Indian Journal of Science and Technology

Abstract

Objectives: The core intent of this paper is to propose a dynamic recommender system(i) to endow with an enhanced understanding of the behaviors and interest of online users and (ii) to tumble information overload by providing guidance to the online users. Methods: Foraging behavior based swarm intelligence is motivated from the dynamic social behavior of swarms. It puts forward innate option for modeling dynamic online usage data. In the preset work, we mainly concentrate on the innate resemblance between Swarm Intelligence and communal behavior. The proposed WebFrieseBee algorithm is inspired from the foraging behavior of the T. biroiFriese bees and the proposed the proposed BestProphecy-annealing algorithm is used for providing recommendations to users. Findings: The proposed WebFrieseBee is motivated from the dynamic social behavior of swarms. It offers collaborative learning and has decentralized control. Moreover, it also has high exploration ability. That is, it put forward innate option for modeling dynamic online usage data. The proposed BestProphecy-annealing algorithm uses a greedy heuristic approach, for providing recommendations to users by identifying a better neighborhood for agents and gives recommendations based on the preferences of these best neighborhoods. The proposed WebFrieseBee may overcome the data redundancies existing in the repeated use of information which are inappropriate and may provide tradeoff between coverage and precision. Improvements: Our proposed dynamic recommender system surmount the grey sheep problem and new user ramp up problem in many traditional recommender systems. Our proposed dynamic recommender system was compared with Ant clustering approach. The Experimental results shows that our approach offers better quality in terms of coverage, precision and F1 Measure than the traditional Ant clustering approaches. Applications: The recommender system is very effective in increasing the utility of e-commerce by minimizing the user surfing time and overload in servers.
Keywords: Recommender Systems, Software Agents, Stimulated Annealing, Swarm Intelligence, User Profiles, Web Usage Mining