Research On Ontology-based Adaptive Learning Method And Application

E-Learning has become an important way in education and training, it provides a well-designed and learner-centered interactive learning environment. In this mode, learners can learn at any time and in any place. The design and development of learning resources involved a variety of digital technologies, this adapts learning resources to a flexible and open learning environment. Many scholars, institutions and enterprises dedicated to research and work in this area. They have made great achievements both in theory and technology, but put forward new demands: In E-Learning system, learners should be able to offer a personalized learning experience, system should provide personalization learning resources to adapt to learner's personalization features, namely, adaptive learning system is needed. In this context, learner model is necessary to represent the learner's preference and need, the system can provide personalized learning content for learners by referring to the information in learner model, so the acquisition of learner characteristics and learner modeling are the two important issues in this paper. While there are new requirements in knowledge management: knowledge should be formalized, shared and reusable; knowledge acquisition should be intelligent, the obtained knowledge should be accurate and comprehensive to meet the needs of learners, knowledge acquisition in the syntax level is not enough; knowledge management should be more flexible in the increasing, updating, processing, transmission and so on, they should be easy to operate; knowledge should be shared and reused. So an adaptive learning system must meet the following three important aspects: first, the knowledge should be formal, explicit, consistent, and can be shared; second, the system need describe accurately a learner's characteristics and needs; third, the system must have the intelligent reasoning engine. The Semantic Web and ontology technologies can meet the above requests in aspect of knowledge representation and learners modeling and reasoning. Then, the adaptive learning system opened a new chapter. In this paper, we research the following four issues: the needs and frame of adaptive learning system; knowledge management; learners modeling; reasoning algorithm.In this paper, we first introduced the relevant basic knowledge of adaptive learning system, including theoretical and technical basis. In theory, we introduced the development process of the learning theory, particularly constructivism learning theory, knowledge view of constructivism, learning view of constructivism, teaching view of constructivism, and the guidance significance of constructivism for the adaptive hypermedia system. In technology, we introduced the relevant basic knowledge of semantic web and ontology, including the background of semantic web, concept definition, development and the state of application. Finally, we introduced the definition, classification methods, construction methods, developing tool, language of the ontology and the key role of the ontology in this paper.In the needs and frame of adaptive learning system, we first analyzed the requirements and the difficulties faced in developing the adaptive learning system, and proposed our frame structure of system, the design idea is based on providing the personalized learning content for every learner. The system consists of seven main modules, including the diagnostic module of the learning state, the presentation module of learning content, the learning evaluation module, the learner model module, the collaboration and communication module, the intelligent question and answering module, the learning tool module. The diagnostic module of the learning state aims to estimate the student's level of learning ability, cognitive ability and mastery of domain knowledge. It is an important basis of the adaptive recommendation of learning content. The diagnostic methods can be the test paper, and the questions in test paper are determined by the measuring theory, it can be diagnosed by the information in learner model. Test results can be as the basis both in learning content recommendation and also the evolution of learner model; in the learning content presentation module, the presentation of learning content is different according to the characteristics of different learners. The main basis is preference and cognitive skill of learner. The learning content is shown to every learner in hypermedia way, the difference is that the level and content of hyperlink are different based on different abilities of learners, namely, the adaptive navigation and adaptive content presentation; the learning evaluation module is to determine if the content and the presentation of learning content appropriate for learner's demand, and if the learning objectives can be achieved and the learning need is meted, the evaluation results not only can provide a basis for further study, but also provide reference of the next learning, this will help the update of learner model, including adaptive rules update, learning resources update, knowledge base update; the learner model module provides accurate description of the learner characteristics, it is the main evidence of adaptive content presentation and adaptive navigation, in addition, system can carry out collaborative learning and learning resources collaborative recommendations according to the similarity of learners'features; the role of collaboration and communication module is to provide a discuss platform for the learners in the learning process, we emphasize personalized learning, for individual student his learning process is entirely personalized, but the collaboration and exchange is