AMALGUM -- A Free, Balanced, Multilayer English Web Corpus

TL;DR

AMALGUM is a large, freely accessible, genre-balanced English web corpus with multiple high-quality automatic annotations, designed to provide a more comprehensive and reliable resource than smaller, manually created datasets.

Contribution

It introduces a sizable, open web corpus with multiple annotation layers, addressing issues of imbalance, licensing, and quality in existing resources.

Findings

Achieves a 'better than NLP' benchmark through multi-layered annotations.

Provides high-quality automatic annotations including dependency trees and discourse structures.

Offers a balanced, large-scale corpus as an alternative to smaller datasets.

Abstract

We present a freely available, genre-balanced English web corpus totaling 4M tokens and featuring a large number of high-quality automatic annotation layers, including dependency trees, non-named entity annotations, coreference resolution, and discourse trees in Rhetorical Structure Theory. By tapping open online data sources the corpus is meant to offer a more sizable alternative to smaller manually created annotated data sets, while avoiding pitfalls such as imbalanced or unknown composition, licensing problems, and low-quality natural language processing. We harness knowledge from multiple annotation layers in order to achieve a "better than NLP" benchmark and evaluate the accuracy of the resulting resource.