Lsislif CRF and Logistic Regression for Opinion Target…和.pdfVIP

Lsislif: CRF and Logistic Regression for Opinion Target Extraction and Sentiment Polarity Analysis Hussam Hamdan Patrice Bellot Frederic Bechet Aix-Marseille University Aix-Marseille University Aix-Marseille University hussam.hamdan@lsis.org patrice.bellot@lsis.org frederic.bechet@lif.univ-mrs.fr Abstract which all entities and their aspects should be ex- tracted and the sentiment towards them should also This paper describes our contribution in Opin- be determined. ion Target Extraction OTE and Sentiment Po- Aspect-Based SA task consists of several sub- larity sub tasks of SemEval 2015 ABSA task. A CRF model with IOB notation has been problems, the document is about many entities adopted for OTE with several groups of fea- which could be for example a restaurant, a laptop, tures including syntactic, lexical, semantic, a printer. Users may refer to an entity by different sentiment lexicon features. Our submission writings, but normally there are not a lot of vari- for OTE is ranked fifth over twenty submis- ations to indicate the same entity, each entity has sions. A Logistic Regression model with a many aspects which could be its parts or attributes. weighting schema of positive and negative la- Some aspects could be another entity such as screen bels have been used for sentiment polarity; of laptop, but most work did not take this case into