NL2026606B1

NL2026606B1 - Method and device for constructing an educational learning environment

Info

Publication number: NL2026606B1
Application number: NL2026606A
Authority: NL
Inventors: Zhang Hao; Wang Tai; Yang Zongkai; Liu Sannyuya; Liu Zhi; Zhou Dongbo; Sun Jianwen
Original assignee: Univ Central China Normal
Priority date: 2020-10-02
Filing date: 2020-10-02
Publication date: 2021-10-20

Abstract

The present disclosure provides a method and a device for constructing an educational learning environment, in which the method includes the following steps. A plurality of spaces in the educational learning environment and learning environment construction parameters of each of the spaces are predefined. A semantic network model of each of the spaces is constructed. Learning subject- related events and learning environment-related events are collected, and a data fusion model of a cyber-physical space semantic level consistency for learning subjects is constructed. An integrated embodied model based on learning scenarios is constructed. The learning environment construction parameters for the learning subjects are set according to the integrated embodied model. Therefore, the disclosure can dynamically customize an educational learning environment for the learning subjects.

Description

METHOD AND DEVICE FOR CONSTRUCTING AN EDUCATIONAL LEARNING ENVIRONMENT

TECHNICAL FIELD The present application relates to information technologies, and more particularly to a method and a device for constructing an educational learning environment.

BACKGROUND information technology develops very fast and has been widely applied in education, which facilitates the digitalization and intelligence of traditional teaching environment. Currently, the learning can be conducted offline in physical spaces such as classrooms, or conducted online using interactive whiteboards, portable tablets and other smart devices. Therefore, in students’ learning environments, offline physical spaces are fused with online spaces, such as digital classrooms, various intelligent terminal devices and teaching appliances.

However, in the prior art, the offline physical space and the online space are independent of each other without seamless interaction. An educational cyber- physical space fusion learning environment cannot be personalized according to students’ behaviors in any one of the spaces. For example, since offline education resources and online education resources are independent of each other, it is hard to adjust the online learning resources to adapt the offline learning as required.

SUMMARY In order to overcome the above-mentioned defects in the prior art, the present disclosure provides a method and a system for constructing an educational learning environment according to students’ behaviors.

In a first aspect, the present disclosure provides a method for constructing an educational learning environment, comprising: (1) predefining a plurality of spaces in the educational learning environment and learning environment construction parameters of each of the spaces, wherein the spaces comprise at least two of a physical space, a network space, a resource space, and a social space; wherein the step 1 comprises: defining the educational learning environment, and clarifying the spaces in the educational learning environment, wherein the spaces at least include: the physical space, the network space, the resource space and the social space; defining the spaces in the educational learning environment, and clarifying meaning of the spaces, a basis for providing the spaces, and contents of the educational learning environment supported in the spaces; and defining presentation forms and presentation mediums of different spaces in a preset learning scenario; (2) predefining service contents of the spaces, and constructing a semantic network model of each of the spaces according to the service contents; (3) collecting learning subject-related events and learning environment-related events, and constructing a data fusion model of a cyber-physical space semantic level consistency for learning subjects; (4) constructing an integrated embodied model based on learning scenarios, wherein the integrated embodied model is for describing relationships between the learning subjects and the learning environment construction parameters in the learning scenarios, and is capable of being dynamically adjusted according to learning subject-related events or learning environment-related events collected in any one of the spaces, and (5) setting the learning environment construction parameters for the learning subjects according to the integrated embodied model.

In an embodiment, the step (2) comprises: (2.1) predefining the service contents of the spaces according to the learning scenarios; performing a semantic calibration on the service contents of each of the spaces; establishing semantic subject units of the service contents; and determining semantic units of the service contents; (2.2) analyzing semantic relationships of the service contents of the spaces; determining sequence relationships, hierarchical relationships and inclusion relationships between the semantic subject units; and constructing a semantic relationship table; and (2.3) constructing a semantic network organization structure according to the semantic relationship table for each of the semantic units of the service contents.

in an embodiment, the step (3) comprises: (3.1) creating embodied objects of a learning subject in different spaces, and determining a consistent expression of the same learning subject in different spaces; (3.2) collecting the learning subject-related events and the learning environment-related events, and converting the learning subject-related events and learning environment-related events into data according to the semantic network model to form a learning behavior data pool; and (3.3) constructing the data fusion model of the cyber-physical space semantic level consistency based on the learning behavior data pool.

In an embodiment, the step (3.3) comprises: extracting and classifying learning behavior data in the learning behavior data pool using a long short-term memory model and a convolutional neural network with three channels comprising a spatial domain, a local time domain and a global time domain, to construct the data fusion model of the cyber-physical space semantic level consistency.

In an embodiment, the step (4) comprises: (4.1) converting the learning subject-related events and the learning environment-related events into standard format data according to the semantic network model and the data fusion model; and (4.2) converting relationships between the learning environment-related events and the learning subject-related events into a graph network model; expressing spatial environments using vertexes of the graph network model, and expressing embodied relationships among different spaces using edges of the graph network model; constructing the integrated embodied model based on the learning scenarios using the graph network mode, wherein the integrated embodied model is used for describing the relationships between the learning subjects and the learning environment construction parameters in the learning scenarios, and is capable of being dynamically adjusted according to learning subject-related events or learning environment-related events collected in any one of the spaces.

in an embodiment, the learning environment construction parameters are one or more of drive parameters of hardware or software of the physical space, the network space or the social space, or resource acquisition parameters of the resource space.

