Aspect-based summarisation using distributed clustering and single-objective optimisation

In the user reviews of various domains, there is an increase in the accumulation of reviews in the web that presents a lot of difficulties to the readers. So it becomes necessary to generate a summary which represents the entire review in a concise manner. It is required for each feature or aspect i...

Full description

Saved in:
Bibliographic Details
Published in:Journal of information science 2020-04, Vol.46 (2), p.176-190
Main Authors: Priya, V, Umamaheswari, K
Format: Article
Language:English
Subjects:
Clustering
Data mining
Evolutionary algorithms
Genetic algorithms
Optimization
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In the user reviews of various domains, there is an increase in the accumulation of reviews in the web that presents a lot of difficulties to the readers. So it becomes necessary to generate a summary which represents the entire review in a concise manner. It is required for each feature or aspect in the reviews for the ease of users. The aspect-based summarisation plays a vital role in the field of opinion mining. This article proposes an aspect summarisation framework using sentence scoring clustering and weight-based single-objective optimisation technique by utilising evolutionary algorithm. The system uses MapReduce framework to incorporate the proposed combiner–based optimised clustering approach. Then a novel single-objective optimisation with genetic algorithm is developed. Its purpose is to retrieve top sentences from each cluster to generate feature-based summary. The accuracy of the system-generated summary is evaluated using the Recall Oriented Understanding for Gisting Evaluation tool kit using human standard reference summaries. The system is able to achieve more promising results when compared with other standard feature–based summarisation systems.
ISSN:0165-5515
1741-6485
DOI:10.1177/0165551519827896