next up previous contents
Next: 目次   目次

概要:

This study examines the similarities of documents by text mining.

Here, the documents are syllabuses from the department of electrical and computer engineering in Gifu National College of Technology.

The results will help motivating students to learn.

To calculate the degree of similarity, the following methods are adopted to our school's syllabuses.

The methods are as follows.

  1. Use Network Analysis.
  2. Use Latent Semantic Analysis.
  3. Use Multi Dimensional Scaling.

As a result, similarity calculation by Network Analysis is possible.

But the results were not accurate.

Next, the degree of similarity was calculated by Latent Semantic Analysis with and without the importance of words.

The analysis with importance of words led to better results than the one without importance did.

The result became intelligible when similarity was visually expressed with the multidimensional scaling method.





Deguchi Lab. 2013年2月28日