|
|
|
     |
                                     |
Return to Text systems and web issues
This paper presents the BINGO! focused crawler, an advanced tool for information portal generation and expert Web search. In contrast to standard search engines such as Google which are solely based on precomputed index structures, a focused crawler interleaves crawling, automatic classification, link analysis and assessment, and text filtering. A crawl is started from a user-provided set of training data and aims to collect comprehensive results for the given topics. The focused crawling paradigm has been around for a few years and many of our techniques are adopted from the information retrieval and machine learning literature. BINGO! is a system-oriented effort to integrate a suite of techniques into a comprehen- sive and versatile tool. The paper discusses its overall architecture and main components, important lessons from early experimentation and the resulting improvements on effectiveness and efficiency, and experimental results that demonstrate the usefulness of BINGO! as a next-generation tool for information organization and search.
@inproceedings {DBLP:conf/cidr/SizovTSWGBZ03, author = {Sergej Sizov and Martin Theobald and Stefan Siersdorfer and Gerhard Weikum and Jens Graupmann and Michael Biwer and Patrick Zimmer}, booktitle = {CIDR}, title = {The BINGO! System for Information Portal Generation and Expert Web Search.}, year = {2003}, url = {db/conf/cidr/cidr2003.html#SizovTSWGBZ03}, ee = {http://www-db.cs.wisc.edu/cidr/program/p7.pdf}, bibsource = {DBLP, http://dblp.uni-trier.de} }  ©2004 Association for Computing Machinery |
