Training data for the shared task Ideology and Power Identification in Parliamentary Debates (2024)

<p>This dataset contains a selection of speeches from <a href="https://www.clarin.eu/parlamint">ParlaMint</a> corpora (version 4.0) as the training set for &nbsp;the shared task on "<a href="https://touche.webis.de/clef24/touche24-web/ideology-and-power-identification-in-parliamentary-debates.html">Ideology and Power Identification in Parliamentary Debates</a>" in <a href="https://clef2024.imag.fr/">CLEF 2024</a>.</p> <p>All files are tab-separated text files with the following fields:</p> <ul> <li>"<em>id</em>" is a unique (arbitrary) ID for each text.</li> <li>"<em>speaker</em>" is a unique (arbitrary) ID for each speaker. There may be multiple speeches from the same speaker.</li> <li>"<em>sex</em>" is the (binary/biological) sex of the speaker. This information is collected from varying sources (typically data published by the respective parliament), and in some cases it may be unspecified or unknown.</li> <li>"<em>text</em>" is the transcribed text of the parliamentary speech. Real examples may include line breaks, and other special sequences escaped or quoted.</li> <li>"<em>text_en</em>" is an automatic English translation of the corresponding text. This field may be empty (obviously) &nbsp;for speeches in English, but the translations may be missing for a small number of non-English speeches as well.</li> <li>"<em>label</em>" is the binary/numeric label. For political orientation, 0 is left and 1 is right. For power identification 0 indicates coalition (or governing party) and 1 indicates opposition.</li> </ul> <p>File names indicate the task and the parliament. We provide data from&nbsp;the following national and regional parliaments.</p> <ul> <li>Austria (at)</li> <li>Bosnia and Herzegovina (ba)</li> <li>Belgium (be)</li> <li>Bulgaria (bg)</li> <li>Czechia (cz)</li> <li>Denmark (dk)</li> <li>Estonia (ee) [only political orientation]</li> <li>Spain (es)</li> <li>Catalonia (es-ct)</li> <li>Galicia (es-ga)</li> <li>Basque Country (es-pv) [only power]</li> <li>Finland (fi)</li> <li>France (fr)</li> <li>Great Britain (gb)</li> <li>Greece (gr)</li> <li>Croatia (hr)</li> <li>Hungary (hu)</li> <li>Iceland (is) [only political orientation]</li> <li>Italy (it)</li> <li>Latvia (lv)</li> <li>The Netherlands (nl)</li> <li>Norway (no) [only political orientation]</li> <li>Poland (pl)</li> <li>Portugal (pt)</li> <li>Serbia (rs)</li> <li>Sweden (se) [only political orientation]</li> <li>Slovenia (si)</li> <li>Turkey (tr)</li> <li>Ukraine (ua)</li> </ul> <p>The number of training instances and the class imbalance differs for each training set. We do not provide a fixed validation split. Please see the <a href="https://touche.webis.de/clef24/touche24-web/ideology-and-power-identification-in-parliamentary-debates.html">shared task website</a> for further description of the data set and the sampling process.</p>