International Journal of Scientific & Engineering Research, Volume 7, Issue 5, May-2016 468 ISSN 2229-5518 IJSER © 2016 http://www.ijser.org Recommendation system and its approaches- A survey Naresh E, Geetha LM, Vijaya Kumar BP Abstract— Data size has been increasing day by day in the E commerce business. This rapid development of technology leads to overload of information. In order to get the required solution from this massive amount of data search engines are used. However this search engine doesn’t provide personalized information to the user. Further, recommendation systems are introduced in order to provide the personalized information to the user. Recommendation system provides suggestions based on the user’s interest. Many approaches are available for this purpose which can be used to create the recommendation list. E commerce websites uses these various approaches with different combinations in order to increase their business by attracting the users. This paper gives the overview of such approaches along with their strengths and weaknesses. Index Terms— Content-based approach, Collaborative approach, Data Mining, E-commerce business, Hybrid approach, Neural network , Recommendation list, Recommendation systems, Search engines. . ——————————  —————————— 1 INTRODUCTION ecommendation system was mostly used in the online shopping to recommend the list of product. Nowadays it is also gaining importance in different domains like news, portal and service providing web pages and so on. Recommendation system helps the user to choose the right information based on their preferences. This recommenda- tion system builds a model using set of data and then this model is used to predict the result for recommendation [13]. As the technology advanced data generation is also increas- ing rapidly. Earlier, search engines were the solution to find the required data. But in the E-Commerce business, this massive generation of data leading to find the approaches that helps the customer to choose the items which are of their interest. Recommendation system is used for this pur- pose. Recommender system takes the input from input and gives the results or suggests services/product which is rel- evant to the customer by using individual or particular group details [11][15]. As the web technology developed, recommender system gained its importance in the E-commerce business. It has also become a research area on developing new approaches for the recommendation system. In order to develop the new approaches and to make the system more efficient, we need to understand the existing approaches and its limita- tions. ———————————————— • Naresh E is working as Assistant professor, Dept of ISE, MSRIT and research scholar in CSE dept at Jain University Bangalore, India. E-mail:nareshkumar.e@gmail.com • Geetha LM is currently pursuing masters degree program in software engineering in MSRIT, Bangalore, India,. E-mail: geetha.lm90@gmail.com • Viajaya Kumar BP is currently professor and head, Dept of ISE, MSRIT, Bangalore. And Senior member in IEEE. As per various survey conducted, many approaches are available for recommendation engine. These are mainly classified into following categories: i. Content-based approach: Recommendation is given based on the user’s preferred items in the past; ii. Collaborative approach: Items are recom- mended based on the interest of similar previ- ous people; iii. Hybrid approach: This is the combined ap- proach of Content-based and Collaborative recommendations. Other than these approaches many other recommendation class techniques are available. They are feature based, be- havior based, citation based, context based, knowledge based, rule based and many other recommendation classes [2]. It is necessary to understand these concepts in order to extend the capabilities of recommendation engine [1]. Stud- ies have revealed that 55% of the recommendation system uses content based filtering. 18% of recommendation sys- tem uses collaborative filtering and graph based recom- mendation approach is used by 16% of the system. Though many approaches are available for recommenda- tion, many studies have revealed that it is hard to say which concept or approach is best. Sometime Content- based approach works better compared to collaborative recommendation approach and sometime it doesn’t. Hence it is difficult to choose the promising approaches in the available techniques. Sometimes it requires the combina- tions of available approaches. Hence, basic understandings R IJSER