Variations of the Similarity Function of TextRank for Automated Summarization

TL;DR

This paper explores alternative similarity functions for the TextRank algorithm, aiming to improve automatic text summarization by proposing new variants that outperform the original method on standard datasets.

Contribution

The authors introduce novel similarity functions for TextRank, demonstrating significant improvements in summarization quality without changing the underlying algorithm.

Findings

Some variants outperform the original TextRank using standard metrics

Proposed functions achieve significant improvements on benchmark datasets

The study broadens the applicability of TextRank through alternative similarity measures

Abstract

This article presents new alternatives to the similarity function for the TextRank algorithm for automatic summarization of texts. We describe the generalities of the algorithm and the different functions we propose. Some of these variants achieve a significative improvement using the same metrics and dataset as the original publication.