Teachers of autistic students must observe the overt behaviors of these students and acquire the necessary skills to analyze the causes of these behaviors. Teachers can build a low-stress environment by considering the characteristics of each student, which can be revealed by analyzing their behavioral factors. However, it is not easy for less-experienced teachers to perform such analysis. Therefore, we are developing a training package that aims to help less-experienced teachers acquire the observation and analytical skills necessary to accurately identify the behavioral factors of each student with autism. We carried out our preliminary experiment by using a prototype system based on the hypothesis that it is possible for even an less-experienced teacher to make correct assumptions about behavioral factors such as stress. The results of our preliminary experiment support this hypothesis.

This paper proposes a topic-independent method for automatically scoring essay content. Unlike conventional topic-dependent methods, it predicts the human score of a given essay without training essays written to the same topic as the target essay. To achieve this, this paper introduces a new measure called MIDF that measures how important and relevant a word is in a given essay. The proposed method predicts the score relying on the distribution of MIDF. Surprisingly, experiments show that the proposed method achieves an accuracy of 0.848 and performs as well as or even better than conventional topic-dependent methods.

This paper describes an edu-mining technique for finding keywords to improve pupils' message- production skills. It automatically finds keywords from blog items pupils create. It then adaptively suggests some of the keywords to pupils when they create new blog items so that they can revise their items using the keywords. Experiments show that it finds and suggests keywords surprisingly well despite its simple algorithm. They also show that pupils in an elementary school in Japan successfully improve the quality of blog items using the suggested keywords.

As an application of NLP to computer-assisted language learning(CALL), we propose a diagnostic processing of Japanese being able to detect errors and inappropriateness of sentences composed by the students in the given situation and the context of the exercise texts. Using LTAG(Lexicalized Tree Adjoining Grammar) formalism, we have implemented a prototype of such a diagnostic parser as a component of a CALL system being developed.