to another aspect of an embodiment of the present invention, there is provided an intelligent terminal, the device comprising a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus;

[0036] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations such as the above-mentioned course recommendation method based on the smart terminal.

[0037] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, wherein the storage medium stores at least one executable instruction. When the executable instruction is executed on the above-mentioned intelligent course recommendation device, the intelligent course recommendation device performs the operations of the above-mentioned intelligent terminal-based course recommendation method.

[0038] An embodiment of the present invention processes a student data set according to a natural nearest neighbor algorithm to obtain a data relationship set, constructs an undirected weighted graph based on the data relationship set, determines the composition and number of each connected subgraph of the undirected weighted graph, and then reduces the connected subgraphs to a preset number of clusters k by merging them together; constructs a similarity matrix using a Gaussian kernel function, and clusters the similarity matrix using an NJW spectral clustering algorithm to obtain a clustering result; generates a student learning tag set based on the clustering result, and generates a course recommendation sequence from the learning tag set.

[0039] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0040] 1. The student data set of the present invention includes student basic data and student behavior data. By clustering the student multidimensional data, the data source of the clustering model is richer and covers the student basic data and student behavior data, making the course recommendation results more accurate.

[0041] 2. The clustering method of the present invention first uses the natural nearest neighbor relationship without setting parameters, can search based on the characteristics of the data set itself, and is less affected by discrete points. It accurately divides the neighborhood of each sample, combines the Gaussian kernel function and the NJW spectral clustering algorithm to cluster the student data set, and more accurately reflects the structural relationship of the data set, thereby obtaining better clustering results.

[0042] 3. The present invention uses the Apriori algorithm to mine association rules. Through this solution, the accurate correlation between students' own learning behavior data can be obtained, and on this basis, accurate learning courses can be recommended to students.

[0043] 4. The present invention also calculates the completion probability of each course in the course recommendation sequence through a preset pine and cypress model, and reflects the completion probability of each course in the course recommendation sequence, providing another dimension of reference for students when making course selections, making this method more humane.

[0044] The above description is only an overview of the technical solutions of the embodiments of the present invention. In order to more clearly understand the technical means of the embodiments of the present invention, they can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiments of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are specifically listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings are only used to illustrate the embodiments and are not to be considered as limiting the present invention. In addition, the same reference symbols are used to represent the same components throughout the drawings. In the drawings:

[0046] Figure 1 A schematic diagram showing a flow chart of a first embodiment of a course recommendation method based on an intelligent terminal provided by the present invention;

[0047] Figure 2 The figure shows a schematic diagram of a process of clustering a similarity matrix using the NJW spectral clustering algorithm provided by the present invention to obtain clustering results;

[0048] Figure 3 The figure shows a schematic structural diagram of the intelligent course recommendation device provided by the present invention;

[0049] Figure 4 The structure of the intelligent terminal provided by the present invention is shown Figure 1 .

[0050] Figure 5 The structure of the intelligent terminal provided by the present invention is shown Figure 2 . DETAILED DESCRIPTION

[0051] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0052] Figure 1 A course recommendation method based on an intelligent terminal is shown, the method comprising:

[0053] 110. Obtain a student dataset. In step 110, the student dataset includes basic student data and behavioral student data. The basic student data includes at least one or more of name, gender, age, major, class, contact information, and place of origin. The behavioral student data includes at least one or more of interests, career planning, basic course learning, professional course learning, and interest course learning. Furthermore, the student dataset can be obtained through the school's student electronic files, which can be stored on the school's server.

[0054] 120. Process the student data set according to the natural nearest neighbor algorithm to obtain a data relationship set, construct an undirected weighted graph according to the data relationship set, and determine the composition and number of each connected subgraph of the undirected weighted graph, and then reduce the connected subgraphs to a preset number of clusters k by merging them with each other; in step 120, the clustering method of the present invention first accurately divides the neighborhood of each sample through the natural nearest neighbor relationship without setting parameters, can search based on the characteristics of the data set itself, and is less affected by discrete points.

[0055] 130. A similarity matrix is ​​constructed using a Gaussian kernel function, and the similarity matrix is ​​clustered using an NJW spectral clustering algorithm to obtain a clustering result. In step 130, the Gaussian kernel function and the NJW spectral clustering algorithm are combined to cluster the student dataset to more accurately reflect the structural relationship of the dataset, thereby obtaining a better clustering result.

