Visual and textual summarization of webpages

Abstract

Search and re-finding tasks are among the most typical applications on internet. In order to make these tasks more efficient, we propose an attractive new scheme to visually summarize web pages. This would allow users to quickly get the idea of what the webpage is all about and helping users to recall the visited web pages. For the purpose, we employ latent semantic analysis to prepare a gist of the textual contents accompanied by title and a relevant image. The image could be internal or external depending upon the contents of the webpage. If the webpage contains dominant internal images, ranking algorithm is employed to select the most dominant internal image for summarization. If not, key phrases are extracted from the textual part for the purpose of searching the entire internet for a relevant external image. After the image search results are obtained, re-ranking algorithm is applied to select the most appropriate image.