In a second aspect, the present disclosure provides a device for constructing an educational learning environment, comprising: a space predefining module for predefining a plurality of spaces in the educational learning environment and learning environment construction parameters of each of the spaces, wherein the spaces comprise at least two of a physical space, a network space, a resource space, and a social space; the space predefining module is configured to define the educational learning environment, and clarify the spaces in the educational learning environment, wherein the spaces at least include: the physical space, the network space, the resource space and the social space; the space predefining module is further configured to define the spaces in the learning situation, and clarify meaning of the spaces, a basis for providing the spaces, and contents of the educational learning environment supported in the spaces; and the space predefining module is further configured to define presentation forms and presentation mediums of different spaces in a preset learning scenario; a semantic network model constructing module for predefining service contents of the spaces, and constructing a semantic network model of each of the spaces according to the service contents; a data fusion model constructing module for collecting learning subject-related events and learning environment-related events, and constructing a data fusion model of a cyber-physical space semantic level consistency for learning subjects; an integrated embodied model constructing module for constructing an integrated embodied model based on learning scenarios, wherein the integrated embodied model is for describing relationships between the learning subjects and the learning environment construction parameters in the learning scenarios, and is capable of being dynamically adjusted according to learning subject-related events or learning environment-related events collected in any one of the spaces; and a learning environment constructing module for setting the learning environment construction parameters for the learning subjects according to the integrated embodied model.

Compared to the prior art, the present invention has the following beneficial effects.

(1) Based on spaces involved in education and teaching, the present invention analyzes the physical space, network space, resource space, social space, etc. involved in the educational scenarios, and explores a semantic analysis method based on the contents of education and teaching services to realize an educational cyber-physical space unified expression and data fusion method with a consistent learning subject, which can construct an educational cyber-physical space fusion learning environment according to students’ behaviors in any one of the spaces.

(2) The present invention introduces a multi-level data fusion method into the application of educational scenarios, making it possible to construct a consistent cyber-physical space learning environment for learners.

(3) The present invention proposes a learning space fusion method for learners, and expression methods such as semantic network model and traditional independent spatial expression methods are fused.

(4) The present disclosure constructs a personalized cyber-physical space 5 learning environment and designs an educational cyber-physical space fusion system for learners, which supports a personalized learning environment under different physical conditions

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a flowchart of a method for constructing an educational learning environment according to an embodiment of the present disclosure.

FIG. 2 schematically shows an application of the method for constructing the educational learning environment according to the embodiment of the present disclosure.

FIG. 3 schematically shows a semantic network model according to an embodiment of the present disclosure.

FIG. 4 schematically shows another semantic network model according to another embodiment of the present disclosure.

FIG. 5 schematically shows the application of a device for constructing the educational learning environment according to an embodiment of the present disclosure.

FIG. 6 schematically shows the device for constructing the educational learning environment according to the embodiment of the present disclosure.

DETAILED DESCRIPTION OF EMBODIMENTS In order to make the object, technical solutions and advantages of the disclosure more clearly, the present disclosure will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the embodiments described herein are only intended to explain the present disclosure, but not to limit the scope of the present disclosure. In addition, the technical features involved in various embodiments of the present disclosure described below can be combined with each other as long as there is no conflict with each other.

The technical solutions provided by the embodiments of the present disclosure can be applied to education informatization, so that students can not only learn in offline physical spaces such as classrooms, but also use interactive whiteboards, portable tablets and other smart devices for online learning. Therefore, in students’ learning environments, offline physical spaces are fused with network spaces, resource spaces and social spaces, such as digital classrooms, various intelligent terminal devices and teaching appliances, etc., so that an educational cyber-physical space learning environment is personalized according to students’ behaviors in any one of the spaces.

As shown in Fig. 1, the embodiment of the disclosure provides a method for constructing an educational learning environment, including the following steps.

(1) A plurality of spaces in the educational learning environment and learning environment construction parameters of each of the spaces are respectively predefined, where the plurality of spaces comprise at least two of a physical space, a network space, a resource space, and a social space.

The learning environment construction parameters are at least one of drive parameters of hardware or software of the physical space, the network space or the social space, or resource acquisition parameters of the resource space.

Through the analysis of educational learning environment, different situations and learning spaces involved in the education learning environment are determined and divided into the physical space, the network space, the resource space, the social space, etc. The specific medium and presentation form of each of the spaces are defined, and typical learning scenarios in each of the spaces and the learning situations under the learning scenarios are determined.

As shown in Fig. 2, the step (1) includes the following steps.

(1.1) An educational learning environment is defined and spaces in the educational learning environment are clarified; where the spaces include: the physical space, the network space, the resource space and the social space; the physical space is an actual environment, the network space (or cloud space, cyberspace) is a virtual space formed by Internet; and the resource space and the social space are virtual spaces formed by knowledge and social relations.

(1.2) The physical space in the educational learning environment is defined, and the meaning of the physical space, contents involved in the physical space and the contents and conditions of the educational learning environment in the physical space are clarified. In this embodiment, it is preferable to define the physical space for the environment of education and learning, and describe the physical world where students are located. The physical space is a physical existence of the actual environment. Due to differences in learning contents, physical locations of the physical space during learning are different, that is, the physical locations can be a classroom, places for learning or activities, such as classrooms, libraries, reading rooms, sports fields, dormitories, lecture halls, etc.

(1.3) The network space in the educational learning environment is defined, the meaning of the network space including contents involved in the network space and the contents of the learning environment supported in the network space are clarified; the network space is the Internet connected by wired or wireless means, and connected to Internet services through corresponding devices; essentially, the network space is a network place for studying through service partitions and organization of different levels of the Internet. The network space is generally used to express the situation of learning with network, processing capabilities, such as network computing capabilities, network storage capabilities, and network service capabilities. Besides, the network space shows a situation of the network for the learning, such as bandwidths and speeds of the network, computing and processing capabilities of the network, storage capacities and read and write speeds of the network, service contents and service qualities of the network, etc.

