Efficient Text Content Extraction and Browsing of WWW Documents Using the Abstract Text Viewer.
Richard H. Fowler, Yavuz Tor, David Navarro, Laura M. Grabowski · International Conference on Internet Computing · 2003
The Abstract Text Viewer (ATV) is an integrated suite of text reading tools for electronic documents designed to increase efficiency and effectiveness of content extraction. ATV reads a HTML formatted document to create more abstract representations, such as a heading structure for overviews. The system uses both well-known techniques for text representation and novel display and content extraction techniques. Initially, documents are displayed with an overview + detail model. The detail window displays the entire document text, possibly highlighting keywords, and an overview window displays document’s headings together with a semantic summary of each section based on keyword extraction. The system provides additional methods to improve readability, such as fish-eye view, zooming, and highlighting.