Research On Semantic Model Of Educational Resources Based On Relational Formal Concept

In recent years, with the further promotion of the national information strategy and the continuous development of digital education, the construction of educational resources at all levels in China has made great achievements. Because of the huge stock,rich content and rapid growth of different kinds of education resources on Web,especially the deep application of Internet Plus thinking in social life, many auxiliary learning modes and means are also evolving and progressing quietly with the technological change. The resource acquisition of terminal learners, as well as their contribution and influence, will be more and more direct. Thus, it will further promote the construction of educational resource sharing scale and strengthen the quality of development. All of these have put forward advanced requirements to education resources on the effective organization and acquisition. However, at the same time, the huge amount of educational resources is in an open, dynamic and changeable environment of Internet. And also, the distribution, types and actual description standard are not unified. For the specific purpose of learning, learners often have difficulties in achieving efficient resource content selection and acquisition. How to use the text information technology to realize automatic identification on resource semantic level,correlation integration in different tissues and different forms of education resource data,and to achieve a more effective organization of educational resources and sharing have very important theoretical and practical value.According to the background above, this paper proposes the educational resources semantic interconnection model—Resource Related Semantic Link Network (RALN),which is based on the relational formal concept. It mainly focuses on the following three key research questions:(1)To introduce more background knowledge and to improve the generality and adaptability of the model for Web educational resources identification and organization,this paper needs to provide a generic semantic frame for the expression and acquisition of background knowledge;(2) To reduce the limitation of text topic model, which is simply composed by statistical features of discrete key words, and to make it more similar to the way of human parallel reading, the paper needs to obtain the context and local semantic features of the resource fragment automatically in order to make better use of the value of background knowledge;(3) In view of the time and space dynamics of individual resources, that is, to different learners and different learning stages, the correlation between resources may be different or changeable, the resource model should not only expresse the content static association, but also have the dynamically expand ability at the same time, which provides dynamic model support for the semantic identification of educational resources and association organizationsIn view of the three key questions above, the main research contents of this paper are:(1) Based on the extension of the traditional Formal context, the Formal Concept Model (RFCM) is constructed. Based on classic three tuple relations, a conceptual model of RFCM is formed,which provides a uniform formal framework to describe the semantic entity concept, including the concept of the subject and object concept, entity relation (predicate concept) and a discrete entity entry. That is,the relationship is expressed as subject concept, object concept and predicate concept. And the subject and object concept is the mathematical set form of relational semantics, including the concept in two aspects-the connotation and extension of the philosophy definition.Based on relational formal concept model, with the open collaborative knowledge base(such as Baidu Encyclopedia, Wikipedia), the paper automatically accesses the large data entry and data labels as formal background sources of the relational formal concept to build background knowledge in the form of relational concept. Under the semantic background provided by the relational formal concept and with the comprehensive consideration of the context and syntactic information, the paper uses graph theory to calculate the relational formal concept connectivity of entity entry, and to realize the extraction of entity relation and the identification of entity concept cooperatively. As a kind of text semantic acquisition and expression model of cooperation and self organization, Relational Formal Concept Model can provide a more flexible form of semantic background and unified semantic frame for subsequent text subject extraction and expression, as well as Web resources related organizations.(2) Based on relational formal concept provided by RFCM, the complete technical route for the definition and acquisition of text topic is designed, and the Relational Formal Concept Topic Model (RFCTM) is constructed. With the background knowledge in form of relational formal concept, RFCTM calculates the correlative relationship between the relational formal concepts and entity entries. And considering the entry context syntax information, RFCTM calculates semantic connectivity of the relational formal concepts in the text to implements acquisition path of the resource topic - from original text entry, entry topic, text discrete topic to text connection topic. Compared to the topic expression with discrete keywords, RFCTM has more flexible semantic granularity and more completed theme framework. The modern system science believes that the nature of the new things is on the basis of existing elements, but it can not be explained completely by the existing elements. Certain inner relation between elements is the way things are. This view also fits for the analysis of the text topic, that is, entry is the foundation of the semantic. However, discrete entry and its concept also cannot completely characterize the text semantic topic. RFCTM simulates the human parallel reading cognitive style based on local features and on the context. RFCTM model can provide a formal topic vector for the Web educational resource based on content, and provides the basic semantic elements for the association link of resources.(3) On the basis of the text topic model, the identification and organization semantic model of the educational resources are constructed, which includes the knowledge association link network (KALN) and the resource association link network(RALN). KALN is the knowledge network constructed by correlation between relational formal concepts. It provides basis semantic background knowledge for the resource identification and association. RALN expresses the semantic relations between resources text fragments. The resource fragments are regarded as independent topic models in the paper. It is also an associated logical node of RALN. And at the same time,it mergers and contracts the same topic patterns in order to reduce the increasing speed of RALN nodes in the process of resource identification and association, to control the size of RALN semantic correlation network and to improve the efficiency of resource identifier and association. Because KALN and RALN are constructed by general open collaborative knowledge base, identification and association for specific educational resource are often either too broad or lack of the support of domain knowledge. And it's difficult to reflect the domain resources association or domain requirements. Thus, it requires a combination of resources for further extension.(4) With the aid of the four hypotheses of the neural network structure theory,the KALN link strength calculation is extended. Simulating the human memory activation diffusion and forgetting inhibition mechanism, this paper also makes a dynamic adaptive adjustment to the association strength of correlation between relational formal concepts with the resource content, which makes it reflect not only the semantic association between relational formal concepts, but also the relational formal concepts co-occurrence semantic association in Web resources. And it also makes the Knowledge Association Link Network more consistent with background field. Based on the structural information of the document resources, RALN is also extended in this paper.The association between the topic modes reflects fine-grained semantic association of the relational formal concepts, as well as the resource topic modes correlation in the same document resources. Finally, the incremental updating algorithm of KALN and RALN is optimized, and the extended learning efficiency of the model is also improved in this paper.The innovative work of this paper is mainly reflected as the following:(1) This paper extends the formal concept analysis, defines the relational formal concept, unifies the semantic expression of the terms, relations and concepts, and provides a new idea for the expression of the text topic.(2) Based on the relational formal concept, this paper calculates the entry connectivity in the text, which provides more contextual information for the semantic meaning of the text.(3) According to the rule of human activation diffusion and forgetting inhibition,this paper constructs and extends the association link network to provide a dynamic model for personalized learning and resource recommendation.