The main factors for the quality of a semantic space in the traditional Chinese intelligent tutoring system using limit concept in calculus as an example

摘要

In recent years, due to the advance of new technologies, the idea about intelligent tutoring system (ITS) has achieved many successes. The quality of the employed semantic space is a crucial element for establishing an ideal ITS. How accurate the student's responses can be identified and how precise the teaching content can be distributed all depend on the characteristics of the semantic space. A semantic space can be derived from the so-called term-document matrix or word-word co-occurrence matrix. However, how the associated factors, such as the number of terms, the number of documents, and the length of the documents, will affect the quality of the semantic space is still an important question. This issue is particular substantial in some application fields, such as the limit concept in Calculus written in traditional Chinese, where the accessibility of documents might be a problem due to its rareness or cost. In this research, several candidate factors were adopted for evaluating their effects on the quality of the semantic space using semantic representation analysis. Also, a corresponding scree plot is proposed to observe the effect of factors. Moreover, this scree plot can serve as a valuable tool and reference during the process of building a semantic space.