(1.4) The resource space in the educational learning environment is defined; the meaning of the resource space is clarified; and the characteristics of the resource space and its supporting elements for the construction of the learning environment are determined. The resource space, which is defined as the space of the contents that students learn and calibrated according to different learning contents (i.e., the disciplines or knowledge meanings), is reflected in a unique resource organization structure space formed by contents of each discipline, such as the knowledge map space of the discipline; and the resource space is generally embodied as the knowledge space formed by the semantics of the resource. Essentially, the resource space is to show semantics spaces of different knowledge contents, and its expression form is a knowledge graph. At present, since there is no unified knowledge graph that can express everything, it is necessary to construct the knowledge graph according to different disciplines and research contents. In the embodiment, in an application scenario, the resource space is expressed as a knowledge graph of specific contents. In particular, this embodiment does not carry out any innovation and research on the generation of the knowledge graph, and thus this embodiment only uses the knowledge graph of well-known contents.

(1.5) The social space in the educational learning environment is defined; the meaning of the social space is clarified, and the characteristics of the social space and its construction elements for the learning environment are determined. The social space is a virtual space in which the students communicate with people when learning, where the communication includes social activities among students, social activities among students and teachers, and other related activities. It may be peer- to-peer communication and group communication formed by online learning forums and social networking tools, etc. The social space is a space for socializing through tools such as forums and instant messaging, in which the space records social contents, such as post text, chat text, voice, or pictures and videos shared in moments of Wechat, and the social contents are mainly the personalized expression of the learners to the learning contents or other learners.

(1.6) Other dimensional spaces in the educational learning environment are defined, and the meaning of the space, the basis for setting up the space, and its elements for constructing the learning environment are clarified. In a preferred embodiment, the learning space includes at least the physical space, the network space, the resource space and the social space described as above, but is not limited to the above-mentioned spaces, which may also include other spaces which are not defined to support the learning process.

(1.7) The presentation form and presentation medium of different spaces in a preset learning scenario are defined. In particular, in the embodiment, the physical space is a place where specific information equipment and conditions are equipped, such as smart classrooms which are equipped with traditional blackboards or whiteboards and various output devices such as acoustic sensors, video sensors, biosensors, lights, projections, etc. The network space expresses cloud storage scenarios and cloud service scenarios, where the cloud storage scenarios include learning contents storage space, learning process storage space, etc.; and the cloud service scenarios include online course autonomous learning scenarios, online classroom learning scenarios, online tutoring teaching scenarios. The learning scenarios of the resource space include online learning scenarios, learning contents construction scenarios, and learning contents management scenarios, where the online learning scenarios are expressed as a learner's online learning process for a certain subject and a certain course; the learning construction scenarios are expressed as a construction scenario of knowledge of specific subjects; and the learning contents management scenarios are configured to manage the user's operation and processing of resources during the learning process. The social space can be divided into different learning scenarios, mainly including: course forum discussion scenarios, chatting scenarios for individual learners, teachers or classes on social networking tools, and in the physical space.

(2) The service contents of the spaces are predefined, and a semantic network model of each of the spaces is constructed according to the service contents.

As shown in Fig. 3, the step (2) includes the following steps.

(2.1) The service contents of the spaces are predefined according to the learning scenarios; a semantic calibration is performed on the service contents of each of the spaces; semantic subject units of the service contents are established; and semantic units of the service contents are determined.

(2.2) Semantic relationships of the service contents of the spaces are analyzed, sequence relationships, hierarchical relationships and inclusion relationships between the semantic subject units are determined, and a semantic relationship table is constructed.

(2.3) A semantic network organization structure is constructed according to the semantic relationship table for each of the semantic units of the service contents, so as to express and store semantics.

The method for constructing the semantic network model of the physical space is as follows.

The physical space in the educational learning environment is defined, the service contents provided by the physical space when providing education and learning services are clarified; the equipment and hardware elements involved in providing the service are clarified, and the content semantics of the physical space is analyzed, thereby constructing the semantic network model. Through the technologies such as intelligent sensing, electronic tags, image/voice collection, video monitoring, etc., classroom teaching scenarios perception achieve intelligent sensing of classroom teaching environment parameters and automatic identification of collected target information to provide classroom recording video and teaching courseware video of different perspectives, and customized synthetic video, and record contextual information such as teaching contents, learning styles, teaching methods, etc. in real time. The multi-scenario online learning process perception will adopt the "activity stream" to track the learner's effective behavior information collection in all activities in real time, to achieve the unified processing, quantification and recording of structured, semi-structured and unstructured behavior data, forming a panoramic learning behavior data pool which is reusable and computable.

Classroom teaching behavior perception can perceive typical classroom teaching behavior information such as classroom interaction, question answering, discussion, etc., between teachers and students and automatically extracts interactive behavior data such as interactive frequencies, interactive subjects and interactive contents in different stages. The generation format of "activity stream", such as [subject, action, object, result, scene, timestamp, authority], is transformed into well-structured data that are applicable for modeling through semantics.

The method for constructing the semantic network model of the network space is as follows.

The network space in the educational learning environment is defined; the services contents provided by the network space when providing education and learning services is clarified; the hardware elements of the network space services and the service elements of the network space are clarified; the content semantics based on services is analyzed to construct a semantic network model of the network space. At present, personalized services in the network space mostly involve personalized learning diagnosis, personalized learning path planning, personalized resource recommendation, and visualization of learning status, but rarely involve personalized learning intervention. Personalized learning in the network space focuses on improving the online experience of netizen groups, mostly based on social computing, complex networks and other technologies to construct netizen group models, and provide customized learning path planning and other personalized service contents for different groups as needed by analyzing group needs and interaction between groups, and classifying groups according to different needs.

The method for constructing the semantic network model of the resource space is as follows.

