Replicability Analysis for Natural Language Processing: Testing Significance with Multiple Datasets

TL;DR

This paper introduces a statistically rigorous framework for analyzing the replicability of NLP algorithm performance across multiple datasets, addressing the limitations of traditional methods and demonstrating its effectiveness in various NLP tasks.

Contribution

It proposes a novel Replicability Analysis framework that improves statistical validity in multi-dataset NLP algorithm comparisons.

Findings

Framework provides more reliable significance testing across datasets.

Empirical validation across four diverse NLP applications.

Highlights limitations of current practices in NLP statistical analysis.

Abstract

With the ever-growing amounts of textual data from a large variety of languages, domains, and genres, it has become standard to evaluate NLP algorithms on multiple datasets in order to ensure consistent performance across heterogeneous setups. However, such multiple comparisons pose significant challenges to traditional statistical analysis methods in NLP and can lead to erroneous conclusions. In this paper, we propose a Replicability Analysis framework for a statistically sound analysis of multiple comparisons between algorithms for NLP tasks. We discuss the theoretical advantages of this framework over the current, statistically unjustified, practice in the NLP literature, and demonstrate its empirical value across four applications: multi-domain dependency parsing, multilingual POS tagging, cross-domain sentiment classification and word similarity prediction.