This paper describes our entry into the TREC 2010 Web track. We extracted and ranked results for both last year’s and this year’s topics from the ClueWeb09 corpus using a parallel processing pipeline that avoids the generation of an inverted file. We describe the components of the parallel architecture and the pipeline and how we ran the TREC experiments, and we present effectiveness results.

PDF file

In  Proc. of the 19th Text Retrieval Conference (TREC)

Publisher  National Institute of Standards and Technology

Details