Analysis of a Novel Hybrid Fine Tuned Weighted Harmonic Mean for Efficient Plagiarism Detection using Particle Swarm

Abstract

Many types of resemblance across lexical and semantic levels are sometimes difficult for current plagiarism detection algorithms to detect. To overcome this drawback, this paper suggests a brand-new Weighted Harmonic Mean model that incorporates Hamming, Cosine, and Jaccard similarity scores. The suggested model makes use of the harmonic means' sensitivity to low scores to emphasize suspicious situations and accentuate small differences. Furthermore, Particle Swarm Optimization is suggested and presented as an effective way to tune weights and enhance performance. It is obtained that the proposed model performs better than the other stated methods.