Word sense discrimination in information retrieval: A spectral clustering-based approach
作者：
Highlights：
• We propose a new unsupervised method that uses word sense discrimination in IR.
• The re-ranking method is based on spectral clustering.
• The effectiveness over queries with ambiguous terms is proved on TREC corpora.
• Our interest regards improving the precision after 5, 10 and 30 retrieved documents.
• The method improves results for queries with poor baseline performance.
摘要
•We propose a new unsupervised method that uses word sense discrimination in IR.•The re-ranking method is based on spectral clustering.•The effectiveness over queries with ambiguous terms is proved on TREC corpora.•Our interest regards improving the precision after 5, 10 and 30 retrieved documents.•The method improves results for queries with poor baseline performance.
论文关键词：Information retrieval,Word sense disambiguation,Word sense discrimination,Spectral clustering,High precision
论文评审过程：Received 29 July 2013, Revised 7 August 2014, Accepted 22 October 2014, Available online 19 November 2014.
论文官网地址：https://doi.org/10.1016/j.ipm.2014.10.007
