Abstract: In this paper the main focus is on identifying opinion features via its distributional disparities across two Corpora domain dependent which is intrinsic domain and domain independent which is extrinsic domain. In the Initial stage candidate features are extracted via syntactic rules, then in the second stage we calculate domain relevance Score of candidate features related to intrinsic domain which is Intrinsic Domain relevance score(IDR Score) And that of extrinsic domain is Extrinsic Domain Relevance Score (EDR Score). At the very last stage we identify the Opinion features by thresholding activity. Those candidate features intrinsic domain relevance score greater than Threshold and those that having score less than extrinsic threshold is identified as opinion feature. We further determine the polarity of opinion feature whether it is positive, negative or neutral
Keywords: Candidate features, Intrinsic Domain Relevance Score (IDR Sore) and Extrinsic Domain Relevance Score (EDR Score),Opinion Features.