A method, apparatus, device, and storage medium for mining autonomous learning behavior patterns

Abstract

This invention relates to the field of autonomous learning technology, specifically to a method, apparatus, device, and storage medium for mining autonomous learning behavior patterns. The method includes: acquiring autonomous learning data; annotating the autonomous learning data using contextual attention metadata specifications and experiential application programming interface (API) learning standards to obtain autonomous learning behavior sequences; extracting feature vectors from the autonomous learning behavior sequences using a time- and context-optimized sequence graph transformation algorithm; and performing frequent pattern mining on the feature vectors to obtain the autonomous learning behavior pattern mining results. This invention comprehensively enriches the data dimensions by annotating autonomous learning data using contextual attention metadata specifications and experiential application programming interface (API) learning standards, making the representation of learning behaviors more comprehensive and accurate. Optimizing the sequence graph transformation algorithm based on context and time factors effectively improves the accuracy and efficiency of subsequent mining, making the mining results more practical and personalized.

Description

Technical Field

[0001] This invention relates to the field of autonomous learning technology, specifically to a method, apparatus, device, and storage medium for mining autonomous learning behavior patterns. Background Technology

[0002] Self-directed learning refers to learners exploring and discovering the relationships between learning content through their own efforts and experiences, and improving their learning abilities through this exploration process. Self-directed learning is a highly efficient and effective learning method. Its core idea is the learner's initiative and enthusiasm. Compared to traditional education methods, self-directed learning better meets learners' needs and adapts to the rapid development of society.

[0003] Self-directed learning is gaining increasing importance in today's information age because it can promote individual self-development and adaptability to changing environments, making it a key issue in education, training, and personal development. However, current research on self-directed learning still has a number of shortcomings, especially in mining behavioral patterns of self-directed learning, where insufficient use of data affects the accuracy of behavioral pattern mining. Summary of the Invention

[0004] In view of this, the present invention provides a method, apparatus, device and storage medium for mining autonomous learning behavior patterns, in order to solve the problem that the accuracy of behavior pattern mining is affected by insufficient utilization of data.

[0005] In a first aspect, the present invention provides a method for mining autonomous learning behavior patterns, the method comprising: acquiring autonomous learning data; annotating the autonomous learning data using contextual attention metadata specifications and experiential application programming interface learning standards to obtain autonomous learning behavior sequences; extracting feature vectors of the autonomous learning behavior sequences using a time- and context-optimized sequence graph transformation algorithm; and performing frequent pattern mining on the feature vectors to obtain autonomous learning behavior pattern mining results.

[0006] The self-directed learning behavior pattern mining method provided in this invention comprehensively enriches the data dimensions by annotating self-directed learning data using contextual attention metadata specifications and experiential application programming interface learning standards, resulting in a more comprehensive and accurate representation of learning behavior. Simultaneously, the sequence graph transformation algorithm is optimized based on contextual and temporal factors, and the optimized algorithm is used for feature vector extraction, effectively improving the accuracy and efficiency of subsequent mining and making the mining results more practical and personalized. Therefore, this method fills the gaps in existing self-directed learning research and provides strong technical support for promoting the in-depth development of self-directed learning.

[0007] In one optional implementation, before extracting the feature vector of the autonomous learning behavior sequence using a time- and context-optimized sequence graph transformation algorithm, the method further includes: processing the autonomous learning behavior sequence based on point sequence splitting and segment sequence splitting to obtain the processed sequence.

[0008] In this embodiment, the autonomous learning behavior sequence is processed by point sequence splitting and segment sequence splitting, which can realize the splitting of the sequence and avoid the problem of parallel sequences or indistinguishable elements in the sequence, which would lead to invalid feature extraction of the sequence.

[0009] In one optional implementation, the autonomous learning data is labeled using the contextual attention metadata specification and the experiential application programming interface (API) learning standard to obtain an autonomous learning behavior sequence. This includes: using the experiential API learning standard to determine the basic statement corresponding to the autonomous learning data, where the statement includes the operator, verb, and object; and based on the autonomous learning data, using the contextual attention metadata specification to determine the attributes of the basic statement to obtain an autonomous learning behavior sequence, where the attributes include verb attributes, object attributes, and conversation attributes.

