Domain Adaptation for Sequence Labeling Tasks with a Probabilistic Language Adaptation Model
Min Xiao, Yuhong Guo · 2013
In this paper, we propose to address the prob-lem of domain adaptation for sequence label-ing tasks via distributed representation learn-ing by using a log-bilinear language adapta-tion model. The proposed neural probabilis-tic language model simultaneously models two different but related data distributions in the source and target domains based on in-duced distributed representations, which en-code both generalizable and domain-specific latent features. We then use the learned dense real-valued representation as augment-ing features for natural language processing systems. We empirically evaluate the pro-posed learning technique on WSJ and MED-LINE domains with POS tagging systems, and on WSJ and Brown corpora with syn-tactic chunking and named entity recognition systems. Our primary results show that the proposed domain adaptation method outper-forms a number of comparison methods for cross domain sequence labeling tasks. 1.