X-SRL: Parallel Cross-lingual Semantic Role Labeling

<h3>Introduction</h3></br>
<p>X-SRL: Parallel Cross-lingual Semantic Role Labeling Department of Computational Linguistics and the <a href="https://www.leibniz-gemeinschaft.de/en/institutes/leibniz-institutes-all-lists/leibniz-institute-for-the-german-language.html">Leibniz Institute for the German Language (IDS)</a>. It consists of approximately three million words of German, French and Spanish annotated for semantic role labeling. The texts are translations of the English portion of <a href="../../../LDC2012T04">2009 CoNLL Shared Task Part 2 (LDC2012T04)</a>. All sentences have annotations for verbal predicates and share the original English <a href="../../../LDC2004T14">Propbank</a> label set across the four languages.</p></br>
<h3>Data</h3></br>
<p>The 2009 CoNLL Shared Task developed syntactic dependency annotations, including the semantic dependency model roles of both verbal and nominal predicates. The following English data used in the shared task:</p></br>
<ul></br>
<li><a href="../../../LDC95T7">Treebank-2 (LDC95T7)</a>: over one million words of annotated English newswire and other text developed by the University of Pennsylvania</li></br>
<li><a href="../../../LDC2012T04">Proposition Bank I (LDC2004T14)</a>: semantic annotation of newswire text from Treebank-2 developed by the University of Pennsylvania</li></br>
<li><a href="../../../LDC2008T23">NomBank v 1.0 (LDC2008T23)</a>: argument structure for instances of common nouns in Treebank-2 and <a href="../../../LDC99T42">Treebank-3 (LDC99T42)</a>, texts developed by New York University</li></br>
</ul></br>
<p>For X-SRL, the English source data was automatically translated using <a href="https://www.deepl.com/en/translator">DeepL</a>. Automatic tokenization, lemmatization, part-of-speech tagging and syntactic parsing were than applied to the text. The data was divided into train, development and test partitions. Semantic labels were transfered for the train and development sections, and the test sentences were validated for translation quality, alignment, label transfer, and filtering.</p></br>
<p>More information on the development process and tools used is available in the included documentation.</p></br>
<p>Annotated data is in the <a href="https://universaldependencies.org/format.html">Universal CoNLL format</a> and encoded in UTF-8.</p></br>
<h3>Sponsorship</h3></br>
<p>The creation of this corpus was funded by the Leibniz ScienceCampus "Empirical Linguistics and Computational Language Modeling" supported by Leibniz Association (grant no. SAS2015-IDS-LWC) and by the Ministry of Science, Research, and Art of Baden-Wurttemberg.</p></br>
<h3>Samples</h3></br>
<p>Please view this <a href="desc/addenda/LDC2021T09.txt">text sample (TXT)</a> and <a href="desc/addenda/LDC2021T09.conll.txt">annotation sample (TXT)</a>.</p></br>
<h3>Updates</h3></br>
<p>None at this time.</p>
Portions © 1987-1989 Dow Jones & Company, Inc., © 2021 Department of Computational Linguistics, Heidelberg University, © 2021 Leibniz Institute for the German Language (IDS), © 1993-1995, 1999, 2004, 2008, 2012, 2021 Trustees of the University of Pennsylvania