Paper: Unsupervised Feature Selection for Relation Extraction

ACL ID I05-2045
Title Unsupervised Feature Selection for Relation Extraction
Venue International Joint Conference on Natural Language Processing
Session poster-demo-tutorial
Year 2005

This paper presents an unsupervised re- lation extraction algorithm, which in- duces relations between entity pairs by grouping them into a “natural” num- ber of clusters based on the similarity of their contexts. Stability-based crite- rion is used to automatically estimate the number of clusters. For removing noisy feature words in clustering proce- dure, feature selection is conducted by optimizing a trace based criterion sub- ject to some constraint in an unsuper- vised manner. After relation clustering procedure, we employ a discriminative category matching (DCM) to find typi- cal and discriminative words to repre- sent different relations. Experimental results show the effectiveness of our al- gorithm.