Intelligent Interaction Support for E-learning
Takashi Yukawa, Yoshimi Fukumur · InTech eBooks · 2010
1 Requirements for iBBS Inherently, a BBS for an e-Learning course cannot generate active discussions of the subject.If a large number of articles are posted on the BBS every day, reading every article becomes burdensome to both teachers and students, and may cause some students to drop out of the discussion.Therefore, it is desirable to reduce the workload for discussion participants.To realize this workload reduction, an intelligent communication support function, which is an important IPN function, is embedded in the BBS.We call this system iBBS, and it is achieved with natural language processing techniques. The IPN Function in iBBSThe IPN function, illustrated in Figure 1, operates as follows.1.A participant (student B in the figure) registers his/her words of interest (keywords) into the system.2. Another participant (student A in the figure) posts an article that includes the content of interest to the student B. 3. The system notifies student B of the posting.The IPN function must notify the posting of the article by including not only the keywords themselves but also similar words.For example, assume that student B registers an "urgent stop" as a keyword, and student A posts an article including the word "emergency stop."The system must be able to notify student B of this article because the article discusses "urgent stopff even though the poster uses word "emergency stopff to express it.A message in the BBS has a title line expressing its subject; however, a novice user often uses an irrelevant title, e.g., "question" or "request."Thus, the system has to check the content of the posted message.Consequently, to apply the IPN function, the system must understand the meanings of words.Although machine-readable dictionaries [3] are commonly used for this purpose, such dictionaries generally consist of words for daily use rather than technical terms used in specialized fields.The authors previously proposed the Concept-based Vector Space Model (CBVSM), which is able to capture the relationships between words used in target texts [14].CBVSM provides a function that discerns the semantic similarity between words that appear in texts; this capability enables the system to process texts as if the technical terms are understood.CBVSM and the procedure to construct the concept base are introduced below, and then an implementation of the IPN function of iBBS is described.A performance (accuracy) evaluation of the IPN function is also presented.E-learning, experiences and future