Skip to Main Content
In recent researches, an anchor text and texts near the anchor in the original pages are used for the purpose of summarizing or classifying the target pages. No research considers these text parts as the people's views about the target page and tries to classify them. In our research, we try to extract the descriptions about the target pages from the original pages. We then classify these descriptions. In other manner, we classify the people's views about the target pages into the hierarchical directory. We call the result the Multi-People's Views Web Directory. In this paper, we concentrate on describing our preliminary survey of semantic text portions. We also explain our method for extracting semantic text portions. The experimental results show that our method achieves high accuracy.
Date of Conference: 19-21 May 2005