Error Diversity Matters: An Error-Resistant Ensemble Method for Unsupervised Dependency Parsing

TL;DR

This paper introduces an error-diversity-aware ensemble method for unsupervised dependency parsing, improving robustness and accuracy by selectively combining diverse models and avoiding error accumulation.

Contribution

It proposes a novel ensemble-selection approach that considers error diversity, leading to better performance and robustness in unsupervised dependency parsing.

Findings

Outperforms individual models and previous ensemble methods

Enhances robustness against weak ensemble components

Significantly improves parsing accuracy

Abstract

We address unsupervised dependency parsing by building an ensemble of diverse existing models through post hoc aggregation of their output dependency parse structures. We observe that these ensembles often suffer from low robustness against weak ensemble components due to error accumulation. To tackle this problem, we propose an efficient ensemble-selection approach that considers error diversity and avoids error accumulation. Results demonstrate that our approach outperforms each individual model as well as previous ensemble techniques. Additionally, our experiments show that the proposed ensemble-selection method significantly enhances the performance and robustness of our ensemble, surpassing previously proposed strategies, which have not accounted for error diversity.