摘要：This article presents a new two-step method to handle and study large closed subsets of WordNet-type dictionaries with the goal of finding possible structural inconsistencies. The notion of closed subset is explained using a WordNet tree. A novel and very fast method to order large relational systems is described and compared with some other fast methods. All the presented methods have been tested using Estonian1 and Princeton WordNet2 largest closed sets. DOI: http://dx.doi.org/10.5128/ERYa9.10
关键词：thesaurus, closed set, seriation, Power Iteration Clustering (PIC), reducing number of crossings, WordNet