[0010] In this embodiment, the basic statements corresponding to the self-learning data are first determined by using the experiential application programming interface learning standard, and then the attributes of the basic statements are determined by using the contextual attention metadata specification, thereby making the representation of learning behavior more comprehensive and accurate.

[0011] In one optional implementation, a time- and context-optimized sequence graph transformation algorithm is used to extract feature vectors of autonomous learning behavior sequences. This includes: performing a first dynamic assignment of hyperparameters to the sequence graph transformation algorithm based on the time factor of the autonomous learning behavior sequence; performing a second dynamic assignment of hyperparameters to the sequence graph transformation algorithm based on the context factor of the autonomous learning behavior sequence; optimizing the hyperparameters in the distance metric function of the sequence graph transformation algorithm using the first and second dynamic assignments to obtain an optimized sequence graph transformation algorithm; and extracting feature vectors of the autonomous learning behavior sequences based on the optimized sequence graph transformation algorithm.

[0012] In this embodiment, an optimized sequence graph transformation algorithm is used for feature vector extraction. When optimizing the sequence graph transformation algorithm, its hyperparameters are dynamically assigned based on the time and contextual factors of the autonomous learning behavior sequence. Thus, compared with the original algorithm, the optimized sequence graph transformation algorithm considers more factors besides the relative position information of sequence elements. For autonomous learning behavior sequences, it includes conversation time factors and contextual factors, thereby improving the accuracy of behavior sequence feature extraction.

[0013] In one optional implementation, frequent pattern mining is performed on the feature vectors to obtain autonomous learning behavior pattern mining results, including: clustering the feature vectors to obtain a clustered dataset; and performing frequent pattern mining on the clustered dataset to obtain autonomous learning behavior pattern mining results.

[0014] In this embodiment, when performing frequent pattern mining, the feature vectors are first clustered, which can improve the efficiency of frequent pattern mining of feature vectors.

[0015] In one alternative implementation, acquiring self-directed learning data includes: acquiring self-directed learning data based on a social cognitive self-directed learning process model, online learning platforms, offline classrooms, self-reports, and physiological measurement data.

[0016] In this embodiment, autonomous learning data is obtained based on the social cognitive autonomous learning process model, online learning platform, offline classroom, self-report and physiological measurement data, thereby making the obtained autonomous learning data include data from multiple sources and comprehensively enriching the data dimensions.

[0017] In one optional implementation, autonomous learning data is obtained based on a social cognitive autonomous learning process model, online learning platforms, offline classrooms, self-reports, and physiological measurement data. This includes: dividing autonomous learning behavior into three stages: pre-planning, execution, and reflection; determining secondary indicators for each stage; determining the data types corresponding to the secondary indicators and the learning experiences corresponding to the data types based on the social cognitive autonomous learning process model, online learning platforms, offline classrooms, self-reports, and physiological measurement data; and determining autonomous learning data based on the learning experiences.

[0018] In this embodiment, when acquiring self-learning data, the learning stages are first divided, and then the secondary indicators corresponding to each stage and the learning experience corresponding to each secondary indicator are determined according to the divided stages, thereby determining the self-learning data. This makes the determined self-learning data more comprehensive and accurate.

[0019] Secondly, the present invention provides a device for mining autonomous learning behavior patterns. The device includes: a data acquisition module for acquiring autonomous learning data; an annotation module for annotating the autonomous learning data using contextual attention metadata specifications and experiential application programming interface learning standards to obtain autonomous learning behavior sequences; an extraction module for extracting feature vectors from the autonomous learning behavior sequences using a time- and context-optimized sequence graph transformation algorithm; and a mining module for performing frequent pattern mining on the feature vectors to obtain autonomous learning behavior pattern mining results.

[0020] Thirdly, the present invention provides a computer device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the autonomous learning behavior pattern mining method of the first aspect or any corresponding embodiment described above.

