Thumbs up? Sentiment Classification using Machine Learning Techniques

2 Previous Work

1. How does sentiment classification differ from genre classification and subjectivity detection?
   Genre and subjectivity techniques can recognize that a document expresses an opinion (e.g. an editorial), but not what the opinion is: positive or negative.
   • Pang et al., 2002, Thumbs up? Sentiment Classification using Machine Learning Techniques (Proceedings of the 2002 Conference on Empirical Methods in Natural Language Processing (EMNLP 2002)), §2, pp. 79–80
2. How was earlier sentiment work typically done, and what did Pang et al. do instead?
   It was largely knowledge-based: semantic orientation of words from linguistic heuristics or seed words, models from cognitive linguistics, or hand-built discriminant-word lexicons. Pang et al. used prior-knowledge-free supervised machine learning to understand the task’s inherent difficulty.
   • Pang et al., 2002, Thumbs up? Sentiment Classification using Machine Learning Techniques (Proceedings of the 2002 Conference on Empirical Methods in Natural Language Processing (EMNLP 2002)), §2, pp. 79–80
3. How did Turney (2002), the closest prior work, classify reviews?
   Unsupervised: the semantic orientation of phrases was estimated from their mutual information with “excellent” vs “poor”, using search-engine hit counts. Movie reviews were the hardest of his domains (65.83% on 120 documents).
   • Pang et al., 2002, Thumbs up? Sentiment Classification using Machine Learning Techniques (Proceedings of the 2002 Conference on Empirical Methods in Natural Language Processing (EMNLP 2002)), §2, pp. 79–80