Retrieval-efficiency trade-off of Unsupervised Keyword Extraction

TL;DR

This paper introduces a novel unsupervised keyword extraction method that merges document parts before graph construction, achieving state-of-the-art accuracy and significantly improved speed, scalable to large biomedical corpora.

Contribution

The paper proposes a new approach that merges document segments prior to graph creation and uses personalized PageRank for enhanced efficiency and performance in keyword extraction.

Findings

Achieves state-of-the-art retrieval performance.

Up to 100 times faster than existing methods.

Scalable to large biomedical datasets with 14 million documents.

Abstract

Efficiently identifying keyphrases that represent a given document is a challenging task. In the last years, plethora of keyword detection approaches were proposed. These approaches can be based on statistical (frequency-based) properties of e.g., tokens, specialized neural language models, or a graph-based structure derived from a given document. The graph-based methods can be computationally amongst the most efficient ones, while maintaining the retrieval performance. One of the main properties, common to graph-based methods, is their immediate conversion of token space into graphs, followed by subsequent processing. In this paper, we explore a novel unsupervised approach which merges parts of a document in sequential form, prior to construction of the token graph. Further, by leveraging personalized PageRank, which considers frequencies of such sub-phrases alongside token lengths during node ranking, we demonstrate state-of-the-art retrieval capabilities while being up to two orders of magnitude faster than current state-of-the-art unsupervised detectors such as YAKE and MultiPartiteRank. The proposed method's scalability was also demonstrated by computing keyphrases for a biomedical corpus comprised of 14 million documents in less than a minute.