Abstract
Abstract In the paper we discuss the possibilities of using hierarchical contextual ontologies for supporting sentiment classification tasks. The discussion focuses on two important research hypotheses: (1) whether it is possible to construct such an ontology from a corpus of textual document, and (2) whether it is possible and beneficial to use inferencing from this ontology to support the process of sentiment classification. To support the first hypothesis we present a method of extraction of hierarchy of contexts from a set of textual documents and encoding this hierarchy into a multi-level contextual ontology. To support the second hypothesis, we present a method of reasoning from the ontology, and results of experimental verification, which show that use of this reasoning method can increase the accuracy of sentiment classification for longer text documents.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
