Bridging Pre-trained Language Models and Hand-crafted Features for Unsupervised POS Tagging

TL;DR

This paper introduces a neural CRF autoencoder model that combines pre-trained contextualized word embeddings and hand-crafted features, significantly improving unsupervised POS tagging performance.

Contribution

It is the first to integrate PLMs and hand-crafted features in a neural CRF autoencoder for unsupervised POS tagging, surpassing previous SOTA results.

Findings

Outperforms previous models on Penn Treebank and Universal Dependencies datasets.

Effectively incorporates ELMo representations into a discriminative model.

Reintroduces hand-crafted features to enhance tagging accuracy.

Abstract

In recent years, large-scale pre-trained language models (PLMs) have made extraordinary progress in most NLP tasks. But, in the unsupervised POS tagging task, works utilizing PLMs are few and fail to achieve state-of-the-art (SOTA) performance. The recent SOTA performance is yielded by a Guassian HMM variant proposed by He et al. (2018). However, as a generative model, HMM makes very strong independence assumptions, making it very challenging to incorporate contexualized word representations from PLMs. In this work, we for the first time propose a neural conditional random field autoencoder (CRF-AE) model for unsupervised POS tagging. The discriminative encoder of CRF-AE can straightforwardly incorporate ELMo word representations. Moreover, inspired by feature-rich HMM, we reintroduce hand-crafted features into the decoder of CRF-AE. Finally, experiments clearly show that our model outperforms previous state-of-the-art models by a large margin on Penn Treebank and multilingual Universal Dependencies treebank v2.0.