Expanding hierarchical contexts for constructing a semantic word network

Ustalov, Dmitry

URL:	http://www.dialog-21.ru/media/3982/dialogue2017_v1...
Additional URL:	http://www.dialog-21.ru/media/3959/ustalovda.pdf
Document Type:	Conference or workshop publication
Year of publication:	2017
Book title:	Computational Linguistics and Intellectual Technologies : Papers from the Annual conference "Dialogue 2017", Moscow, May 31-June 3, 2017
The title of a journal, publication series:	Dialogue
Volume:	16,1
Page range:	369-381
Conference title:	Computational Linguistics and Intellectual Technologies 2017
Location of the conference venue:	Moscow, Russia
Date of the conference:	May 31 - June 3, 2017
Place of publication:	Moscow, Russia
Publishing house:	RSUH
ISSN:	2221-7932 , 2075-7182
Related URLs:	http://www.dialog-21.ru/media/4160/ustal...
Publication language:	English
Institution:	School of Business Informatics and Mathematics > Wirtschaftsinformatik III (Ponzetto 2016-)
Subject:	004 Computer science, internet
Keywords (English):	lexical semantics , hyponym , hypernym , subsumption , semantic network , crowdsourcing , Russian
Abstract:	A semantic word network is a network that represents the semantic relations between individual words or their lexical senses. This paper proposes Watlink, an unsupervised method for inducing a semantic word network (SWN) by constructing and expanding the hierarchical contexts using both the available dictionary resources and distributional semantics’ methods for is-a relations. It has three steps: context construction, context expansion, and context disambiguation. The proposed method has been evaluated on two di erent datasets for the Russian language. The former is a well-known lexical ontology built by the group of expert lexicographers. The latter, LRWC (“Lexical Relations from the Wisdom of the Crowd”), is a new resource created using crowdsourcing that contains both positive and negative human judgements for subsumptions. The proposed method outperformed the other relation extraction methods on both datasets according to recall and F1-score. Both the implementation of the Watlink method and the LRWC dataset are publicly available under libré licenses.