The resource space in the educational learning environment is defined, the service contents provided by the resource space when providing education and learning services is clarified; the hardware elements and service elements of the resource space service are clarified; a graph relationship of the resource semantic in the resource space is analyzed to construct a semantic network model. Based on the contents of course knowledge, including nodes (domain concepts, knowledge fragments) and inter-node relationships (relationships among domain concepts, relationships between courses, relationships between knowledge fragments and domain concepts, etc.), a semantic network of knowledge is constructed; the text contents are judged sentence by sentence using the existing single-sentence semantic type classification method to discover the indicator sentences of various types of domain concepts; secondly, a lexical chain method is adopted to analyze the lexical chain of the adjacent sentences of the indicator to find a lexical dependency relationship between sentences; finally, based on the analyzed results of the lexical chain, a starting point and an ending point of domain concept text are labeled using sequence labeling models such as Conditional Random Fields (CRF), Hidden Model HMM (HMM), Maximum Entropy Markov Model (MEMM), and a domain concept boundary division method is proposed by comparatively analyzing the labeled results and the analyzed results of lexical chain.

On this basis, according to the text characteristics such as word frequency, word order, context, etc., the term competitive learning method is used to identify the core terms, and the multi-class classifier is used to identify the semantics type of the domain concept.

In the tracking of each learner's online learning process, according to the characteristics of the user's click behavior, click resource type (text, image, exercise/test question of domain concept), forum discussion knowledge sharing method, etc., combining with the learning resource metadata (including keywords), the degree of mastery of knowledge concepts learned by users, etc., and using conditional constrained sequential pattern mining methods, learners’ click behavior patterns are discovered, learner’s cognitive strategies are analyzed, and the typical mass learners model is discovered; and by borrowing Felder-Silverman's learning style classification method, different learner styles are summarized.

The method of constructing the semantic network model of the social space is as follows: the social space in the educational learning environment is defined, and the contents of the education and learning services provided by the social space are clarified; the hardware elements and service elements of the resource space service are clarified, and the individuals of the social space and their relationship semantics are analyzed to construct the semantic network model. (3) Learning subject-related events and learning environment-related events are collected, and a data fusion model of a cyber-physical space semantic level consistency for learning subjects is constructed.

As shown in Fig. 4, the step (3) includes the following steps. (3.1) Embodied objects of a learning subject in different spaces are created, and a consistent expression of the same learning subject in different spaces is determined.

The embodied objects of unified learning subjects in different spaces are constructed to form learner subjects; individual static characteristics (background information, early knowledge ability, learning style, etc.) and dynamic features (current knowledge ability, learning motivation, cognitive level, emotional attitude, interest preference, etc.) in multiple learning scenarios (online resource browsing, collaborative mutual evaluation, question and answer, offline classroom interaction, outdoor learning, etc.) are analyzed; individual key learning features are extracted. Based on theoretical methods such as contents analysis, cognitive classification theory, sentiment analysis, deep learning, etc., key elements representing individual characteristics are extracted from the multi-source learning activity data, and key factors affecting individual learning process are analyzed, thereby exploring the characteristics and personalities of learners in different scenarios in terms of knowledge ability, cognitive level, emotional attitude, etc. Taking the timing, scene and deep semantic characteristics of the learning state features into account, the learning state of individuals under a specific time and space is identified. Finally, a dynamic unified model of learners based on specific instructional designs and ability evaluation is constructed.

(3.2) The learning subject-related events and the learning environment-related events are collected, and the learning subject-related events and learning environment-related events are converted into data according to the semantic network model to form a learning behavior data pool.

The learning subject-related events and the learning environment-related events are learner behaviors/actions collected by various sensors and network monitoring devices in the environment or monitoring parameters related to the learning environment. For example, through the technologies such as intelligent sensing, electronic tags, image/voice collection, video monitoring, etc., the classroom teaching scenarios perception achieve intelligent sensing of classroom teaching environment parameters and automatic identification of collected target information to provide classroom recording video and teaching courseware video of different perspectives, and customized synthetic video, and records contextual information such as teaching contents, learning methods, teaching methods, etc. in real time. The muliti-scenario online learning process perception will adopt the "activity stream” to track the learner's effective behavior information collection in all activities in real time, to achieve the unified processing, quantification and recording of structured, semi-structured and unstructured behavior data, forming a panoramic learning behavior data pool which is reusable and computable. The classroom teaching behavior perception can perceive typical classroom teaching behavior information such as classroom interaction, question and answer, discussion, etc. between students and teachers, and automatically extracts interactive behavior data such as interactive frequencies, interactive subjects and interactive contents in different stages. The generation formats "activity stream”, such as [subject, action, object, result, scene, timestamp, authority], is transformed into well-structured data that are applicable for modeling through semantics.

(3.3) The data fusion model of the cyber-physical space semantic level consistency is constructed based on the learning behavior data pool. In the embodiment, the semantics of objects and learning behaviors of the same learning subject in different spaces is clarified; the semantics of different spaces describing the same event, the same behavior and activities is analyzed; the data entities of each semantic object are clarified, and cyber-physical space data fusion standards are constructed. Different classroom teaching scenarios, different network platforms and social organization spaces carry out the automatic extraction, intelligent identification and automatic record for classroom teaching situations, teaching subjects, teaching status and learning behavior, so that structured, semi-structured and unstructured behavior data are uniformly processed, quantified and recorded, thereby forming a panoramic learning behavior data pool which is reusable and computable.

A long short-term memory model and three-stream convolutional neural networks (three-stream CNNs) with three channels of space, local time domain and global time domain are used to extract and classify the learning behavior data, thereby constructing a data fusion model of the cyber-physical space semantic level consistency.

