Bootstrapped Lexicon of English Polarity Shifters

<p>We provide a bootstrapped lexicon of English polarity shifters and their shifting direction. We cover verbs, nouns and adjectives. Our lexicon provides 2521 shifters among a vocabulary of 9145 words, taken from WordNet v3.1 (Miller et al., 1990).</p> <p>We also provide a dataset of 2631 verb phrases that are annotated for shifting polarities. The phrases are taken from the Amazon Product Review Data corpus (Jindal &amp; Liu, 2008).</p> <p><strong>Data</strong></p> <p><strong>1. Polarity Shifter Lexicon</strong></p> <p>A list of 9145 words, annotated for whether they are polarity shifters. Contains 2631 shifters and 6514 non-shifters.</p> <ul> <li>File:&nbsp;<code>shifters.txt</code></li> <li>The lexicon is a comma-separated value (CSV) table</li> <li>Each line follows the format&nbsp;<code>POS,LEMMA,SHIFTER_LABEL,SOURCE</code>. <ul> <li><code>POS</code>: The part of speech of the word (<code>verb</code>,&nbsp;<code>noun</code>,&nbsp;<code>adj</code>)</li> <li><code>LEMMA</code>: The lemma representation of the word in question. Multiword expressions are separated by an underscore (<code>WORD_WORD</code>).</li> <li><code>SHIFTER_LABEL</code>: Whether the word is a polarity shifter (<code>SHIFTER</code>) or a non-shifter (<code>NONSHIFTER</code>)</li> <li><code>SOURCE</code>: Whether the word was part of the gold standard (<code>GOLD_STANDARD</code>) or was bootstrapped (<code>BOOTSTRAPPED</code>). All labels, both from gold standard and bootstrap output, were verified by a human annotator.</li> </ul> </li> </ul> <p><strong>2. Sentiment Verb Phrases</strong></p> <p>A set of verb phrases, annotated for the polarity of the verb phrase and the polarity of a polar noun that it contains. Can be used to evaluate whether a polarity classifier correctly recognizes polarity shifting. The file starts with 400 phrases containing shifter verbs, followed by 2231 phrases containing non-shifter verbs.</p> <ul> <li>File:&nbsp;<code>sentiment_phrases.txt</code></li> <li>Every item consists of: <ul> <li>The sentence from which the VP and the polar noun were extracted.</li> <li>The VP, polar noun and the verb heading the VP.</li> <li>Constituency parse for the VP.</li> <li>Gold labels for VP and polar noun by a human annotator.</li> <li>Predicted labels for VP and polar noun by RNTN tagger (Socher et al., 2013) and&nbsp;<code>LEX_gold</code>&nbsp;approach.</li> <li>Items are separated by a line of asterisks (*)</li> </ul> </li> </ul> <p><strong>Attribution</strong><br> This dataset was created as part of the following publication:</p> <p><a href="http://marc.schulder.info/">Schulder, Marc</a>&nbsp;and&nbsp;<a href="http://www.coli.uni-saarland.de/~miwieg/">Wiegand, Michael</a>&nbsp;and&nbsp;<a href="http://ruppenhofer.de/">Ruppenhofer, Josef</a>&nbsp;(2020).&nbsp;<strong>&quot;Automatic Generation of Lexica for Sentiment Polarity Shifters&quot;</strong>. In:&nbsp;<em>Natural Language Engineering</em>.&nbsp;<a href="https://doi.org/10.1017/S135132492000039X">doi:10.1017/S135132492000039X</a></p> <p>If you use the data in your research or work, please cite the publication.</p>