[0021] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the autonomous learning behavior pattern mining method of the first aspect or any corresponding embodiment described above. Attached Figure Description

[0022] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating the autonomous learning behavior pattern mining method according to an embodiment of the present invention;

[0024] Figure 2 This is a descriptive model for autonomously learning multi-source data based on xAPI and CAM specifications according to an embodiment of the present invention;

[0025] Figure 3 This is a flowchart illustrating another method for mining autonomous learning behavior patterns according to an embodiment of the present invention;

[0026] Figure 4 This is a structural block diagram of an autonomous learning behavior pattern mining device according to an embodiment of the present invention;

[0027] Figure 5 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] According to an embodiment of the present invention, an embodiment of a method for mining autonomous learning behavior patterns is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0030] This embodiment provides a method for mining autonomous learning behavior patterns, which can be used in electronic devices such as computers, mobile phones, and tablets. Figure 1 This is a flowchart of a method for mining autonomous learning behavior patterns according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps:

[0031] Step S101: Obtain self-directed learning data. Specifically, self-directed learning data includes learner behavioral data during the self-directed learning process, such as how learners learn and what content they learn. This data can be generated online or offline. Furthermore, learners' physiological data during the learning process can be monitored to better understand and analyze their self-directed learning behavior.

[0032] Step S102 involves annotating the self-learning data using the Contextualized Attention Metadata (CAM) specification and the Experience API (xAPI) learning standard to obtain a sequence of self-learning behaviors. The Experience API (xAPI) learning standard is a next-generation learning technology specification that can collect learning behavior data (online and offline), breaking the limitation of only recording the courseware reading process. The acquired self-learning data can be recorded using xAPI. During recording, xAPI reports and sends statements to the learning record storage in the form of "noun + verb + object" (i.e., "someone - did - what"). The Contextualized Attention Metadata (CAM) specification is data used to describe a series of user behaviors in a specific context. Therefore, CAM can be used to add contextual descriptions to the statements recorded by xAPI, thereby annotating the self-learning data and obtaining a sequence of self-learning behaviors. This sequence describes multiple complete self-learning behaviors contained in the self-learning data.

[0033] Step S103 involves extracting feature vectors from the autonomous learning behavior sequence using a time- and context-optimized sequence graph transform algorithm. Specifically, the sequence graph transform (SGT) algorithm can extract sequence features based on the relative position information between sequences without increasing computational cost. The specific extraction process can be implemented using relevant technologies and will not be elaborated here. In this embodiment, in addition to considering the relative position information of sequence elements, the influence of time and context is further considered; that is, the sequence graph transform algorithm is optimized using time and context, and the optimized algorithm is used to extract feature vectors. This improves the accuracy of feature vector extraction.

[0034] Step S104 involves performing frequent pattern mining on the feature vectors to obtain the autonomous learning behavior pattern mining results. Specifically, frequent pattern mining refers to mining patterns that frequently occur simultaneously in the feature vectors. Frequent pattern mining mainly includes generating candidate itemsets and calculating support, then selecting itemsets with support greater than a threshold as frequent patterns. In this embodiment, a frequent pattern mining algorithm from related technologies can be used to perform frequent pattern mining, thereby obtaining the autonomous learning behavior pattern mining results.

[0035] The self-directed learning behavior pattern mining method provided in this invention comprehensively enriches the data dimensions by annotating self-directed learning data using contextual attention metadata specifications and experiential application programming interface learning standards, resulting in a more comprehensive and accurate representation of learning behavior. Simultaneously, the sequence graph transformation algorithm is optimized based on contextual and temporal factors, and the optimized algorithm is used for feature vector extraction, effectively improving the accuracy and efficiency of subsequent mining and making the mining results more practical and personalized. Therefore, this method fills the gaps in existing self-directed learning research and provides strong technical support for promoting the in-depth development of self-directed learning.

[0036] This embodiment provides a method for mining autonomous learning behavior patterns, the process of which includes the following steps:

[0037] Step S201: Obtain self-learning data. For details, please refer to [link to relevant documentation]. Figure 1 Step S101 of the illustrated embodiment will not be described again here.

[0038] Step S202 involves labeling the self-learning data using the contextual attention metadata specification and the experiential application programming interface learning standard to obtain a sequence of self-learning behaviors. For details, please refer to [link to relevant documentation]. Figure 1 Step S102 of the illustrated embodiment will not be described again here.

