Simple is not Enough: Document-level Text Simplification using Readability and Coherence

TL;DR

This paper introduces SimDoc, a document-level text simplification system that jointly optimizes for simplicity, readability, and coherence, addressing a gap in existing sentence-focused approaches.

Contribution

It extends professional simplification corpora with new annotations and evaluates a multi-objective training approach for document-level simplification.

Findings

Effective in zero-shot, few-shot, and fine-tuning settings

Improves coherence and readability in document simplification

Highlights challenges of document-level simplification

Abstract

In this paper, we present the SimDoc system, a simplification model considering simplicity, readability, and discourse aspects, such as coherence. In the past decade, the progress of the Text Simplification (TS) field has been mostly shown at a sentence level, rather than considering paragraphs or documents, a setting from which most TS audiences would benefit. We propose a simplification system that is initially fine-tuned with professionally created corpora. Further, we include multiple objectives during training, considering simplicity, readability, and coherence altogether. Our contributions include the extension of professionally annotated simplification corpora by the association of existing annotations into (complex text, simple text, readability label) triples to benefit from readability during training. Also, we present a comparative analysis in which we evaluate our proposed models in a zero-shot, few-shot, and fine-tuning setting using document-level TS corpora, demonstrating novel methods for simplification. Finally, we show a detailed analysis of outputs, highlighting the difficulties of simplification at a document level.