Text Mining and Logic Model for University Midterm Plans

摘要

National universities in Japan are required to create midterm plans based on logic models. However, universities are still in trouble about what item should be written about five viewpoints of the logic model - purpose, input, activity, output, and outcome. In this paper, we paid attention to the analysis of the third term midterm plan by the expert who extracted the typical sentences which represent the viewpoints. We applied the machine learning for those extracted sentences to identify the characteristic (feature) words and pattern of each viewpoint. It turned out that the activities have simple pattern and can be recognized with high identification performance. However, the output and the outcome are hard to describe the viewpoints.