The three-stream CNNs framework including three channels of space, local time domain and global time domain is used to extract the spatiotemporal features of learners’ behaviors. The three-stream CNNs framework includes 4 convolutional layers (Conv 1-4), where 2 convolutional layers (Norm 1 and Norm 2) are normalized, and connected to 2 pooling layers (Pooling 1 and Pooling 2); after the three channels (the spatial channel, the local time domain channel and the global time domain channel) undergo convolution and pooling operations, deep features are obtained; where the spatial channel CNNs perform deep learning for learner behavior images; the local time domain channels CNNs perform deep learning for optical flow features;

and the global time domain channel CNNs perform deep learning for learner's behavior and differential image products.

With the advantages in dealing with time series problems, the long short-term memory model (LSTM) is introduced into the training of the classification module to recognize the learner's behavior. The features extracted by 3DCNN are input into the LSTM model for learning. Time-domain information is introduced into sequence learning, thereby bringing more accurate results to the classification. Then, a spatial pyramid pooling layer (SPP) is added between the fully connected layer (FC) and the LSTM model. After the SPP layer calculates the feature maps of different sizes, a fixed-length feature vector can be obtained. Then, through the fully connected layer, 3DCNN+LSTM is learned to classify a single type of features. Finally, the classification results of the independent features of the three channels are voted to obtain behavior action categories, so as to summarize and store the corresponding behavior data in the form of tags. Through the two-way long short-term memory neural network to learn the data of learners in different learning spaces, the single form features of the learning process data and the shared features that show the correlation between different data forms can be effectively learned, so that relevance between relevant learning contents and learning behavior data at different times and in different spaces can be captured, thereby aggregating the data with the corresponding “activity stream” specifications. The relationship between different data is automatically sorted out using the learned multi-level features. On this basis, Deep Boltzmann Machine (DBM) is used to fuse feature-level data from different learning spaces.

(4) An integrated embodied model is constructed based on learning scenarios, where the integrated embodied model is used for describing relationships between the learning subjects and the learning environment construction parameters in the learning scenarios, and is capable of being dynamically adjusted according to learning subject-related events or learning environment-related events collected in any one of the spaces.

Specifically, the step (4) includes the following steps.

(4.1) The learning subject-related events and the learning environment-related events are converted into standard format data according to the semantic network model and the data fusion model.

(4.2) The relationships between the learning environment-related events and the learning subject-related events are converted into a graph network model; the spatial environments are expressed using vertexes of the graph network model, and embodied relationships among different spaces are expressed using edges of the graph network model; the integrated embodied model is constructed based on the learning scenarios using the graph network model, where the integrated embodied model is used for describing relationships between the learning subjects and the learning environment construction parameters in the learning scenarios, and is capable of being dynamically adjusted according to the learning subject-related events or the learning environment-related events collected in any one of the spaces.

Specifically, an integrated embodied model is constructed based on the integration of scene-based space and environment structure, resource and content semantics, learning activities and behaviors, data of the embodied learning environment is fused using the graph network model, that is, a physical embodied model of the intelligent partner system.

Using the graph neural network, the embodied learning environment and embodied relationship are expressed as follows: G=(N.E) ; where G (term: graph) is a set of embodied environments and embodied relationship of the cyber-physical space fusion; N (term: vertex) is expressed as a spatial set of embodied learning environments; and E (term: edge) is a set of embodied relationships. ne[n] represents an adjacent space of a space n (represented by vertices), and co[n] relates the relationship of the space n (represented by edges). The space n and the corresponding attribute of the relationship (n_1,n_2) are respectively expressed as | neRMI_N) and {(n_1, n_2)ERA(L E), | represents the tensor formed by stacking all the attributes in the graph.

The embodied learning environments of multiple spaces and the relationships therebetween are generalized into a graph neural network model, where the vertex is used to express the spatial environment, and the edges are used to express the embodied relations of different spaces. Based on the computational model of the graph neural network, the calculation rule for perception and fusion of embodied data of multiple spaces is constructed to obtain a solution of personalized personal environments.

Based on technologies such as voice interaction, action interaction, holographic visual interaction, wearable sensor interaction, etc., data perception of all elements of the environment and the entire process of behavior is constructed to realize cyber-

physical space embodied interaction of learners through the multi-model interaction technology. The learner's cognitive results are perceived through the embodied learning environment, and a feedback mechanism of the cognitive results to the embodied learning environment and intelligent services is constructed. Cognitive practice results are fed back to the learning environment to realize customization of the learning environment; and cognitive results are fed back to the intelligent service to realize the resource customization of the intelligent service and achieve customized learning.

The cyber-physical space embodied model further includes data standard specification of fusion of spatial hierarchy, resource graph and learning interaction behavior.

Aiming at unstructured, distributed, heterogeneous, and source-scattered features of data, under the guidance of learning science and education and teaching related theories, an educational cyber-physical space data fusion standard system is formed, where the system includes education-related standards, such as subject standards, resource standards, and evaluation standards, management standards, teaching process standards, data processing and data quality standards, and data interoperability standards. Through the unified conversion gateway for education big data standardization, the standardized processing of heterogeneous education data is realized, including the information extraction, data storage and retrieval of structured, unstructured and semi-structured data, as well as the steps of data cleaning and data verification, so that usable and trusted data sources for data modeling, analysis and application are provided. An information association model of cross-scenario and cross-temporal education big data is established by acquiring the multi-source data entities and their multi-level association relationships, so as to realize the information extraction and aggregation of muiti-source heterogeneous educational data, so that data after processed can meet the needs of applications such as data analysis and modeling. Data exchange and sharing services based on application requirements are realized, and an on-demand data exchange for multiple data sources is supported through unified standard data interfaces and standard data formats, including data aggregation, data distribution, data update, data conversion, etc., thereby supporting identity authentication, user authorization, transmission encryption, data integrity, data credibility, and data validity.