[0039] Step S203 involves processing the self-learning behavior sequence based on point sequence splitting and segment sequence splitting to obtain the processed sequence. Specifically, self-learning behavior sequences formed based on CAM and xAPI specifications suffer from sequence overlap. For example, multiple self-learning behavior sequences may be contained within the same time point or time period. This problem arises because the smallest granularity of the sequence is a single behavior of a student over a continuous period of time, while the smallest granularity of the annotation is the generation of a single behavior. Learners may generate multiple behaviors at the same time, and the generation of a single behavior may be contained within multiple time periods. These situations can lead to sequence overlap and the inability to distinguish the types of elements within the sequence, resulting in ineffective feature extraction and consequently affecting frequent pattern mining. Therefore, to address the above problems, this embodiment employs point sequence splitting and segment sequence splitting to process the self-learning behavior sequence.

[0040] Point sequence decomposition involves breaking down multiple self-directed learning behaviors occurring simultaneously within a given time frame, with each behavior having a time interval (TimeInfo) of 0. From a behavior sequence perspective, this can be interpreted as students rapidly performing multiple self-directed learning behaviors. Specifically, a learner may generate multiple self-directed learning behaviors within a given time frame. Point sequence decomposition can separate these behaviors. In this decomposition, a student's attention being focused on a specific screen and clicking on an object within the same time frame can be considered a single self-directed learning behavior, thus enabling the breakdown of multiple self-directed learning behaviors occurring simultaneously within a given time frame.

[0041] When splitting a sequence into segments, after traversing all sequences, sequences with overlapping time intervals (TimeInfo) are split according to the shortest possible time interval, thus transforming long, overlapping sequences into short, non-overlapping sequences. The shortest possible time interval here refers to the time interval corresponding to the smallest sequence among the different overlapping time intervals.

[0042] Therefore, the point sequence splitting described above can split multiple self-learning behavior sequences that may be contained within the same time point or time period. The segment sequence splitting described above can split sequences that may be generated by learners at the same time, or sequences that may be generated in multiple time periods. Thus, point sequence splitting and segment sequence splitting can solve the problems of parallel sequences and indistinguishable elements in the sequence.

[0043] Step S204: A time- and context-optimized sequence graph transformation algorithm is used to extract feature vectors from the autonomous learning behavior sequence; for details, please refer to [link to relevant documentation]. Figure 1 Step S103 of the illustrated embodiment will not be described again here.

[0044] Step S205 involves performing frequent pattern mining on the feature vectors to obtain the results of autonomous learning behavior pattern mining. For details, please refer to [link to relevant documentation]. Figure 1 Step S104 of the illustrated embodiment will not be described again here.

[0045] This embodiment provides a method for mining autonomous learning behavior patterns, which includes the following steps:

[0046] Step S301: Obtain self-learning data.

[0047] Specifically, step S301 includes:

[0048] Step S3011 involves acquiring self-directed learning data based on the social cognitive self-directed learning process model, online learning platforms, offline classrooms, self-reports, and physiological measurement data. Specifically, the social cognitive self-directed learning process model can be based on Zimmerman's model, which posits that self-directed learning involves the interaction between self, behavior, and environment. This model provides an in-depth analysis of the components and underlying psychological mechanisms of self-directed learning, offering a valuable entry point for studying students' self-directed learning. According to the self-directed learning model theory proposed in this model, only when learners can effectively utilize strategies to control the three elements of self, behavior, and environment can it be considered self-directed learning, and only then can the learner be considered a self-directed learner, and only then can the strategies employed by the learner be considered self-directed learning strategies. Therefore, when acquiring self-directed learning data based on this model, data on self, behavior, and environment can be obtained.

[0049] Online learning platforms can be platforms that enable learners to engage in self-directed learning, such as online learning websites where learners can log in and study independently. Offline classrooms can be classroom settings where teachers and learners conduct face-to-face instruction and learning. Self-reports can be summaries and reflections on the learning process by learners after completing it. Physiological measurement data can be changes in students' physiological parameters during the learning process, which can more accurately reflect their self-directed learning status.

[0050] In an optional implementation, step S3011 includes:

[0051] Step a1 divides self-directed learning behavior into three stages: planning, execution, and reflection.

[0052] Step a2: Determine the secondary indicators corresponding to each stage.

[0053] Step a3: Based on the social cognitive autonomous learning process model, online learning platform, offline classroom, self-report and physiological measurement data, determine the data types corresponding to the secondary indicators and the learning experiences corresponding to the data types.