[0056] 140. Generating a learning label set for a student based on the clustering results, and generating a course recommendation sequence from the learning label set. In step 130, generating a learning label set for a student based on the clustering results, and generating a course recommendation sequence from the learning label set, includes: extracting common labels from each cluster of the clustering results as the student's learning label, and generating a learning label set;

[0057] For the learning label items in the learning label set, the Apriori algorithm is used to determine the frequent item sets in the learning label items. Each frequent item set k generates 2*(2k-1) association rules. Each association rule is sorted according to the confidence level to obtain a course recommendation sequence.

[0058] 150, calculate the completion probability of each course in the course recommendation sequence through the preset cypress model, and push the course recommendation sequence and the completion probability of each course to the user; wherein, the cypress distribution parameter of the preset cypress model is any one or more of the student's academic performance, the completion rate of courses in the same field, and the student's interest level; after obtaining the course recommendation sequence, it also includes calculating the completion probability of each course in the course recommendation sequence through the preset cypress model; wherein, the cypress distribution parameter of the preset cypress model is any one or more of the student's academic performance, the completion rate of courses in the same field, and the student's interest level. In this embodiment, the Poisson model is a generalized linear model form of regression analysis, which is used to model count data and contingency tables. Poisson regression assumes that the response variable Y has a Poisson distribution, and assumes that the logarithm of its expected value can be modeled by a linear combination of unknown parameters. The Poisson regression model is sometimes called a log-linear model, especially when used to model contingency tables. The present invention also calculates the completion probability of each course in the course recommendation sequence through a preset pine and cypress model, and reflects the completion probability of each course in the course recommendation sequence, providing students with another dimension of reference when making course selections, making this method more applicable.

[0059] In the present invention, the student dataset includes both basic student data and behavioral student data. By clustering the multidimensional student data, the clustering model's data source is enriched, encompassing both basic student data and behavioral data, resulting in more accurate course recommendation results. Furthermore, the clustering method of the present invention first uses the natural nearest neighbor relationship, which eliminates the need for parameter settings, enables searches based on the dataset's inherent characteristics, and is less affected by discrete points. It accurately demarcates the neighborhood of each sample, then combines the Gaussian kernel function and the NJW spectral clustering algorithm to cluster the student dataset, more accurately reflecting the dataset's structural relationships and resulting in superior clustering results.

[0060]

[0061]

[0062] Table 1

[0063] Table 1 shows the differences between the present invention and the prior art. The prior art disclosed in Patent Document 1 cannot determine the student's concentration during browsing, which easily leads to clustering noise. The student dataset of the present invention includes student basic data and student behavior data. By clustering the student multidimensional data, the data source of the clustering model is richer and covers both student basic data and student behavior data, making the course recommendation results more accurate.

[0064] In an optional manner, the student data set is processed according to the natural nearest neighbor algorithm to obtain a data relationship set, which can be obtained in the following manner:

[0065] Define the r neighborhood,

[0066] Among them, findKNN(x i , n) represents the KNN search function, which returns x i The nth nearest neighbor; KNN r (xi) represents a subset of the student dataset X.

[0067] Define the natural nearest neighbor, based on the r neighborhood, if point X is in the r neighborhood of point Y and point Y is also in the r neighborhood of point X, then X and Y are called each other's natural nearest neighbors;

[0068]

[0069] The stable search state is defined as follows: the natural nearest neighbor algorithm reaches a stable search state if and only if the following conditions are met:

[0070]

[0071] Among them, r is the number of search rounds. If the above formula is satisfied, it means that the search round is stable.

[0072] In an optional manner, an undirected weighted graph is constructed based on the data relationship set, specifically including: taking the data relationship set NaN1 as the edge relationship set, taking each data set of the student data set as the vertex relationship set, and the undirected weighted graph is: NSG = (V, NaN1), wherein the undirected weighted graph is composed of many connected subgraphs, and there are one or more edges between any two points in the same connected subgraph that connect the two points directly or indirectly. Among them, NaN1 represents the set of edges connecting two vertices. Consider each data in the student data set X as a vertex. If two vertices x i and x j If an edge is inserted between them, the edge is placed in the NaN1 set.

[0073] In an optional method, the composition and number of each connected subgraph of the phase-free weighted graph are determined, and then the connected subgraphs are merged with each other to reduce them to a preset k number, specifically including: obtaining the number of connected subgraphs based on the undirected weighted graph, and comparing the number of connected subgraphs with the preset number of clusters; if the number of connected subgraphs is greater than the preset number of clusters, inserting a relationship edge between two preset vertices, and updating the undirected weighted graph and the number of connected subgraphs until the number of connected subgraphs is equal to the preset number of clusters.

[0074] In an optional approach, a similarity matrix is ​​constructed using a Gaussian kernel function and is calculated using the following formula:

[0075]

[0076] Among them, max{d ih :(i,h)∈NaN1} is the longest edge in the same connected subgraph.

