2020/05/27 by Katharina Kann, Arya D. McCarthy, Kann, Katharina +5
Computer Science · #Computation and Language (cs.CL) #FOS: Computer and information sciences #Natural Language Processing Techniques #Speech and dialogue systems #Topic Modeling
paper · pdf · doi:10.48550/arxiv.2005.13756
openalex publication_date 2020/05/27 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
In this paper, we describe the findings of the SIGMORPHON 2020 shared task on\nunsupervised morphological paradigm completion (SIGMORPHON 2020 Task 2), a\nnovel task in the field of inflectional morphology. Participants were asked to\nsubmit systems which take raw text and a list of lemmas as input, and output\nall inflected forms, i.e., the entire morphological paradigm, of each lemma. In\norder to simulate a realistic use case, we first released data for 5\ndevelopment languages. However, systems were officially evaluated on 9 surprise\nlanguages, which were only revealed a few days before the submission deadline.\nWe provided a modular baseline system, which is a pipeline of 4 components. 3\nteams submitted a total of 7 systems, but, surprisingly, none of the submitted\nsystems was able to improve over the baseline on average over all 9 test\nlanguages. Only on 3 languages did a submitted system obtain the best results.\nThis shows that unsupervised morphological paradigm completion is still largely\nunsolved. We present an analysis here, so that this shared task will ground\nfurther research on the topic.\n