Document clustering for knowledge discovery using nature-inspired algorithm

As the internet is overload with information, various knowledge based systems are now equipped with data analytics features that facilitate knowledge discovery.This includes the utilization of optimization algorithms that mimics the behavior of insects or animals.This paper presents an experiment on document clustering utilizing the Gravitation Firefly algorithm (GFA).The advantage of GFA is that clustering can be performed without a pre-defined value of k clusters.GFA determines the center of clusters by identifying documents with high force.Upon identification of the centers, clusters are created based on cosine similarity measurement.Experimental results demonstrated that GFA utilizing a random positioning of documents outperforms existing clustering algorithm such as Particles Swarm Optimization (PSO) and K-means