SHELLFBK: An Information Retrieval-based System For Multi-Domain Sentiment Analysis

摘要: This paper describes the SHELLFBK system that participated in SemEval 2015 Tasks 9, 10, and 11. Our takes a supervised approach builds on techniques from information retrieval. The algorithm populates an inverted index with pseudo-documents encode dependency parse relationships extracted sentences training set. Each record stored is annotated polarity domain of sentence it represents. When or new has to be computed, converted query used retrieve most similar retrieved instances are scored for relevance query. relevant instant assign label sentence. While results well-formed encouraging, performance obtained short texts like tweets demonstrate more work needed this area.