With the increase in the customer reviews, feedbacks, suggestions posted in the web forum, blogs led to the emergence of spam. Spam detection is important for both the customer and service providers to arrive at a proper decision while purchasing as well as marketing the product. Most of the research works has been developed only for sentiment classification for the past few decades which favors the spammers to write fake reviews. Hence it is important to detect the spam reviews but the major issues in spam review detection are the high dimensionality of feature space which contains redundant, noisy and irrelevant features. To resolve this, optimization method for selecting subset of features is necessary. Hence, this paper proposes Hybridization of Improved Binary Particle Swarm Optimization (iBPSO) and Binary Flower Pollination Algorithm (BFPA) utilized with Naive Bayes and k-NN for optimization process to improve the classification performance. Experimentation result proves that hybrid iBPSO BFPA outperformed the existing approach by obtaining the maximum accuracy of 94.43% for review spam dataset when compared with existing Cuckoo Search NB(CS) and Shuffled Frog Leaping Algorithm NB (SFLA) which achieved only 81.87% and 88.23%. The experimental result proves that the proposed hybrid method increases the classification accuracy
Digital Object Identifier (DOI)
Rajamohana, SP. and Umamaheswari, K.
"A Hybrid Approach to Optimize Feature Selection Process Using iBPSO- BFPA for Review Spam Detection,"
Applied Mathematics & Information Sciences: Vol. 11:
5, Article 22.
Available at: https://digitalcommons.aaru.edu.jo/amis/vol11/iss5/22