Incorporating word embeddings in unsupervised morphological segmentation
Abstract© The Author(s), 2020. Published by Cambridge University Press. We investigate the usage of semantic information for morphological segmentation since words that are derived from each other will remain semantically related. We use mathematical models such as maximum likelihood estimate (MLE) and maximum a posteriori estimate (MAP) by incorporating semantic information obtained from dense word vector representations. Our approach does not require any annotated data which make it fully unsupervised and require only a small amount of raw data together with pretrained word embeddings for training purposes. The results show that using dense vector representations helps in morphological segmentation especially for low-resource languages. We present results for Turkish, English, and German. Our semantic MLE model outperforms other unsupervised models for Turkish language. Our proposed models could be also used for any other low-resource language with concatenative morphology.
CitationÜstün, A., & Can, B. (2020). Incorporating word embeddings in unsupervised morphological segmentation. Natural Language Engineering, 1-21. doi:10.1017/S1351324920000406
PublisherCambridge University Press (CUP)
JournalNatural Language Engineering
DescriptionThis is an accepted manuscript of an article published by Cambridge University Press in Natural Language Engineering on 10/07/2020, available online: https://doi.org/10.1017/S1351324920000406 The accepted version of the publication may differ from the final published version.
SponsorsThis research was supported by TUBITAK (The Scientific and Technological Research Council of Turkey) with grant number 115E464.
Except where otherwise noted, this item's license is described as https://creativecommons.org/licenses/by-nc-nd/4.0/