Representation Learning for Resource-Constrained Keyphrase Generation

TL;DR

This paper introduces a resource-efficient keyphrase generation method that leverages retrieval-based statistics and pre-trained language models, enabling effective low-resource and zero-shot domain adaptation.

Contribution

It proposes a novel data-oriented approach with salient span recovery and prediction objectives, improving keyphrase generation in low-resource settings.

Findings

Effective in low-resource keyphrase generation

Enhances zero-shot domain adaptation

Generates absent keyphrases close to large-data models

Abstract

State-of-the-art keyphrase generation methods generally depend on large annotated datasets, limiting their performance in domains with limited annotated data. To overcome this challenge, we design a data-oriented approach that first identifies salient information using retrieval-based corpus-level statistics, and then learns a task-specific intermediate representation based on a pre-trained language model using large-scale unlabeled documents. We introduce salient span recovery and salient span prediction as denoising training objectives that condense the intra-article and inter-article knowledge essential for keyphrase generation. Through experiments on multiple keyphrase generation benchmarks, we show the effectiveness of the proposed approach for facilitating low-resource keyphrase generation and zero-shot domain adaptation. Our method especially benefits the generation of absent keyphrases, approaching the performance of models trained with large training sets.