Finding subjectivity clues; sentence and clause-level classification Reactions
Sentiment Retrieval Using Generative Models authors: Eguchi, K. & Lavrenko, V. read by: John Knox
Sentiment retrieval: combined IR and
sentiment classification Novel idea Does it work? No reported recall, low precision…but competitive with similar systems
Automatic Identification Of Sentiment Vocabulary read by: Michael Lipschultz authors: Wilson, T., Wiebe, J., Hoffmann, P.
Two step: neutral-polar then determine
polarity Discussion of the role “not” and “will” Reduce some errors by allowing neutral terms in stage 2 Relation to “Identifying Subjective Adjectives through Web-based Mutual Information,” Baroni, M. & Vegnaduzzo, S. This work is concerned with finding opinions Wilson et. al is concerned with the next step
Using Emoticons to reduce Dependency in Machine Learning Techniques authors: Read J. read by: Yaw Gyamfi
Lack of rigor in analysis Rare work in that it looks at temporal
dependency…but is it persuasive? Title is misleading – emoticons are used more for reduction of annotation costs
Identifying Expressions of Opinion in Context authors: Breck, E., Choi, Y. & Cardie, C. read by: Matt McGettigan
Reviewer impressed with performance (close
to human annotators)…but… Concern with evaluation standards Subjective phrases as a natural extension from subjective adjectives
Feature Subsumption for Opinion Analysis authors: Riloff, E., Patwardhan, S. & Wiebe, J. read by: Mahesh
Considering dependencies among features
can add considerable performance Considering POS as subsuming unigrams
Mining the Peanut Gallery: Opinion Extraction and Semantic Classification of Product Reviews authors: Dave K., Lawrance S. and Pennock D.
Supposedly comparing IR vs. Machine
learning techniques, but the IR approach skews heavily towards machine learning approach Comparison to emoticon paper: explicit rating (self tagging) vs. automatic identification Granularity: some features that help at e.g. sentence level are less useful at the document level
Extracting Appraisal Expressions authors: Bloom, K., Garg, N. & Argamon., S. read by: Danielle Mowery
Significantly different annotation schema
compared to MPQA Author evaluated system output post facto Can’t evaluate precision Ample opportunity for bias
Major themes IR for sentiment analysis Problems with evaluation standards Weak standards Lack of rigor in analysis Not enough data supplied (e.g. accuracy only)
Increasing sophistication of features Multi-stage approaches Dependencies among features Levels of tagging (phrase/sentence/document)