Model and Data Transfer for Cross-Lingual Sequence Labelling in Zero-Resource Settings
This addresses the problem of applying supervised models to unlabelled languages for NLP researchers, offering insights into effective transfer methods, though it is incremental as it builds on existing techniques.
The paper tackled cross-lingual sequence labelling in zero-resource settings by comparing data-based and model-based transfer techniques, finding that high-capacity multilingual language models in a zero-shot setting consistently outperform data-based approaches, with results indicating up to competitive performance differences.
Zero-resource cross-lingual transfer approaches aim to apply supervised models from a source language to unlabelled target languages. In this paper we perform an in-depth study of the two main techniques employed so far for cross-lingual zero-resource sequence labelling, based either on data or model transfer. Although previous research has proposed translation and annotation projection (data-based cross-lingual transfer) as an effective technique for cross-lingual sequence labelling, in this paper we experimentally demonstrate that high capacity multilingual language models applied in a zero-shot (model-based cross-lingual transfer) setting consistently outperform data-based cross-lingual transfer approaches. A detailed analysis of our results suggests that this might be due to important differences in language use. More specifically, machine translation often generates a textual signal which is different to what the models are exposed to when using gold standard data, which affects both the fine-tuning and evaluation processes. Our results also indicate that data-based cross-lingual transfer approaches remain a competitive option when high-capacity multilingual language models are not available.