gzvulon / Serelex

A tool for semantic relation extraction. The program finds pairs of semantically related words based on the text definitions coming from the Wikipedia articles (other texts may be also used). The extraction method implemented in this system is based on three similarity measures (cosine, gloss overlap, and Karaulov's measure) between texts and two nearest-neighbor algorithms (KNN and Mutual KNN). The tool is a cross-platform console application.