| In recent years, research in the domain of Natural Language Understanding (NLU) has benefited from the methodological compilation of context corpora with rich syntax such as the Brown corpora of the Penn Treebank, and machine-readable semantic lexicons such as WordNet. It has also benefited from developments of parsers for syntactic information presentation and semantic meaning retrieval using such corpora. Both types of activities have also advanced the progress in the comprehension of semantic meanings of natural contexts.; The comprehension of semantic meanings is at the heart of the literature of modern linguistics. Inspired largely by the work of the current view of natural language processing, the focus of language processing is isolated linguistic items, research in statistical literature linguistics has focused primarily on the statistical representation of linguistic occurrences. With the help of a wide range of context corpora, this statistical representation ensures a virtually comprehensive coverage of lexical combinational occurrences, which allows a computer program to identify the recurrence of a combination and to further predict the meaning of the recurrence. Due to the limitations of current computer technology, however, representing a lexical combination is restricted to a finite length. Since the statistical representation is confined to describing a finite term-span context, which impedes a computer program from the understanding of contextual meaning, we focus attention on obtaining an approximate but simpler and more satisfactory solutions through soft computing techniques; in particular, fuzzy set theory.; The proposed fuzzy methodology is based on the definition of fuzzy formalism. This involves the comprehensive study of the interplay between the syntactic structure of a verb and the semantic roles that the verb plays. The theory underlying fuzzy formalism is that the syntactic representations associated with a particular verb indicates an explicit range of semantic meanings. Through applying fuzzy formalism to the training corpora, the quality of quantitative representation of linguistic items is improved. (Abstract shortened by UMI.)... |