[0054] Step a4: Determine self-directed learning data based on learning experience.

[0055] Specifically, to more accurately acquire self-directed learning data, self-directed learning behavior is divided into three stages: pre-planning, execution, and reflection, as shown in Table 1. Pre-planning includes secondary indicators such as goal setting, developing a learning plan, and selecting strategies, manifested as changes in students' electroencephalograms (EEGs) of alpha, beta, and theta waves, and text descriptions. The execution stage includes implementing the plan and self-monitoring, manifested as focusing attention on accessing resources, seeking help from teachers, and specific learning experiences such as changes in heart rate variability (HRV), galvanic skin response (GSR), and alpha, beta, and theta waves. The self-reflection stage includes self-evaluation and strategy adjustment, manifested as self-satisfaction ratings on scales and writing reflection logs.

[0056] Table 1

[0057]

[0058] Therefore, after generating Table 1 above, we can acquire self-learning data based on the relevant data in Table 1.

[0059] Step S302: The autonomous learning data is labeled using the contextual attention metadata specification and the experiential application programming interface learning standard to obtain the autonomous learning behavior sequence.

[0060] Specifically, step S302 includes:

[0061] Step S3021: The experiential application programming interface (API) learning standard is used to determine the basic statement corresponding to the self-directed learning data. The statement includes an operator, a verb, and an object. Specifically, the basic statement can be represented in the form of <Actor + Verb + Object>. The Actor (also called the executor) is mainly a single learner, with each learner corresponding to a unique student ID. The basic statement describes the behavioral data of a single learner's self-directed learning. The Verb represents the action taken by the executor. This embodiment summarizes 35 commonly used behavioral verbs, as shown in Table 2 below.

[0062] Table 2

[0063] viewed read reviewed created defined estimated-duration performed-offline started paused suspended resumed completed scheduled talked annotated rated favorited selected shared unshared unfavorited discarded modified-annotation set earned expected enabled disabled personalized arranged removed displayed

[0064] The Object refers to the object of learner activities, including four categories: online learning platform, offline classroom, self-report, and physiological indicators. In this embodiment, a total of 28 objects are formed, as shown in Table 3. The Simple collection represents a mixture of two types of objects, and the Mixed collection includes a mixture of more than two types of objects.

[0065] Table 3

[0066] slide deck file slide game media book module course device note idea project goal task performance step strategy event doubt solution to-review reward resource plan vocabulary-word

[0067] Step S3022: Based on the autonomous learning data, the attributes of the basic statement are determined using the contextual attention metadata specification to obtain the autonomous learning behavior sequence. The attributes include verb attributes, object attributes, and session attributes. Specifically, the autonomous learning behavior sequence obtained after adding attributes to the basic statement can be represented as [In a session[ActorVerb[in a context]Object[from a feed](Optional key)]]. This sequence can be adopted... Figure 2 The diagram illustrates a descriptive model for multi-source data (autonomous learning data) for autonomous learning based on the xAPI and CAM specifications. Here, `context` exists as a property of the `Verb` property, describing the context in which the learner's behavior occurs. For example, if a learner uploads a reflection document to a course within a learning management system, the `context` will contain basic information about that course. `feed` exists as a property of the `Object` property, describing information such as the creation, editing, and modification of electronic documents, or information like the subject, difficulty, and duration of an exam paper. `session` represents the system session information for this learning experience, describing learner information (`ActorInfo`), session time (`TimeInfo`), and learning space or system information (`SysInfo`). Additionally, optional properties such as `result`, `id`, and `authority` can be added. Based on the above description, the resulting sequence of autonomous learning behaviors could be: during a 45-minute math class (`SysInfo`) (`session`), 9th-grade student Xiaoming (`ActorInfo`) discusses a learning task about solving quadratic functions (`feed`) with classmates in a group discussion (`context`).

[0068] Step S303 involves extracting feature vectors from the autonomous learning behavior sequences using a time- and context-optimized sequence graph transformation (SGT) algorithm. In the definition of the Sequence Graph Transformation (SGT) algorithm, a sequence set S is introduced, with each sequence labeled s. These sequences consist of a set of elements V called an "alphabet" or event set. Within this framework, s represents the index position of a specific event in the event set V. A function φ is defined. k (d) is a function used to describe the relative position information between any two event elements in sequence s. In the function, k is a hyperparameter, and d represents the relative distance between the two event elements.

