Collaborative Recommender System based on Improved Firefly Algorithm

Bharti Sharma, Adeel Hashmi, Charu Gupta, Amita Jain


A recommendation system aims to capture the taste of the customer and predict relevant items which he/she may be interested in buying. There are many algorithms for generating recommendations in literature, however, most of them are non-optimal and do not have the capability to handle big data. In this paper, a collaborative recommendation system is proposed based on improved firefly algorithm. The firefly algorithm is used to generate optimal clusters which provide effective recommendations. The proposed algorithm works in two phases: Phase I which generates the clusters with firefly algorithm and Phase II gives real time recommendations. The firefly algorithm has been implemented in Apache Spark to give it the capability of handling big data. The combination of improved firefly-based clustering and Apache Spark makes it much faster and optimal than the state-of-the-art recommendation models. For experiments, movie-lens dataset has been utilized and different evaluation metrics have been used for performance analysis. The results show that the proposed method gives better results compared to existing methods.


Clustering, collaborative filtering, firefly algorithm, recommender system, swarm intelligence

Full Text: PDF