JOWTS: A differential evolution based approach for Microblog Summarization with advanced population enhancement techniques

Abstract

Social networking services have emerged as the main sources for real-time information about events happening. It has been observed that pertinent information gleaned from tweets during catastrophic events can be helpful in a variety of ways. Therefore, it is necessary to create an automated microblog summarization system. The proposed approach JOWTS, confluence of a wide range of evolutionary computation techniques such as the well-known differential evolutionary algorithm JADE (DE/current-to-pbest/1), Opposition-based Learning (OBL) and Whale Optimization Algorithm (WOA), employs multi-objective optimization for microblog summarization. The summarization task is formulated as a multi-objective optimization problem and combination of objectives such as tweet length & importance of tweets (through tf-idf technique) in a dataset are optimized at the same time. For evaluation, datasets relevant to disaster events are employed and the results are compared to different alternative methodologies utilizing ROUGE measures. When compared against the contemporary evolutionary techniques, it was observed that JOWTS improves ROUGE-1, 2, L scores by 3.86%, 8.53% and 4.69% respectively.