[0077] In an optional manner, the similarity matrix is ​​clustered using the NJW spectral clustering algorithm to obtain a clustering result, specifically including:

[0078] 210, calculating a degree matrix using the similarity matrix, and obtaining a Laplace matrix based on the degree matrix and the similarity matrix;

[0079] 220, calculate the first k largest eigenvectors of the Laplacian matrix {z1,z2,…,z k}, build matrix Z and normalize matrix Z to obtain matrix Y;

[0080] 230, all data points y in the i-th row of matrix Y are transformed into i Clustering results {A1, A2, ..., A k}.

[0081]

[0082] And obtain the Laplace matrix according to the degree matrix and the similarity matrix: L=D -1 / 2 AD -1 / 2 , calculate the first k largest eigenvectors of the Laplacian matrix {z1,z2,…,z k}, establish the matrix Z, where Z={z1,z2,…,z k}∈R n*k And normalize the matrix Z to obtain the matrix Y, whose formula is:

[0083]

[0084] Let y i ∈R kFor the i-th row of matrix Y, all data points y in the i-th row of matrix Y are transformed into i Clustering results {A1, A2, ..., A k}.

[0085] In an optional manner, generating a learning tag set of students according to the clustering result, and generating a course recommendation sequence from the learning tag set includes:

[0086] According to each cluster of the clustering result, common labels are extracted as learning labels of students, and a learning label set is generated; common features of students are found in each cluster, and the common features are used as the learning label set.

[0087] For the learning label items in the learning label set, the Apriori algorithm is used to determine the frequent item sets in the learning label items. Each frequent item set k generates 2*(2k-1) association rules. Each association rule is sorted according to the confidence level to obtain a course recommendation sequence. The Apriori algorithm is the first association rule mining algorithm. It uses an iterative method of layer-by-layer search to find the relationship between item sets in the database to form rules. The process consists of connection and pruning. The concept of item set in this algorithm is a set of items. A set containing K items is a k-item set. The frequency of an item set is the number of transactions containing the item set, which is called the frequency of the item set. If an item set meets the minimum support, it is a frequent item set. The association rules in the label items can be learned by the Apriori algorithm; and the association rules are sorted according to the confidence level. For example, if there are five courses, A, B, C, D, and F, and the confidence levels of A, B, C, D, and F are 90%, 80%, 60%, 50%, and 20%, respectively, then the order of ranking is A, B, C, D, and F, with A being the most recommended course. This invention uses the Apriori algorithm to mine association rules. This solution can accurately determine the correlations between a student's various learning behavior data, and based on this, accurately recommend courses for the student.

[0088] Example 2:

[0089] According to another aspect of the present invention, an intelligent course recommendation device is provided. Figure 3The device includes: a data acquisition module 310 for executing step 110 in Example 1 to obtain basic student data and student behavior data; a clustering calculation module 320 for executing steps 120 and 130 in Example 1 to cluster student data sets; a course recommendation module 330 for executing step 140 in Example 1 to recommend courses to students based on the clustering results. A data push module 340 is used to execute step 150 in Example 1 above for recommending courses and completion probabilities; wherein the pine and cypress distribution parameters of the preset pine and cypress model are any one or more of the student's academic performance, the completion rate of courses in the same field, and the student's interest level. An embodiment of the present invention processes a student data set according to a natural nearest neighbor algorithm to obtain a data relationship set, constructs an undirected weighted graph based on the data relationship set, determines the composition and number of each connected subgraph of the undirected weighted graph, and then reduces the connected subgraphs to a preset number of clusters k by merging them together; constructs a similarity matrix using a Gaussian kernel function, and clusters the similarity matrix using an NJW spectral clustering algorithm to obtain a clustering result; generates a student learning tag set based on the clustering result, and generates a course recommendation sequence from the learning tag set.

[0090] Example 3:

[0091] Figure 4 The schematic diagram of the structure of an embodiment of the intelligent terminal of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the intelligent terminal.

[0092] like Figure 4 As shown, the intelligent terminal may include: a processor (processor) 402 , a communication interface (Communications Interface) 404 , a memory (memory) 406 , and a communication bus 408 .

[0093] Processor 402, communication interface 404, and memory 406 communicate with each other via communication bus 408. Communication interface 404 is used to communicate with other devices, such as client devices or other server network elements. Processor 402 is used to execute program 410, which may specifically perform the steps described in the aforementioned embodiment of the smart terminal-based course recommendation method.

[0094] Specifically, the program 410 may include program code including computer-executable instructions.