equally important, learners not only need to express their views in the process of learning, but also need to refer to other people's point of view, this kind of exchange and discussion is essential for learning, which enables learners to understand the knowledge in different perspectives and enrich their own cognitive structure; the role of intelligent question and answering module is to acquire the learner's feedbacks and problems encountered in the study, in the network learning environment, the student and instructor are separated in space, and therefore system need to provide answering the questions to help students make their own decisions, the System can know if the learning effect is good and if the arrangement of learning content is reasonable by mining students'questions; learning support tool is primarily designed to assist students to annotate on the learning content in the learning process, the main tools include annotation tool and search tool, annotation tool is used to add the personalized key words on content fragments so that students can focus on these places in learning, the search tool is used to search information and find the relevant documents and help students better understanding. The seven modules are indispensable to each other in close cooperation and support adaptive learning effectively; they are beneficial experience and references in the development of learning system in education domain, while this structure can be the reference for the other adaptive hypermedia system. Finally, in order to understand the structure of the system better, we analyzed the workflow of the system running based on the above structure, discussed the input information, processing and output information of each step.In knowledge representation and management, we used ontology for representation. The ontology is formal, explicit, consistent, and sharable, these features can meet the knowledge representation requirements. We used OWL DL ontology language which is based on description logic to represent the ontology, it not only guaranteed formal but also have a good support in intelligent reasoning. The domain knowledge was divided into four layers: learning target layer, domain ontology layer, learning resources layer, learning resources describe layer and learning resource storage layer. Domain ontology layer include three sub-layers: concept ontology layers, learning methods concept layers, and application ontology layers. The concept ontology layers includes the key domain concepts, the learning methods concept layers includes the learning methods designed by the domain experts, the application ontology layers includes the strategic knowledge. This four layers structure has the advantage of enhancing the system's maintainability and scalability, such a structure has more conducive for knowledge management. Knowledge base can adapt to different versions of the tutorial book published by different publishers, it also can be modified simply, and converts to a new flexible curriculum for different type learners.In learner characteristics modeling, ontology-based representation method is used. This method is the algorithm-oriented, it is a specific data structure and formal description. Based on the analysis of the previous research results and relevant national standards, the learner characteristic model includes five aspects: learner basic information, learner prior knowledge level, learner preference, learner cognitive skill, and learner test scores. These five aspects have different roles in the system, they are the reference for the adaptive recommendation. In these features, the learner prior knowledge level is the most important feature, it is not only an important reference in choosing the learning objects precisely and evaluation of learning outcomes, but also the reference of master learner's state in learning. The accurate analysis of learner preference is the premise of active learning for learner, and only providing the preferred learning resources for learners, the learners can active learn and have a good learning. The accurate expression of cognitive skill ensures that system can provide moderate difficulty learning resources and avoid the learners'frustration in the learning process. The phase test scores can explain the learners'learning attitude. In addition, the relevant network characteristics of learners in learning activities have also have a certain influence on learning effect, including learners'technology skills, information literacy, the attitude of computer network, adaptation in network environment. Although these network characteristics have a certain degree affection on learning, but after all it is very small, we ignored them. Finally, we gave the representation of these features.In the inference algorithm, we proposed two algorithms based on the analysis of the existing algorithms, one is a semantic similarity algorithm based on semantic distance, and the other is a recommendation algorithm based on the learning resource score and semantic distance between concepts. The second algorithm is better than the first algorithm, the main reason is: we considered the learning resource score gave by the learners. This way can improve the effect of the first algorithm which only depends on the semantic similarity between concepts.In this paper, we discussed the framework of adaptive learning system, representation and management of knowledge, learner characteristics modeling, and adaptive reasoning method deeply, proposed a series of solutions, improve the system recommendation performance, improved the effectiveness of learning content recommendation, increased the learner's satisfaction, our work provides a useful reference for the research on adaptive learning system.