As for teaching subject standards, the basic information of the teaching subject is described, and the continuous recording and data association of the active subject across platforms and systems are realized. The teaching subject is the subject of teaching activities, including students, parents, teachers, teaching researchers and teaching managers. Teaching resource standards include a unified description, encapsulation and reorganization of teaching resources of different forms, different granularities, and different formats, such as courses, videos, and exercises. The teaching resource standard not only includes the metadata description of the resource attributes, but also includes the semantics attributes of the resources to support automatic recognition and processing of machines, so that the personalized intelligent push of resources is realized. The teaching process standard is to describe any interaction or related experience between the teaching subject and teaching contents (such as courses, resources, etc.}, teaching environment (such as traditional classrooms, outdoor learning environment), and other participants in the teaching activities during the teaching process. semantic The teaching process standards are oriented to traditional teaching environments (such as schools, classrooms) and non-traditional teaching environments (such as online learning environments, outdoor learning environments). It mainly includes how and when teaching activities occur, contextual information, and the results of the teaching process. Data processing standards mainly regulate data collection, preprocessing, analysis, visualization, and access. Data quality standards mainly propose specific management requirements and corresponding index requirements for data quality to ensure the quality of data in various links such as generation, storage, exchange and use, and lay a good foundation for the application of educational big data. Education data interoperability standards are mainly aimed at the heterogeneity of education data to achieve the connection between massive data sets and the interoperability requirements of coupling, fusion, migration and information extraction of education data.

(5) The learning environment construction parameters for the learning subjects are set according to the integrated embodied model. (5.1) An automatic construction technology of personalized cyber-physical space learning environment for learners is established; based on technologies such as voice interaction, action interaction, holographic visual interaction, wearable sensor interaction, etc., data perception of all elements of the environment and the entire process of behavior is constructed to realize cyber-physical space embodied interaction of learners through the multi-mode! interaction technology. The learner's cognitive results are perceived through the embodied learning environment, and a feedback mechanism of the cognitive results to the embodied learning environment and intelligent services is constructed. Cognitive results are fed back to the learning environment to realize customization of the learning environment.

The situation of the learning space and the behavior of the learning subject are perceived; data such as environmental data, learning process data, learning behavior data is obtained; cyber-physical space embodied entities are constructed and are expressed as vertexes of the graph network; the edge structures of the graph network are formed by the relationship between the spatial embodied entities to describe the relationship between the vertexes, and finally an educational cyber- physical space learning environment graph network structure is formed. Using the topology calculation method of the graph, an educational cyber-physical space data fusion algorithm is developed to achieve subject-based data fusion.

Learner feature extraction algorithm, learner state recognition algorithm, learner portrait technology and learning analysis technology are applied to analyze the state of the learning process of learning subjects in multiple spaces and realize the visualization of the learning process of learning subjects in multiple spaces. In the learner's learning process, data mining and in-depth analysis are performed to draw the learner's learning curve, so as to diagnose the students’ knowledge structure in detail, find blind spots in learning, and design personalized learning plan that is more targeted at students’ weak knowledge to provide learners with personalized learning diagnosis.

(5.2) An educational learning environment creation device that meets the learner's embodied requirements is developed and integrated; an intelligent learning partner system for embodied recognition is perceived and fused through the cyber- physical space personal cognitive data perception and fusion module to realize perception and fusion for the learning environment and learning process data; the learner learning analysis module based on multi-source data is adopted to realize the recognition of the characteristics and status of the learning subject, so as to accurately characterize the learner and its learning behavior feedback, the intelligent tutoring module is applied to realize autonomous learning.

According to the individual needs of learners, data mining and semantic search are used to continuously improve the course knowledge map. Combining the learning process, forgetting rules and success rate, the node relationships in the course knowledge map are dynamically updated, and genetic algorithms (GA) are used to optimize the path, so that a learning path adapted to the learner's personality is outputted. Through the optimized learning path and high-quality adaptive learning resources, the generalized resource recommendation method is applied to achieve personalized intelligent tutoring for learners.

FIG. 5 schematically shows the application of a device for constructing the educational learning environment according to an embodiment of the present disclosure, where the device is connected to a video module, a network module and an Internet of things (loT) module, etc. in the learning environment through interfaces. The video module, the network module and the loT module, etc. are configured to collect learning subject-related events and learning environment- related events in different spaces, and are controlled to construct parameters to configure the learning environment according to the educational learning environment of the device.

As shown in Fig. 6, the device includes the following modules.

(1) A space predefining module for predefining a plurality of spaces in the educational learning environment and learning environment construction parameters of each of the spaces, wherein the spaces include at least two of a physical space, a network space, a resource space, and a social space. The space predefining module is configured to define the educational learning environment, and clarity the spaces in the educational learning environment. The spaces at least include: the physical space, the network space, the resource space and the social space. The space predefining module is further configured to define the spaces in the educational learning environment, and clarify meaning of the spaces, a basis for providing the spaces, and contents of the educational learning environment supported in the spaces. The space predefining module is further configured to define presentation forms and presentation mediums of different spaces in a preset learning scenario.

(2) A semantic network model constructing module for predefining service contents of the spaces according to a learning scenario, and constructing a semantic network model of each of the spaces according to the service contents.

Specifically, the semantic network model constructing module includes the following modules.

(2.1) A spatial content semantic calibrating module for predefining the service contents of the spaces according to the learning scenario, performing a semantic calibration on the service contents of each of the spaces, establishing semantic subject units of the service contents, and determining semantic units of the service contents.

(2.2) A spatial content semantic analyzing module for analyzing semantic relationships of the service contents of the spaces; clarifying sequence relationships, hierarchical relationships and inclusion relationships between the semantic subject units to construct a semantic relationship table.

(2.3) A spatial content semantic network module for constructing a semantic network organization structure according to the semantic relationship table for each of the semantic subject units of the service contents.