[0095] Processor 402 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The one or more processors included in the device may be processors of the same type, such as one or more CPUs, or may be processors of different types, such as one or more CPUs and one or more ASICs.

[0096] The memory 406 is used to store the program 410. The memory 406 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0097] Program 410 may be specifically invoked by processor 402 to cause the device to perform the following operations:

[0098] Get the student dataset;

[0099] The student dataset is processed according to the natural nearest neighbor algorithm to obtain a data relationship set, an undirected weighted graph is constructed based on the data relationship set, and the composition and number of each connected subgraph of the undirected weighted graph are determined, and then the connected subgraphs are merged to reduce the number of clusters to a preset number k;

[0100] The similarity matrix is ​​constructed by Gaussian kernel function, and the similarity matrix is ​​clustered by NJW spectral clustering algorithm to obtain clustering results;

[0101] Generating a learning tag set of students according to the clustering results, and generating a course recommendation sequence from the learning tag set;

[0102] The completion probability of each course in the course recommendation sequence is calculated through the preset Pine and Cypress model, and the course recommendation sequence and the completion probability of each course are pushed to the user; wherein, the Pine and Cypress distribution parameters of the preset Pine and Cypress model are any one or more of the student's academic performance, the completion degree of courses in the same field, and the student's interest level.

[0103] In some implementations: the smart terminal can be a computer, tablet, learning machine or other course content output device. Before the course content is output, the course recommendation content can be generated based on the student information. The user can then receive and enter the course based on the recommended content. In a specific example, see Figure 5 , the smart terminal is the learning panel 500.

[0104] Example 4:

[0105] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, wherein the storage medium stores at least one executable instruction. When the executable instruction is executed on the above-mentioned intelligent course recommendation device, the intelligent course recommendation device performs the operations of the above-mentioned intelligent terminal-based course recommendation method.

[0106] The student data set of the present invention includes student basic data and student behavior data. By clustering the student multidimensional data, the data source of the clustering model is richer and covers student basic data and student behavior data, making the course recommendation results more accurate. The clustering method of the present invention first accurately divides the neighborhood of each sample through the natural nearest neighbor relationship without setting parameters, can search based on the characteristics of the data set itself, and is less affected by discrete points. The student data set is clustered in combination with the Gaussian kernel function and the NJW spectral clustering algorithm to more accurately reflect the structural relationship of the data set, thereby obtaining a better clustering result. The present invention uses the Apriori algorithm to mine association rules. Through this solution, the accurate correlation between the various learning behavior data of the students themselves can be obtained, and on this basis, accurate recommendations for learning courses are made to students. The present invention also calculates the completion probability of each course in the course recommendation sequence by presetting the pine and cypress model, and reflects the completion probability of each course in the course recommendation sequence, providing another dimension of reference for students when making course selections, making the method more humane.

[0107] The algorithms or displays provided herein are not inherently related to any particular computer, virtual system or other device. In addition, the embodiments of the present invention are not directed to any particular programming language.

[0108] In the description provided herein, numerous specific details are set forth. However, it is understood that embodiments of the present invention may be practiced without these specific details. Similarly, in order to streamline the present invention and aid in understanding one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, various features of embodiments of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. The claims that follow the detailed description are hereby expressly incorporated into that detailed description, with each claim itself serving as a separate embodiment of the present invention.

[0109] Those skilled in the art will appreciate that the modules in the devices of the embodiments can be adaptively changed and installed in one or more devices different from the embodiments. The modules, units, or components in the embodiments can be combined into one module, unit, or component, and furthermore, they can be divided into multiple submodules, subunits, or subcomponents. Except that at least some of such features and / or processes or units are mutually exclusive.

[0110] It should be noted that the above embodiments illustrate rather than limit the invention, and that alternative embodiments may be devised by a person skilled in the art without departing from the scope of the appended claims. In the claims, any reference signs placed between brackets should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention may be implemented by means of hardware comprising several different elements and by means of appropriately programmed computers. In a unit claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third etc. does not indicate any order. These words may be interpreted as names. The steps in the above embodiments should not be understood as limiting the order of execution unless otherwise specified.

Claims

1. A course recommendation method based on intelligent terminals. It is characterized in that The method comprises: Get the student dataset; The student data set is processed according to the natural nearest neighbor algorithm to obtain a data relationship set, an undirected weighted graph is constructed according to the data relationship set, and the composition and number of each connected subgraph of the undirected weighted graph are determined, and then the connected subgraphs are merged and reduced to a preset cluster number k; The similarity matrix is ​​constructed by Gaussian kernel function, and the similarity matrix is ​​clustered by NJW spectral clustering algorithm to obtain clustering results; Generating a learning tag set of students according to the clustering results, and generating a course recommendation sequence from the learning tag set; The completion probability of each course in the course recommendation sequence is calculated by a preset pine and cypress model, and the course recommendation sequence and the completion probability of each course are pushed to the user; wherein the pine and cypress distribution parameters of the preset pine and cypress model are any one or more of the student's academic performance, the completion degree of courses in the same field, and the student's interest level; The student data set includes student basic data and student behavior data.