[0069] As can be seen from the above definition of the SGT algorithm, the core of the SGT algorithm lies in the function φ that calculates the relative position information of two event elements. k (d)

[0070] Specifically, step S303 includes:

[0071] i. Step S3031: The hyperparameters of the sequence graph transformation algorithm are dynamically assigned a first value based on the time factor of the autonomous learning behavior sequence. Specifically, when constructing the autonomous learning behavior sequence, session information including session time (TimeInfo) is considered. Therefore, when optimizing the SGT algorithm, the hyperparameter k can be dynamically assigned based on the session time (TimeInfo). The dynamic assignment of the hyperparameter k based on the TimeInfo session time factor is performed according to the following formula, where t... m , t l This represents the TimeInfo point in time where two sequence elements m and l occur. As can be seen from the following formula, the range of values ​​for k is [1, 10].

[0072] k l,m =10-9e -(m-tl)

[0073] Step S3032: The hyperparameters of the sequence graph transformation algorithm are dynamically assigned a second value based on contextual factors of the autonomous learning behavior sequence. Specifically, when constructing the autonomous learning behavior sequence, conversational information including contextual factors is considered. Therefore, when optimizing the SGT algorithm, the hyperparameter k can be dynamically assigned based on the contextual factors. The dynamic assignment of the hyperparameter k based on the contextual factors is performed according to the following formula, where z represents the frequency distribution of the context occurring between two sequence elements m and l, and n... l It is the frequency of occurrence of sequence element l in the context, n m It is the frequency of occurrence of sequence element m. From the following formula, we know that the range of k is [1, 10].

[0074] k l,m =10-9z

[0075]

[0076] Step S3033: The hyperparameters in the distance metric function of the sequence graph transformation algorithm are optimized using the first dynamic assignment and the second dynamic assignment to obtain the optimized sequence graph transformation algorithm; specifically, combining the first dynamic assignment and the second dynamic assignment, the distance metric function φ of the SGT algorithm is optimized using the following formula. k (d(l,m))= -k|m-l| The hyperparameter k is optimized, and the range of k is [1, 10] as shown by the following formula.

[0077] k l,m =10-9e -(tm-tl) -8z

[0078] Step S3034: Extract feature vectors from the autonomous learning behavior sequence based on the optimized sequence graph transformation algorithm. Specifically, the process of feature vector extraction based on the optimized SGT algorithm is similar to that of feature vector extraction using the SGT algorithm itself, and can be implemented with reference to the SGT algorithm; therefore, it will not be elaborated further here.

[0079] Step S304: Perform frequent pattern mining on the feature vector to obtain the autonomous learning behavior pattern mining results.

[0080] Specifically, step S304 includes:

[0081] Step S3041 involves clustering the feature vectors to obtain a clustered dataset. When clustering the feature vectors, the K-means clustering algorithm or other clustering algorithms can be used. This embodiment does not specifically limit the clustering algorithm used, as long as it achieves clustering. Furthermore, after clustering, the Calinskiharabaz coefficient and silhouette coefficient can be used to evaluate the clustering effect. The Calinskiharabaz coefficient calculates the quality of a cluster based on the dispersion of the clustering results and the separation between clusters. The silhouette coefficient combines the cohesion and separation of clusters, enabling the evaluation of the performance of the same clustering algorithm on different datasets or different clustering algorithms on the same dataset. Based on the evaluation of the clustering effect, clustering algorithms can be selected or evaluated.

[0082] Step S3042 involves performing frequent pattern mining on the clustered dataset to obtain the results of autonomous learning behavior pattern mining. Specifically, frequent pattern mining can be implemented using relevant algorithms, such as Latent Semantic Analysis (LSA). The specific mining process can be implemented by referring to relevant technologies, and will not be elaborated here.

[0083] As a specific application embodiment of the present invention, such as Figure 3 As shown, this method for mining self-directed learning behavior patterns can be implemented using the following process:

[0084] (1) Obtain online and offline self-directed learning data (self-directed learning behavior data), including offline classroom videos, questionnaires, online self-directed learning activity logs, eye movement, heart rate data, etc.