The semantic network constructing module performs a semantic calibration and hierarchical organization of contents of different spaces, constructs a semantic network model based on the semantic relationship between spatial contents, establishes a content-based organizational structure, and establishes a semantically consistent expression of cyber-physical space services and contents; and converts the connection between cyber-physical space learning environment, learning behavior and cognitive results (learning results) into a graph network model; where a vertex of the graph network model expresses the cognitive result (knowledge) of a space and expresses learning behavior; a certain subnet in the graph network expresses a certain learning space (that is, an embodied learning environment), and the aggregation operation of the nodes between the subnets is the personalized solution of the learning space. A spanning tree after the multi-subnet personalized solution is expressed as a personalized cyber-physical space model. The frame and feedback mechanism of the embodied interaction are the constraints of aggregate computing.

(3) A data fusion model constructing module for collecting learning subject- related events and learning environment-related events of different spaces, and constructing a data fusion model of a cyber-physical space semantic level consistency for learning subjects.

Specifically, the data fusion model constructing module includes the following modules.

(3.1) A learning subject consistency verification module for creating embodied objects of a learning subject in different spaces, and determining a consistent expression of the same learning subject in different spaces.

(3.2) A learning subject cyber-physical space data collection module for collecting the learning subject-related events and the learning environment-related events, and converting the learning subject-related events and learning environment-

related events into data according to the semantic network model to form a learning behavior data pool. (3.3) A data fusion module for constructing the data fusion model of the cyber- physical space semantic level consistency based on the learning behavior data pool.

The data fusion module of the learner subject constructs an educational cyber- physical space data fusion method based on the learner subject and the semantic contents of the same level of learning, learning services and personality characteristics, to realize the cyber-physical space instance data representation of the learner subject based on the same semantics; standardized and unified conversion gateway is constructed to realize the standardized processing of heterogeneous educational data, including information extraction, data storage and retrieval of structured, unstructured and semi-structured data, as well as steps of data cleaning and data verification, etc., thereby providing usable and trusted data sources for modeling, analysis and applications of data.

The pre-processing process includes steps such as data cleaning, data verification, and normalization processing to provide usable and trusted data sources for data modeling, analysis, and applications of data.

Erroneous data and redundant data in the data is cleared; the consistency and completeness of the original data are verified; and the data are processed using unified standards, such as text information format conversion, consistency of unit of measurement, etc., thereby making the data set to be processed more complete, and transforming heterogeneous and complex data into analyzable and applicable information. (4) An integrated embodied model constructing module for constructing an integrated embodied model based on learning scenarios, where the integrated embodied model is used for describing relationships between the learning subjects and the learning environment construction parameters in the learning scenarios, and is capable of dynamically adjusting according to learning subject-related events or learning environment-related events newly collected in any one of the spaces. (5) A learning environment constructing module for setting the learning environment construction parameters for the learning subjects according to the integrated embodied model.

The learning environment construction module constructs an educational cyber- physical space fusion learning environment with learners as the main body, realizes the fusion expression, display and data services of different spaces, and proposes a personalized learning environment.

Cyber-physical space personalized environment construction: the construction parameters of different learning spaces are calculated based on the learner's subject, and the parameter-driven environment construction technology is applied to realize the environment setting and customization of different learning spaces.

Learners can learn in different spaces, where their learning experience should be expressed as a unified whole, and its overall effect needs to be embodied in each space through the embodied model. In a certain space, the resources or abilities of other spaces can be obtained through the embodied object, and activities in a space can be fed back to different spaces through the embodied object. For example, the learning behavior of the cloud space can be fed back to the learning activities of the physical space, so that the learning is carried out through embodied objects of multiple spaces and is no longer the learning behavior or process of a single space.

Cyber-physical space learning environment fusion: a learning environment fusion device is developed, in which a non-physical space is fused in a certain physical space, and holographic technology and multi-modal interaction technology are applied to realize the integrated presentation of multiple spaces to provide a personalized intelligent learning environment. According to the learner's personalized characteristics, the cyber-physical space learning environment that matches the personality is customized, such as the learning environment that conforms to the learner's personal interaction mode (such as voice interaction, somatosensory interaction, etc.), which emphasizes the impact of feedback on the learning process on the user's sense, touch, vision, etc.

The method and device for constructing the educational learning environment of the present disclosure has the following beneficial effects.

(1) Based on spaces involved in education and teaching, the present disclosure analyzes the physical space, network space, resource space, social space, etc. involved in the educational scenarios, and explores a semantic analysis method based on the contents of education and teaching services to realize an educational cyber-physical space unified expression and data fusion method with a consistent learning subject, which can construct an educational cyber-physical space fusion learning environment according to students’ behaviors in any one of the spaces.

(2) The present disclosure introduces a method of multi-level data fusion into the application of educational scenarios, making it possible to construct a consistent cyber-physical space learning environment with learners as the learning subject.

(3) The present disclosure proposes a learning space fusion method for learners, and expression methods such as semantic network model and traditional independent spatial expression methods are fused. (4) The present disclosure constructs a personalized learning environment and designs an educational cyber-physical space fusion system for learner subjects, which supports a personalized learning environment under different physical conditions. it should be understood that the steps are not necessarily executed in the order indicated by the arrows.

Unless specified, the execution order of these steps is not limited herein. it should be understood that described above is only preferred embodiments of the present disclosure, but not intended to limit the scope of the present disclosure.

Any modification, equivalent replacement and improvement made within the spirit and principle of the present disclosure shall fall within the scope of present disclosure defined by the appended claims.