2. According to the course recommendation method based on the intelligent terminal according to claim 1, It is characterized in that The student basic data includes at least one or more of name, gender, age, major, class, contact information, and place of origin; and the student behavior data includes at least one or more of interests, career planning, basic course learning status, professional course learning status, and interest course learning status.

3. According to the course recommendation method based on the intelligent terminal according to claim 2, It is characterized in that The step of constructing an undirected weighted graph according to the data relationship set specifically includes: Taking the data relationship set NaN1 as the edge relationship set, and each data set of the student data set as the vertex relationship set, the undirected weighted graph is: NSG = (V, NaN1), wherein the undirected weighted graph is composed of many connected subgraphs, and there are one or more edges between any two points in the same connected subgraph that directly or indirectly connect the two points.

4. According to the course recommendation method based on the intelligent terminal according to claim 3, It is characterized in that The determining of the composition and number of each connected subgraph of the phase-free weighted graph, and then reducing the connected subgraphs to a preset number of k by merging the connected subgraphs, specifically includes: Acquire the number of connected subgraphs according to the undirected weighted graph, and compare the number of connected subgraphs with the preset number of clusters; If the number of connected subgraphs is greater than the preset number of clusters, a relationship edge is inserted between two preset vertices, and the undirected weighted graph and the number of connected subgraphs are updated until the number of connected subgraphs is equal to the preset number of clusters.

5. According to the course recommendation method based on the intelligent terminal according to claim 4, It is characterized in that The similarity matrix is ​​constructed by using the Gaussian kernel function and is calculated by the following formula: Among them, max{d ih :(i,h)∈NaN1} is the longest edge in the same connected subgraph.

6. The course recommendation method based on the intelligent terminal according to claim 5, It is characterized in that The similarity matrix is ​​clustered by the NJW spectral clustering algorithm to obtain the clustering result, specifically including: Calculating a degree matrix through a similarity matrix, and obtaining a Laplace matrix according to the degree matrix and the similarity matrix; Compute the first k largest eigenvectors of the Laplacian matrix {z 1 ,z 2 ,…,z k }, create matrix Z and standardize matrix Z to obtain matrix Y; All data points y in the i-th row of matrix Y are transformed into i Clustering results 1 ,A 2 ,…,A k }.

7. The course recommendation method based on the intelligent terminal according to claim 6, It is characterized in that Generating a learning tag set of students according to the clustering result, and generating a course recommendation sequence from the learning tag set, including: Extracting common labels from each cluster of the clustering results as learning labels for students, and generating a learning label set; For the learning label items in the learning label set, the Apriori algorithm is used to determine the frequent item sets therein, and each frequent item set k generates 2*(2k-1) association rules. Each association rule is sorted according to the confidence level to obtain a course recommendation sequence.

8. An intelligent course recommendation device, It is characterized in that The device comprises: Data acquisition module, used to obtain student data sets; The clustering calculation module is used to process the student data set according to the natural nearest neighbor algorithm to obtain a data relationship set, construct an undirected weighted graph according to the data relationship set, and determine the composition and number of each connected subgraph of the undirected weighted graph, and then reduce the connected subgraphs to a preset cluster number k by merging them with each other; construct a similarity matrix through a Gaussian kernel function, and cluster the similarity matrix through the NJW spectral clustering algorithm to obtain a clustering result A course recommendation module, used to generate a set of learning labels for students according to the clustering results, and generate a course recommendation sequence from the set of learning labels; The data push module is used to calculate the completion probability of each course in the course recommendation sequence through the preset Pine and Cypress model, and push the course recommendation sequence and the corresponding completion probability to the user; wherein the Pine and Cypress distribution parameters of the preset Pine and Cypress model are any one or more of the student's academic performance, the completion degree of courses in the same field, and the student's interest level.

9. A smart terminal, It is characterized in that The device comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform the operation of the course recommendation method based on the intelligent terminal as described in any one of claims 1-7.

10. A computer-readable storage medium, It is characterized in that The storage medium stores at least one executable instruction. When the executable instruction is executed on the intelligent course recommendation device according to claim 9, the intelligent course recommendation device executes the operation of the course recommendation method based on the intelligent terminal according to any one of claims 1 to 8.

Images