Generating Diverse Descriptions from Semantic Graphs

TL;DR

This paper introduces a stochastic graph-to-text model with an ensemble approach to generate diverse, high-quality descriptions from semantic graphs, along with a new metric to evaluate diversity and quality.

Contribution

It presents a novel stochastic model with an ensemble for diverse text generation from semantic graphs and a new automatic evaluation metric for diversity and quality.

Findings

Ensemble of stochastic models produces diverse outputs.

Generated descriptions maintain quality comparable to state-of-the-art.

Model evaluated on WebNLG datasets in English and Russian.

Abstract

Text generation from semantic graphs is traditionally performed with deterministic methods, which generate a unique description given an input graph. However, the generation problem admits a range of acceptable textual outputs, exhibiting lexical, syntactic and semantic variation. To address this disconnect, we present two main contributions. First, we propose a stochastic graph-to-text model, incorporating a latent variable in an encoder-decoder model, and its use in an ensemble. Second, to assess the diversity of the generated sentences, we propose a new automatic evaluation metric which jointly evaluates output diversity and quality in a multi-reference setting. We evaluate the models on WebNLG datasets in English and Russian, and show an ensemble of stochastic models produces diverse sets of generated sentences, while retaining similar quality to state-of-the-art models.