Extracting Product Features and Opinions from Reviews - Natural Language Processing and Text Mining - page 25

mantic orientation of a word (in the context of a product feature) given its
prior semantic orientation provided by an initially supplied word list. opine's
approach, while independently developed, amounts to a more general version
of the approach taken by [7]: opine automatically computes both the prior
and final word semantic orientation using a relaxation labeling scheme which
accommodates multiple constraints. [23] uses a supervised approach incorpo-
rating a large set of features in order to learn the types of linguistic contexts
which alter a word's prior semantic orientation. The paper's task is different
than the one addressed by opine and [7], as it involves open-domain text and
lacks any information about the target of a particular opinion.
[13] suggests using the magnitude of the PMI-based SO score as an indi-
cator of the opinion's strength while [24, 25] use a supervised approach with
large lexical and syntactic feature sets in order to distinguish among a few
strength levels for sentence clauses. opine's unsupervised approach combines
Turney's suggestion with a set of strong ranking constraints in order to derive
opinion phrase rankings.
2.5 Conclusions and Future Work
opine is an unsupervised information extraction system which extracts fine-
grained features, and associated opinions, from reviews. opine's use of the
Web as a corpus helps identify product features with improved precision com-
pared with previous work. opine uses a novel relaxation-labeling technique to
determine the semantic orientation of potential opinion words in the context
of the extracted product features and specific review sentences; this technique
allows the system to identify customer opinions and their polarity with high
precision and recall. Current and future work includes identifying and analyz-
ing opinion sentences as well as extending opine's techniques to open-domain
text.
2.6 Acknowledgments
We would like to thank the members of the KnowItAll project for their com-
ments. Michael Gamon, Costas Boulis, and Adam Carlson have also pro-
vided valuable feedback. We thank Minquing Hu and Bing Liu for providing
their data sets and for their comments. Finally, we are grateful to Bernadette
Minton and Fetch Technologies for their help in collecting additional reviews.
This research was supported in part by NSF grant IIS-0312988, DARPA
contract NBCHD030010, ONR grant N00014-02-1-0324 as well as gifts from
Google and the Turing Center.
References
1. Turney, P.D.: Thumbs up or thumbs down? semantic orientation applied to
unsupervised classification of reviews. In: Procs. of ACL. (2002) 417-424