Quixote: Building XML Repositories from Topic Specific Web Documents.
Christina Yip Chung, Michael Gertz, Neel Sundaresan · 2001
Despite major advancements in information retrieval techniques employed by today's Web search engines, building applications that allow users to efficiently manage, query, and utilize large collections of related Web documents from diverse, highly heterogeneous sources is still a hard problem. Even in the case where potentially related documents that pertain to the same topic can be gathered efficiently using, e.g., a focused Web crawler, the documents are still heterogeneous both in terms of structure and presentation, due to different authorship. More importantly, the documents are marked up in HTML for visual rendering purposes, thus hampering sophisticated query schemes different from simple keyword-based searches. In this paper, we outline the concepts and methods underlying Quixote, a system that allows users to rapidly build XML document repositories from large collections of topic specific HTML documents. Such documents are assumed to be gathered by a top...