Paper: Evaluating Distant Supervision for Subjectivity and Sentiment Analysis on Arabic Twitter Feeds

ACL ID W14-3624
Title Evaluating Distant Supervision for Subjectivity and Sentiment Analysis on Arabic Twitter Feeds
Venue Workshop on Arabic Natural Language Processing
Session
Year 2014
Authors

Supervised machine learning methods for automatic subjectivity and sentiment anal- ysis (SSA) are problematic when applied to social media, such as Twitter, since they do not generalise well to unseen topics. A possible remedy of this problem is to ap- ply distant supervision (DS) approaches, which learn from large amounts of auto- matically annotated data. This research empirically evaluates the performance of DS approaches for SSA on Arabic Twitter feeds. Results for emoticon- and lexicon- based DS show a significant performance gain over a fully supervised baseline, es- pecially for detecting subjectivity, where we achieve 95.19% accuracy, which is a 48.47% absolute improvement over previ- ous fully supervised results.