next up previous contents
Next: 目次   目次

概要:

Today, documents and books are computerized rapidly with social informatization. It is very favorable condition for the text mining which deals many documents, and this technique will have been used from now on. The purpose of this study is to search for proper method of calculating similarity between documents by the text mining. To calculate the similarity, the following methods are adopted to our school's syllabuses.

  1. Use direct term matching method.
  2. Use Latent Semantic Analysis.
  3. Use Probabilistic Latent Semantic Analysis.

As a result, suitable connections among syllabuses are discovered successfully like the course diagram. In addition, proper connections that aren't described on the subject diagram, are found out. These results are appeared mainly when the syllabuses are applied LSA, but PLSA that is said to be superior to the direct term matching and LSA, didn't give full play to its ability.





Deguchi Lab. 2011年3月4日