An insight into task of opinion mining

Abstract

Recent years have brought the burst of popularity of community websites across the internet of opinionated text on web. Users express their views and opinions regarding products and services. These opinions are subjective information which represents user’s sentiments, feelings or appraisal related to the same. People use such opinion rich sources to formalize knowledge and analyze it for further reuse. This leads to emergence of new field opinion mining which differs from traditional fact based information mining which are generally done by current search engines. With introduction of Blog track in TREC 2006, a considerable work has been done in this field which comprises of opinion mining at sentence level, passage or document level and feature level. This paper presents an insight into task of opinion mining. We find that task of opinion mining is directly related to degree of formalism of language used in data sources.