Efficienct Probing for Extracting New Topic Contents from Hidden Web Sites
Takanori Mouri, Hiroyuki Kitagawa · 2004
There are many information sources which provide their database contents through query interfaces. Hidden Web sites are typical examples. Usually, their database contents dynamically change, new documents on emerging topics being ap- pended. In applications like topic detection and trend analysis, we want to discover newly emerging contents in the databases. However, it is very difficult for ordinary users to detect them only through the query interfaces without support by the database contents administrators. We proposed a method to automatically discover such content. The proposed method generates biased query probes using a classifier to be issued to a given text database with a keyword-based query interface. In this paper, we improve a method for more efficient probing. We evaluate its effectiveness with preliminary experiments.