[0085] (2) Label the self-learning data based on the data specifications of CAM and xAPI.

[0086] (3) The self-learning behavior sequence (initial self-learning behavior sequence) obtained after annotation is processed by point sequence splitting or segment sequence splitting to optimize the behavior sequence and obtain the optimized self-learning behavior sequence.

[0087] (4) Use the optimized SGT algorithm to extract sequence feature vectors and obtain the sequence features of autonomous learning behavior.

[0088] (5) Clustering algorithm is used to cluster the sequence feature vectors to obtain the clustered autonomous learning behavior sequence features.

[0089] (6) Use LSA to perform frequent pattern mining on the clustered dataset to obtain frequent patterns of autonomous learning behavior, and analyze the mining results of frequent patterns.

[0090] This embodiment also provides a self-learning behavior pattern mining device, which is used to implement the above embodiments and preferred embodiments, and will not be repeated as already described. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0091] This embodiment provides a device for mining autonomous learning behavior patterns, such as... Figure 4 As shown, it includes:

[0092] Data acquisition module 41 is used to acquire self-learning data;

[0093] The annotation module 42 is used to annotate the autonomous learning data using the contextual attention metadata specification and the experiential application programming interface learning standard to obtain the autonomous learning behavior sequence;

[0094] Extraction module 43 is used to extract feature vectors of autonomous learning behavior sequences using a time- and context-optimized sequence graph transformation algorithm;

[0095] Mining module 44 is used to perform frequent pattern mining on feature vectors to obtain autonomous learning behavior pattern mining results.

[0096] In one optional embodiment, the apparatus further includes a sequence processing module for processing the autonomous learning behavior sequence based on point sequence splitting and segment sequence splitting to obtain a processed sequence.

[0097] In one optional implementation, the annotation module is specifically used to: determine the basic statements corresponding to the self-learning data using the experiential application programming interface learning standard, wherein the statements include the operator, verbs, and objects; and based on the self-learning data, determine the attributes of the basic statements using the contextual attention metadata specification to obtain a sequence of self-learning behaviors, wherein the attributes include the attributes of the verbs, the attributes of the objects, and the conversation attributes.

[0098] In one optional implementation, the extraction module is specifically used to: perform a first dynamic assignment on the hyperparameters of the sequence graph transformation algorithm based on the time factor of the autonomous learning behavior sequence; perform a second dynamic assignment on the hyperparameters of the sequence graph transformation algorithm based on the context factor of the autonomous learning behavior sequence; optimize the hyperparameters in the distance metric function of the sequence graph transformation algorithm using the first and second dynamic assignments to obtain an optimized sequence graph transformation algorithm; and extract the feature vector of the autonomous learning behavior sequence based on the optimized sequence graph transformation algorithm.

[0099] In one optional implementation, the mining module is specifically used to: cluster the feature vectors to obtain a clustered dataset; and perform frequent pattern mining on the clustered dataset to obtain autonomous learning behavior pattern mining results.

[0100] In one optional implementation, the data acquisition module includes an acquisition submodule for acquiring self-learning data based on a social cognitive self-learning process model, online learning platforms, offline classrooms, self-reports, and physiological measurement data.

[0101] In one optional implementation, the acquisition submodule is specifically used to: divide self-directed learning behavior into three stages: pre-planning, execution, and reflection; determine the secondary indicators corresponding to each stage; determine the data types corresponding to the secondary indicators and the learning experiences corresponding to the data types based on the social cognitive self-directed learning process model, online learning platform, offline classroom, self-report, and physiological measurement data; and determine self-directed learning data based on the learning experiences.

[0102] Further functional descriptions of the above modules are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0103] This invention also provides a computer device having the above-described features. Figure 4 The device shown is for mining autonomous learning behavior patterns.

[0104] Please see Figure 5 , Figure 5 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 5 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 5 Take a processor 10 as an example.

[0105] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0106] The memory 20 stores instructions executable by at least one processor 10 to cause at least one processor 10 to perform the method shown in the above embodiments.

