Findings of the E2E NLG Challenge

TL;DR

This paper reports on the first shared task for end-to-end natural language generation in spoken dialogue systems, evaluating diverse approaches on a dataset with high lexical and syntactic complexity.

Contribution

It provides a comprehensive comparison of 62 systems from 17 institutions, assessing their ability to generate high-quality, diverse language output.

Findings

Seq2seq models perform competitively with rule-based systems.

Diverse approaches can handle complex discourse phenomena.

The shared task benchmarks current capabilities in end-to-end NLG.

Abstract

This paper summarises the experimental setup and results of the first shared task on end-to-end (E2E) natural language generation (NLG) in spoken dialogue systems. Recent end-to-end generation systems are promising since they reduce the need for data annotation. However, they are currently limited to small, delexicalised datasets. The E2E NLG shared task aims to assess whether these novel approaches can generate better-quality output by learning from a dataset containing higher lexical richness, syntactic complexity and diverse discourse phenomena. We compare 62 systems submitted by 17 institutions, covering a wide range of approaches, including machine learning architectures -- with the majority implementing sequence-to-sequence models (seq2seq) -- as well as systems based on grammatical rules and templates.