Claims

Conclusions

A method of constructing an educational learning environment, comprising: (A) pre-defining multiple rooms in an educational learning environment and learning environment construction parameters of each of the rooms; wherein the spaces comprise at least two of a physical space, a network space, a resource space, a social space, wherein step (A) comprises: defining the learning environment, and clarifying the spaces in the learning environment, wherein the spaces are least include: the physical space, the network space, the resource space, and the social space; defining the spaces in the learning environment, and clarifying the meaning of the spaces, a basis for providing the spaces and the content of the learning environment supported in the spaces; and defining presentation formats and presentation media of different spaces in a preset learning scenario; (B) predefining service content of the spaces, and constructing semantic network model of each of the spaces according to the service content; (C) collecting subject-related events and learning environment-related events, and constructing a data fusion model of a cyber-physical space semantically consistent level for learning subjects; (D) constructing an integrated embodied model based on learning scenarios, where the integrated embodied model is intended to describe a relationship between the learning subjects and the learning environment construction parameters in the learning scenarios, and can be dynamically adapted according to learning subject related events or learning environment related events gathered in one of the spaces; and (E) setting the learning environment construction parameters for the learning subject according to the integrated embodied model.

Method according to claim 1, characterized in that step B) comprises: (B.1) pre-defining service content of the rooms according to the learning scenarios; performing a semantic calibration on the service content of each of the spaces; determining semantic subject units of the service content; and determining semantic units of the service content; (B.2) analyzing semantic relationships of the service content of the spaces; determining sequence relationships, hierarchical relationships, and inclusion relationships between the subject semantic entities; and constructing a semantic relationship table; and (B.3) constructing a semantic network organization structure according to the semantic relationship table for each of the semantic units of the service content.

A method according to any one of claims 1-2, characterized in that step (C) comprises: (C.1) constructing embodied objects of a learning subject in different spaces, and determining a consistent expression of the same learning subject in different spaces; (C.2) collecting subject-related events and learning environment-related events, and converting the learning subject-related events and learning environment-related events into data according to the semantic network model to form a learning behavior data pool; and (C.3) constructing the cyber-physical space data fusion model semantically consistent level based on the learning behavior data pool.

A method according to claim 3, characterized in that the step {C.3} comprises: Extracting and classifying learning behavior data from the learning behavior data pool by a long-short-term memory model and a three-channel convolutional neural network comprising a spatial domain, a local time domain, and a geographic time domain, for constructing the cyber-physical space data fusion model semantically consistent level.

The method according to any one of claims 1-4, characterized in that the step (D) comprises: (D.1) converting the learning subject related events and the learning environment related events into standard data format according to the semantic network model and the data fusion model; and (D.2) converting relationships between the learning environment related events and the learning subject related events into a graphical network model; expressing spatial environments using vertices of the graphical network model, and expressing embodied relationships between different spaces using edges of the graphical network model; constructing the integrated embodied model based on the learning scenarios using the graphical network model, where the integrated embodied model serves to describe the relationships between the learning subjects and the learning environment construction parameters in the learning scenarios, and can be dynamically adjusted according to the learning subject-related events or the learning environment related events gathered in one of the rooms.

Method according to any one of claims 1-5, characterized in that the learning environment construction parameters are one or more drive parameters of hardware, or software of the physical space, the network space, or the social space or resource acquisition parameters of the resource space.

An apparatus for constructing an educational learning environment, comprising: - a space-defining module for pre-defining multiple spaces in the educational learning environment and learning environment construction parameters of each of the spaces, the spaces being at least two of a physical space, a network space, a resource space, and a social space; the space define module is configured to define the educational learning environment, and clarify the spaces in the educational learning environment, wherein the spaces include at least: the physical space, the network space, the resource space, and the social space; the space definition module is further configured to define the spaces in the educational learning environment, and clarify the meaning of the spaces, provide a foundation for the spaces, and the content of the educational learning environment supported in the spaces ; and the room defining module is further configured to define presentation shapes and presentation media of different rooms in a predetermined learning scenario; - a semantic network construction module, for pre-defining service content of the rooms, and constructing a semantic network model for each of the rooms according to the service content - a data fusion model construction module for collecting learning topic related events and learning environment related events , and constructing a data fusion model of a cyber-physical space semantically consistent level for learning subjects; - an integrated embodied model construction module for constructing an integrated embodied model based on learning scenarios, where the integrated embodied model serves to describe relationships between the learning subjects and the learning environment construction parameters in the learning scenarios, and can be dynamically adapted according to learning topic related events or learning environment related events gathered in one of the rooms; and - a learning environment construction module for setting the learning environment construction parameters for the learning subjects according to the integrated embodied model.

Device according to claim 7, characterized in that the semantic network construction module comprises: - a spatial content semantic calibration module for pre-defining service content of the spaces according to the learning scenarios; performing a semantic calibration on the service content of each of the spaces; determining semantic subject units of the service content; and determining semantic units of the service content; - a spatial content semantic analysis module for analyzing semantic relationships of the service content of the spaces; clarifying sequence relationships, hierarchical relationships, and inclusion relationships between the subject semantic entities, and constructing a semantic relationship table; and - a spatial content semantic network construction module for constructing a semantic organizational structure according to the semantic relationship table for the semantic units of the service content of the spaces.

Apparatus according to claim 7 or 8, characterized in that the data fusion model construction module comprises: - a learning subject consistency module for creating embodied objects of a learning subject in different spaces, and determining a consistent expression of the same learning subject in different spaces; - a learning subject cyber-physical space data collection module for collecting subject related events and learning environment related events, and converting the learning subject related events and learning environment related events into data according to the semantic network model to form a learning behavior data pool; and - a data fusion module for constructing the data fusion model of the cyber-physical space semantically consistent level based on the learning behavior data pool.

Apparatus according to any one of claims 7-9, characterized in that the learning environment construction parameters are one or more float parameters of hardware, or software of the physical space, the network space, or the social space or resource acquisition parameters of the resource space.