[0107] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device as shown by a landing page for an app. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, which can be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0108] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0109] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0110] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0111] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0112] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for mining autonomous learning behavior patterns, characterized in that, The method includes: Acquire self-learning data; The autonomous learning data is labeled using the contextual attention metadata specification and the experiential application programming interface learning standard to obtain the autonomous learning behavior sequence; The feature vector of the autonomous learning behavior sequence is extracted using a time- and context-optimized sequence graph transformation algorithm. Frequent pattern mining is performed on the feature vectors to obtain the autonomous learning behavior pattern mining results; The feature vector of the autonomous learning behavior sequence is extracted using a time- and context-optimized sequence graph transformation algorithm, including: The hyperparameters of the sequence graph transformation algorithm are dynamically assigned based on the time factor of the autonomous learning behavior sequence. Specifically, the hyperparameters are dynamically assigned based on the session time, and the values ​​are determined according to the following formula: In the formula, This indicates the point in time at which the sessions occur between the two sequence elements m and l; The hyperparameters of the sequence graph transformation algorithm are dynamically assigned a second value based on contextual factors of the autonomous learning behavior sequence. The dynamic assignment of hyperparameters based on contextual factors is performed according to the following formula: In the formula, This represents the frequency distribution of scenarios occurring between two sequence elements m and l. It is a sequence element Frequency of occurrence It is the frequency of the context in which sequence element m occurs; The hyperparameters in the distance metric function of the sequence graph transformation algorithm are optimized by using the first dynamic assignment and the second dynamic assignment, resulting in the optimized sequence graph transformation algorithm; The feature vector of the autonomous learning behavior sequence is extracted based on the optimized sequence graph transformation algorithm.

2. The method according to claim 1, characterized in that, Before extracting the feature vector of the autonomous learning behavior sequence using a time- and context-optimized sequence graph transformation algorithm, the method further includes: The autonomous learning behavior sequence is processed based on point sequence splitting and segment sequence splitting to obtain the processed sequence.

3. The method according to claim 1, characterized in that, The autonomous learning data is labeled using the contextual attention metadata specification and the experiential application programming interface learning standard to obtain a sequence of autonomous learning behaviors, including: The experiential application programming interface learning standard is used to determine the basic statement corresponding to the self-learning data, and the statement includes the operator, verb and object; Based on the self-learning data, the attributes of the basic statements are determined using the contextual attention metadata specification to obtain the self-learning behavior sequence. The attributes include verb attributes, object attributes, and conversation attributes.

4. The method according to claim 1, characterized in that, Frequent pattern mining is performed on the feature vectors to obtain autonomous learning behavior pattern mining results, including: Cluster the feature vectors to obtain a clustered dataset; Frequent pattern mining is performed on the clustered dataset to obtain the results of autonomous learning behavior pattern mining.

5. The method according to claim 1, characterized in that, Acquire self-learning data, including: Self-directed learning data is obtained based on a social cognitive self-directed learning process model, online learning platforms, offline classrooms, self-reports, and physiological measurement data.

6. The method according to claim 1, characterized in that, Self-directed learning data was acquired based on a social cognitive self-directed learning process model, online learning platforms, offline classrooms, self-reports, and physiological measurement data, including: Self-directed learning behavior is divided into three stages: planning, execution, and reflection. Determine the secondary indicators corresponding to each stage; Based on the social cognitive autonomous learning process model, online learning platforms, offline classrooms, self-reports, and physiological measurement data, the data types corresponding to the secondary indicators and the learning experiences corresponding to the data types are determined. Self-directed learning data is determined based on the learning experience described.

7. A device for mining autonomous learning behavior patterns, characterized in that, The apparatus for performing the method as claimed in claim 1 includes: The data acquisition module is used to acquire self-learning data; The annotation module is used to annotate the autonomous learning data using the contextual attention metadata specification and the experiential application programming interface learning standard to obtain the autonomous learning behavior sequence; The extraction module is used to extract the feature vector of the autonomous learning behavior sequence using a time- and context-optimized sequence graph transformation algorithm; The mining module is used to perform frequent pattern mining on the feature vector to obtain the results of autonomous learning behavior pattern mining.

8. A computer device, characterized in that, include: A memory and a processor are interconnected, the memory stores computer instructions, and the processor executes the autonomous learning behavior pattern mining method according to any one of claims 1 to 6 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the autonomous learning behavior pattern mining method according to any one of claims 1 to 6.

Images