Coreference Resolution for Swedish and German using Distant Supervision
Alexander Wallin, Pierre Nugues · DSpace repository (University of Tartu) · 2017
Coreference resolution is the identification of phrases that refer to the same entity in a text.Current techniques to solve coreferences use machine-learning algorithms, which require large annotated data sets.Such annotated resources are not available for most languages today.In this paper, we describe a method for solving coreferences for Swedish and German using distant supervision that does not use manually annotated texts.We generate a weakly labelled training set using parallel corpora, English-Swedish and English-German, where we solve the coreference for English using CoreNLP and transfer it to Swedish and German using word alignments.To carry this out, we identify mentions from dependency graphs in both target languages using hand-written rules.Finally, we evaluate the end-to-end results using the evaluation script from the CoNLL 2012 shared task for which we obtain a score of 34.98 for Swedish and 13.16 for German and, respectively, 46.73 and 36.98 using gold mentions.