Discovering secrets from texts: A self-organizing map perspective

Ordinary text documents such as text books, medical records, reports, research articles, Web pages, and mails dominate the information storage and dissemination in our daily life. However, it is difficult to process and extract useful information from them due to their unstructured nature. Different methodologies have been devised to discover various types of knowledge underlying massive amount of text documents. In this talk, I will first address the general problem of text mining which aims to discover interesting knowledge from unstructured text documents. The application of the self-organizing map (SOM) model, which is well reputed as a good tool for data clustering, in resolving this issue is then discussed. I will cover the basic schemes of SOM as well as some others with upgraded features. Different aspects of text mining using these schemes will also be shown with experimental results.