German Local Protest News (GLPN) dataset

<p>This dataset contains excerpts from newspaper articles of four German local newspapers labelled for relevancy in protest event analysis.</p> <p>It can be used to train machine learning models to detect news articles containing mentions of protest event for political analysis.</p> <p>For using a model trained on this data, it is recommended to preprocess new data in similar ways like this dataset.</p> <p>To retrieve the excerpts, we the following steps have been taken:</p> <ol> <li>split articles into sentences</li> <li>tag sentences that match the following regular expression: protest_regex = re.compile(r&#39;protest|versamm|demonstr|kundgebung|kampagne|soziale bewegung|hausbesetz|streik|unterschriftensammlung|hasskriminalit&auml;t|unruhen|aufruhr|aufstand|boykott|riot|aktivis|widerstand|mobilisierung|petition|b&uuml;rgerinitiative|b&uuml;rgerbegehren|aufmarsch&#39;, re.UNICODE | re.IGNORECASE)</li> <li>tag sentences predecessing or succeeding tagged sentences</li> <li>concatenate all tagged sentences to the excerpt.</li> </ol> <p>See the following code on github for an example:</p> <ul> <li><a href="https://github.com/Leibniz-HBI/protest-event-analysis/blob/main/utils.py">https://github.com/Leibniz-HBI/protest-event-analysis/blob/main/utils.py</a> contains the function &quot;reformat_df&quot; that preprocesses a column named &quot;text&quot; of a given dataframe</li> <li><a href="https://github.com/Leibniz-HBI/protest-event-analysis/blob/main/task-A_prediction.py">https://github.com/Leibniz-HBI/protest-event-analysis/blob/main/task-A_prediction.py</a> contains an example on how to apply a model on new data</li> </ul> <p>Experiments on this dataset are described in the following paper:</p> <p>&gt; Wiedemann, G., Dollbaum, J. M., Haunss, S., Daphi, P., Meier, L. D. (2022): A Generalized Approach to Protest Event Detection in German Local News, In: Proceedings of the 13th International Conference on Language Resources and Evaluation (LREC 2022). Marseille, France. European Language Resources Association (ELRA).</p> <p>In case of questions on the dataset, please contact Gregor Wiedemann at the Leibniz-Institute for Media Research (HBI): g.wiedemann@leibniz-hbi.de</p> <p>&nbsp;</p>