Using Centroid Keywords and Word Mover's Distance for Single Document Extractive Summarization

Dauken Seitkali, Rustam Mussabayev

2019

Closed access Metadata (CSL-JSON)
DOI 10.1145/3342827.3342852
Data confidence 85 high
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

This paper presents unsupervised method of single document extractive summarization. The main idea behind the method is in selecting sentences based on Word Mover's Distance Similarity between each sentence and set of centroid keywords. This approach leverages both compositional property of word embeddings and advantages of recently discovered powerful text to text distance metric. ROUGE results on DUC 2002 data set showed that quality of produced summaries can compete with well-known state of the art systems. In this work we also discuss limitations of gold summaries in evaluating quality of summarization systems.