Enhanced continuous and discrete multi objective particle swarm optimization for text summarization

Abstract

Reviews from various domains is being posted in web increasingly day by day. Analyzing this enormous content would be useful in decision making for various stakeholders. Text summarization techniques generate concise summaries including sentiments which are useful in analyzing the large content. So text summarization systems become significant in analyzing this huge content. The summaries are generated based on important features using multi objective approaches where sufficient literature is not available. Major limitations of text summarization systems are scalability and performance. Two variants of multi objective optimization techniques such as Discrete and Continuous which work under the principles of particle swarm optimization (PSO) for extractive summarization of reviews had been proposed for performance improvement. The performance is validated using Recall-Oriented Understanding for Gisting Evaluation (ROUGE), Success Counting (SC) and Inverted Generational Distance (IGD). Based on the experimental results it is found that the system is effective using multi-objective PSO algorithm when compared to other state-of-art approaches like Liu’s approach feature based binary particle swarm optimization and etc. for feature based review summarization.