Hierarchical Multi Task Learning with Subword Contextual Embeddings for Languages with Rich Morphology

TL;DR

This paper introduces a hierarchical multi-task learning approach using subword contextual embeddings to improve NLP tasks like dependency parsing and NER for morphologically rich languages, demonstrating significant performance gains.

Contribution

It proposes a novel hierarchical multi-task learning framework that incorporates subword contextual embeddings, a combination not previously explored, to enhance NLP task performance in morphologically complex languages.

Findings

Outperforms previous state-of-the-art on Turkish DEP and NER tasks.

Achieves 18.86% and 4.61% F-1 improvements over previous multi-task models.

Shows consistent performance improvements across five different MTL settings.

Abstract

Morphological information is important for many sequence labeling tasks in Natural Language Processing (NLP). Yet, existing approaches rely heavily on manual annotations or external software to capture this information. In this study, we propose using subword contextual embeddings to capture the morphological information for languages with rich morphology. In addition, we incorporate these embeddings in a hierarchical multi-task setting which is not employed before, to the best of our knowledge. Evaluated on Dependency Parsing (DEP) and Named Entity Recognition (NER) tasks, which are shown to benefit greatly from morphological information, our final model outperforms previous state-of-the-art models on both tasks for the Turkish language. Besides, we show a net improvement of 18.86% and 4.61% F-1 over the previously proposed multi-task learner in the same setting for the DEP and the NER tasks, respectively. Empirical results for five different MTL settings show that incorporating subword contextual embeddings brings significant improvements for both tasks. In addition, we observed that multi-task learning consistently improves the performance of the DEP component.