Learning Object Querying with Ontology Concept Expansion － Case Study Using JLOO

• Main Authors: Yue-Shiun Tsai, 蔡岳勳
• Other Authors: Tzong-I Wang
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/12135028594531778948

碩士 === 國立成功大學 === 工程科學系碩博士班 === 94 === 　　In the past few years, the vigorous development of Internet makes enormous information resources obtainable with a single click. However this huge amount of resources also cause recognition burden on users when trying to grab what they need. Although search engines are convenient in collecting as if relevant documents, they do not help much in identifying desirable ones. People who search still have to spend a lot of time on filtering and choosing information they needed. For a novice learner trying to utilize learning resources from repositories on the network, it is already difficult to figure out how to find suitable teaching materials, let alone facing such lot of irrelevant course objects. 　　This study focuses on developing a methodology for helping learners who are novel to a specific field. The thesis uses java programming language as an example for demonstration. The Java Learning Object Ontology (JLOO) is used as system ontology that has being developed as a guide in this field for the learners. The proposed methodology uses ontology to infer a learner’s intention before using a concept expansion algorithm that, base on the learners’ intention, to include more relevant concepts as the query when retrieving learning objects. The most relevant learning objects could be fetched even when the learners are not familiar with the domain. 　　The concept expansion algorithm uses traditional tf-idf formula to calculate the relevant degree of the basic concepts on an ontology hierarchy, which are matched with keywords of a user query. It then calculates the influence of other concepts on the semantic path of basic concepts. Some user intention trees are establish, from which a most suitable one is selected. The result one is further expanded according to system logs that record the past learning behaviors of the user and other users. Finally, the most relevant learning objects are fetched and recommended to the user.