RALPH » Building RGrep: A Case Study on Domain-Specific Search Engines
Web search is a rapidly growing space, and one of the hardest aspects of web search is determining query intent. Domain specific search engines have emerged as a potential solution to this lack of context, but current approaches do not scale to the thousands of domains one can define. RGrep is an experimental search engine that extracts methods for automatic content discovery and relevancy scoring from social bookmarking information. If successful, these extractions can be generalized to allow for automatic creation of topical search engines with a high degree of relevancy and precision. This talk will briefly describe the problem area, and then discuss some of the very practical challenges faced when building a search engine.