Reranking Documents with Antagonistic Terms

摘要

For the participation of the IICS group at the domain-specific task (GIRT) of the CLEF campaign 2006, we employ cooccurrence of search terms and antagonistic terms (antonyms or cohyponyms) in documents to derive values for reranking an initial result set. A reranking test on GIRT 2004 data showed a significant increase in mean average precision (MAP), i.e. a change from 0.2446 MAP to 0.2986 MAP. Precision for the submitted runs for the domain-specific (DS) task did not change significantly, but the setup for the best experiment included a reranking of result documents (0.3539 MAP). For reranking a result set with an already high MAP (provided by the Berkeley group), a significant decrease in precision was observed (MAP dropped from 0.4343 to 0.3653).