The data release (lexicon.csv) contains a list of 400 noun compounds (such as "business magazine" or "fist fight") that were annotated for sentiment using crowdsourcing as a part of a research project at ETS. A positive/negative/neutral rating for each compound and for each of the words separately is provided. The data was used to study "combinatorics" of sentiment, namely, how sentiment profiles (pos/neg/neu) of single words combine to yield the profile for the compound. The paper describing our approach and results should be cited if this dataset is used:
Beata Beigman Klebanov, Jill Burstein, and Nitin Madnani. 2013. Sentiment Profiles of Multiword Expressions in Test-taker Essays: The Case of Noun-noun Compounds. ACM Trans. Speech Lang. Process. 10(3):12. DOI=http://dx.doi.org/10.1145/2483969.2483974
@article{BKBM2013,
author = {Beigman Klebanov, Beata and Burstein, Jill and Madnani, Nitin},
title = {Sentiment Profiles of Multiword Expressions in Test-taker Essays: The Case of Noun-noun Compounds},
journal = {ACM Trans. Speech Lang. Process.},
volume = {10},
number = {3},
year = {2013},
pages = {12:1--12:15},
articleno = {12},
doi = {10.1145/2483969.2483974}
}Each of the 13 columns in lexicon.csv are explained in the table below.
| column name | explanation | example |
|---|---|---|
| nn compound | the text of the compound | airport security |
| DEV/TEST | whether this is part of DEV or TEST (see paper) | DEV |
| pos | positivity score for the compound (between 0 and 1) | 0.381 |
| neg | negativity score for the compound (between 0 and 1) | 0.095 |
| neu | neutrality score for the compound (between 0 and 1) | 0.524 |
| noun1 | the left noun in the compound | airport |
| noun1.pos | positivity score for the left noun alone (between 0 and 1) | 0.15 |
| noun1.neg | negativity score for the left noun alone (between 0 and 1) | 0 |
| noun1.neu | neutrality score for the left noun alone (between 0 and 1) | 0.85 |
| noun2 | the right noun in the compound | security |
| noun2.pos | positivity score for the right noun alone (between 0 and 1) | 0.65 |
| noun2.neg | negativity score for the right noun alone (between 0 and 1) | 0 |
| noun2.neu | neutrality score for the right noun alone (between 0 and 1) | 0.35 |
Noun-Noun Compound Sentiment Lexicon by Educational Testing Service is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
