Exploiting News Article Structure for Automatic Corpus Generation of Entailment Datasets

TL;DR

This paper introduces a method to automatically generate NLI datasets from news articles for low-resource languages, creates the NewsPH-NLI dataset for Filipino, and evaluates transformer models to improve NLP performance in such languages.

Contribution

It presents a novel methodology for dataset creation from news articles, introduces the NewsPH-NLI dataset, and benchmarks new transformer models for Filipino NLP.

Findings

NewsPH-NLI is the first Filipino entailment dataset.

Pretrained transformers improve low-resource language NLP.

Transfer learning performance varies with data scarcity.

Abstract

Transformers represent the state-of-the-art in Natural Language Processing (NLP) in recent years, proving effective even in tasks done in low-resource languages. While pretrained transformers for these languages can be made, it is challenging to measure their true performance and capacity due to the lack of hard benchmark datasets, as well as the difficulty and cost of producing them. In this paper, we present three contributions: First, we propose a methodology for automatically producing Natural Language Inference (NLI) benchmark datasets for low-resource languages using published news articles. Through this, we create and release NewsPH-NLI, the first sentence entailment benchmark dataset in the low-resource Filipino language. Second, we produce new pretrained transformers based on the ELECTRA technique to further alleviate the resource scarcity in Filipino, benchmarking them on our dataset against other commonly-used transfer learning techniques. Lastly, we perform analyses on transfer learning techniques to shed light on their true performance when operating in low-data domains through the use of degradation tests.