Domain Adaptation of Thai Word Segmentation Models using Stacked Ensemble

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 3841–3847, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 3841 Domain Adaptation of Thai Word Segmentation Models using Stacked Ensemble Peerat Limkonchotiwat*, Wannaphong Phatthiyaphaibun§, Raheem Sarwar‡, Ekapol Chuangsuwanich†, Sarana Nutanong* *School of Information Science and Technology, VISTEC, Thailand §Faculty of Interdisciplinary Studies, Khon Kaen University, Thailand ‡RGCL, University of Wolverhampton, United Kingdom †Department of Computer Engineering, Chulalongkorn University, Thailand {Peerat.l s19,snutanon}@vistec.ac.th,wannaphong@kkumail.com, R.Sarwar4@wlv.ac.uk, ekapolc@cp.eng.chula.ac.th Abstract Like many Natural Language Processing tasks, Thai word segmentation is domain-dependent. Researchers have been relying on transfer learning to adapt an existing model to a new domain. However, this approach is inappli- cable to cases where we can interact with only input and output layers of the models, also known as “black boxes”. We propose a filter-and-refine solution based on the stacked- ensemble learning paradigm to address this black-box limitation. We conducted extensive experimental studies comparing our method against state-of-the-art models and transfer learning. Experimental results show that our proposed solution is an effective domain adap- tation method and has a similar performance as the transfer learning method. 1 Introduction Word Segmentation (WS) is an essential process for several Natural Language Processing (NLP) tasks such as Part-of-Speech (PoS) tagging and Machine Translation (MT). The accuracy of WS significantly affects the accuracy of these NLP tasks, as shown in experimental results from Nguyen et al. and Chang et al. While WS is considered relatively simple in English, it is still an open problem in languages without explicitly defined word delimiters, such as Thai, Chinese, and Japanese. However, unlike Chinese and Japanese, Thai WS did not receive much research attention. There are only six no- table publications (Chormai et al., 2019; Nararat- wong et al., 2018; Kongyoung et al.; Noyunsan et al.; Thanadechteemapat and Fung; Tongtep and Theeramunkong) on Thai WS for the past ten years. On the other hand, there are at least eight papers from well-established conferences on Chinese and Japanese WS (Li et al., 2019; Aguirre and Aguiar, 2019; Zhou et al.; Ma et al., 2018; Gong et al., 2017; Chen et al., 2017; Zhou et al., 2017; Cai et al., 2017) within only the last two years. This investigation focuses on the segmentation of Thai words since it is a challenging problem that has an excellent opportunity to improve, especially in the area of domain adaptation. Like many NLP tasks, Thai WS is domain- dependent. For instance, Chormai et al. (2019) recorded an accuracy drop from 91% to 81% when their model trained on a generic domain corpus (Ko- sawat et al., 2009) was tested on a social media one (bact’ et al., 2019). Results from our analysis (Section 3) also conform to these findings. One way to solve the domain dependency prob- lem is through Transfer Learning (TL), which is a common technique in domain adaptations (Schus- ter et al.; Chang et al.). However, TL may not be applicable when working with a commercial API or a model that does …

Content truncated to 500 words