2787612
doi
10.5281/zenodo.2787612
oai:zenodo.org:2787612
user-eu
Stefano Menini
FBK
Marco Baroni
ICREA
Luisa Bentivogli
FBK
Raffaella Bernardi
Università di Trento
Roberto Zamparelli
Università di Trento
The SICK (Sentences Involving Compositional Knowledge) dataset for relatedness and entailment
Marco Marelli
Università di Milano Bicocca
info:eu-repo/semantics/openAccess
Creative Commons Attribution Non Commercial Share Alike 3.0 Unported
https://creativecommons.org/licenses/by-nc-sa/3.0/legalcode
computational linguistics, entailment, sentence similarity, sentence relatedness, compositional semantics, distributional semantics
<p>The SICK data set consists of about 10,000 English sentence pairs, generated starting from two existing sets: the <a href="http://nlp.cs.illinois.edu/HockenmaierGroup/data.html">8K ImageFlickr data set</a> and the <a href="http://www.cs.york.ac.uk/semeval-2012/task6/index.php?id=data">SemEval 2012 STS MSR-Video Description data set</a>. We randomly selected a subset of sentence pairs from each of these sources and we applied a 3-step generation process: first, the original sentences were normalized to remove unwanted linguistic phenomena; the normalized sentences were then expanded to obtain up to three new sentences with specific characteristics suitable to CDSM evaluation; as a last step, all the sentences generated in the expansion phase were paired with the normalized sentences in order to obtain the final data set.</p>
<p>Each sentence pair was annotated for relatedness and entailment by means of crowdsourcing techniques. The <strong>sentence relatedness score</strong> (on a 5-point rating scale) provides a direct way to evaluate CDSMs, insofar as their outputs are meant to quantify the degree of semantic relatedness between sentences; the categorizations in terms of the <strong>entailment relation between the two sentences</strong> (with <em>entailment, contradiction</em>, and <em>neutral</em> as gold labels) is also a crucial aspect to consider, since detecting the presence of entailment is one of the traditional benchmarks of a successful semantic system.</p>
<p>In the final set, gold scores for relatedness and entailment were distributed as follows: the relatednes scoring resulted in 923 pairs within the [1,2) range, 1373 pairs within the [2,3) range, 3872 pairs within the [3,4) range, and 3672 pairs within the [4,5] range; the entailment annotation led to 5595 <em>neutral</em> pairs, 1424 <em>contradiction</em> pairs, and 2821 <em>entailment</em> pairs.</p>
<p><strong>Files</strong></p>
<ul>
<li>SICK.zip (main file)</li>
<li>SICK_Annotated.zip (a version of the data set annotated for the expansion rule which was used in each case)</li>
<li>SICK_subsets.zip (a Indexes specifying further classifications, used in the JLRE 2016 publication)</li>
</ul>
<p> </p>
Zenodo
2014-05-26
info:eu-repo/semantics/other
2787611
user-eu
award_title=Compositional Operations in Semantic Space; award_number=283554; award_identifiers_scheme=url; award_identifiers_identifier=https://cordis.europa.eu/projects/283554; funder_id=00k4n6c32; funder_name=European Commission;
1579893909.751399
217584
md5:b03210036cc64a4cf1bc7a0525357001
https://zenodo.org/records/2787612/files/SICK.zip
246005
md5:1c0e709f59f92e4cfa4b0a613027ac1b
https://zenodo.org/records/2787612/files/SICK_Annotated.zip
16083
md5:eeef24604e01a05e76406b290e241359
https://zenodo.org/records/2787612/files/SICK_subsets.zip
public
10.5281/zenodo.2787611
isVersionOf
doi