An FCA System based on Artificial Intelligence and an Ontology Model
Full recordShow full item record
This paper proposes a Formal Concept Analysis (FCA) system for Learning Object Repositories (LORs) in order to help extracting concepts of learning objects. In our approach, a LOR is a system that stores learning objects on the Web and/or their metadata and a mediator is a system that collects metadata of learning objects and classifies learning objects by concepts and their words or phrases in order to help searching and retrieving these learning objects. Our FCA system is based on an ontology model and Artificial Intelligence (AI) techniques and it extends well-known educational metadata standards based on IEEE LOM (Learning Object Metadata) such as SCORM, CANCORE, Normetic, etc. Our Formal Concept Analysis system is composed of three main components, (i) a concept definition model, (ii) a concept extraction model and (iii) a concept classification model. The concept definition model provides functions to define, index, search concepts and their related words or phrases by using AI technique. The concept extraction model uses metadata exported from existing LOR tools such as PALOMA. These metadata are imported into our mediator by using standard document formats such as XML. Title, keywords and description based on XML format are extracted and compared with existing concepts by using AI techniques. The Result of this step is a set of concepts which are extracted from metadata. However, our concept extraction model is a semi-automatic system which some results have to be validated by human. At some steps the system goes back to the prior step to give other definition of concept. In our system, concept extraction can be redone until meet a good solution. The concept classification model provides functions to annotate and classify concepts by using an existing knowledge management model such as SKOS (Simple Knowledge Organisation System) based on an ontology model. Standard words and their annotations or meaning from a semantic network such as WordNet are used for describing these concepts. Our ontology structure is created to relate concepts based on WordNet and defined concepts from our concept definition model. This approach is implemented using Formal Concept Analysis system in order to extract, classify concepts of learning objects in our LOR and using Artificial Intelligence techniques to help concept extraction. To experiment with this system, our mediator is developed as a prototype by using PALOMA as a LOR management tool, Protégé as a ontology management tool, Corese (Conceptual Resource Search Engine), SeWeSe (Semantic Web Server and JSP Library) as a semantic search engine and TuProlog as an Artificial Intelligent language. Our user interfaces are created by using Java Server Pages (JSP) technologies.