Automating assignment of semantic relations in TANKA

Title: Automating assignment of semantic relations in TANKA
Authors: Bhatia, Ramanjot Singh
Date: 2006
Abstract: Theory of Semantic Relations is one of the oldest linguistic theories dating back to at least 5th century BC. Over the years a multitude of different theories and sets of relations have been developed in order to capture the semantics of a language. This has further gained importance with the desire to have artificial intelligence systems with the ability to comprehend human language. The focus of this thesis is such a system, TANKA. TANKA is a knowledge engineering aid which helps extract semantic relations from a semi-technical text as pairs of syntactic units and assigns them semantic relations. This thesis presents a feasibility study of automating the assignment of semantic relations in TANKA. The thesis looks at different methods and resources that will help to progress the system from a semi-automatic process to a fully automatic process. For the purposes of analyzing the feasibility of this task, this thesis constructs a new dataset and looks into the use of two feature-rich parsers as sources of syntactic features. It also studies the utility of semantic resources such as WordNet, LDOCE and Roget's in order to find features for automating the assignment of semantic relations. A variety of combinations of these are tested to get an insight into the feasibility of this task.
CollectionTh├Ęses, 1910 - 2010 // Theses, 1910 - 2010
MR18400.PDF4.51 MBAdobe PDFOpen