Using frame-based resources for sentiment analysis within the financial domain

摘要: User-generated data in blogs and social networks have recently become a valuable resource for sentiment analysis the financial domain, since they been shown to be extremely significant marketing research companies public opinion organizations. In order identify bullish bearish sentiments associated with stocks, we propose fine-grained approach that returns continuous score $$[-\,1,+\,1]$$ range. Our supervised leverages frame-based ontological which produces feature sets such as lexical features, semantic features their combination. One of outcome our suggests used might successfully applied within domain achieving better results than traditional methods do not embody semantics. We also show higher performance based solely on evaluation specific substrings message, rather extracted from whole text microblog message through resource. compared system semi-supervised unsupervised approaches indicate outperforms others. Last but least, is general can top any existing method polarity detection.