International Journal of Computer Science Trends and Technology (IJCST) – Volume 2 Issue 4, Jul-Aug 2014 ISSN: 2347-8578 www.ijcstjournal.org Page 1 RESEARCH ARTICLE OPEN ACCESS An Efficient Recommender System using Hierarchical Clustering Algorithm Prabhat Kumar 1 , Sherry Chalotra 2 Research Scholar 1&2 , Department of Computer Science Guru Nanak Dev Engineering College, Ludhiana Punjab-India ABSTRACT The massive growth of information these days has created the need for information filtering techniques that help users filter out extraneous content to identify the right information they need to make important decisions. The right information they need to make important decisions. Recommender systems are one approach to this problem, based on presenting potential items of interest to a user rather than requiring the user to go looking for them. Recommender system is a subclass of information retrieval system and information filtering system that seek to predict the 'rating' or 'preference' that user would give to an item. The concept of recommender system grows out of the idea of the information reuse and persistent preferences. Recommender systems have recently gained much attention as a new business intelligence tool for e-commerce business. Applying a recommender system for an online retailer store helps to enhance the quality of service for customers and increase the sale of products and services. In order to recommend items for particular requests the system has to perform large searching, sorting, and filtering and huge matrix operations. This will be a very time consuming operation even for smaller searching operations. Therefore there will be a need of an efficient framework to predict or recommend an item within time bounds. Almost all the recommenders proposed earlier uses continuous algorithms but the nature of the items is discrete and for computer systems the performance of discrete algorithms is much better as compared to continuous algorithms. In this paper a User-User based Collaborative Recommender system has been proposed that makes use of discrete cluster algorithms to enhance the recommendations and improve running time. Keywords: - Recommender System, Collaborative Filtering, Hierarchical Clustering, Jaccard Index. I. INTRODUCTION Servers are central to Businesses. In this information age every Business house owns or lease a server for their business logic. With the increase in Business and business logic the load on a server are increasing with enormous rate. Apart from that each server has to perform various other tasks as responding to client, managing enterprise level calculations [6], handling large databases, security and authentication. But since the domain of these tasks and there algorithms are well known and most of them are standardized thus the time complexity can be improved. This helps the system engineers to cope up with the load issues with the correct optimization and algorithms. But it has been observed that the recommender systems can increase the sales of a business house by 8-10%. Hence the race to incorporate the recommender system has begun. Now a day’s all the business houses are incorporating Recommender Systems to their servers for competitive advantage. But there exists no known algorithm for prediction or recommendation. Instead we use statistical data and then try to come to a conclusion based on certain hypothesis. For this the server has to collect, clean, load, store, parse and perform long computation to enormous set of data. The calculation includes various very large matrix operations and there is no known efficient algorithm for matrix operations. Hence the computation takes huge amount of time and space. To perform a matrix operation on data that currently amazon has is sufficient to overpower all the computers in the world working together. This is true even if we make computers out of every atom in this world. Hence there is a craving need of some efficient algorithms. There are various algorithm proposed in the literature most of them are trying to improve the algorithm using K-means algorithm because K-means algorithm [2] is easier to understand and it is very simple to implement. The simpler the algorithm is the better it will perform. It has been observed in many cases such as image compression technique JPEG 2000. But developing a recommender system is somewhat different from other software engineering tasks. K-means is continuous algorithm and works better if data set is continuous in nature. But this is certainly not in case of recommendation because the nature of items is discrete. And researcher tends to use Euclidean Distance [9] which is computationally very expensive since it requires both power and square root functions. And both of these functions require more than 200