Mining Technique Defined For Improving User-Based Recommendations in Diverse Environment (MTIURD)
Sangeetha G M1, Prasanna Kumar M2

1Sangeetha G M, Department of CSE, EWIT, VTU, Bangalore, India
2Prasanna Kumar, Assistant Professor, Department of CSE, EWIT, VTU, Bangalore, India.
Manuscript received on February 05, 2013. | Revised Manuscript received on February 27, 2013. | Manuscript published on March 05, 2013. | PP: 344-351 | Volume-3 Issue-1, March 2013. | Retrieval Number: A1396033113/2013©BEIESP
Open Access | Ethics and Policies | Cite
© The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Recommender systems are being extensively used in the present generation. Today’s consumer are facing with millions of goods and services when shopping online. Recommender systems help consumers by making product recommendations that are likely to be of interest to the user such as books, CDs, movies, restaurants, online news articles, and other services. Recommender systems are gradually increasingly harder to find the relevant contents of information in the vast abundant current age of information overload. Thus, recommender systems are needed to help individual users find the most relevant items or products or data sets from an abundant number of choices, collection. Through this gradually increase sales by exposing users to what they might like. E.g. In real time or real world applications consider a product say laptop, the laptop present in numerous patterns with different applications in number depending upon different user’s requirements. Thus providing a user or the customer with relevant information about the product as per their requirements with the help of recommender systems would ease the work of an user. Hence we can conclude saying that the volume of information available in the current age is huge to individual users (for e.g., e-commerce sites applications such as Amazon, Netflix) and hence focusing in developing some recommendation techniques within both industry and academia. Most, research to date is focusing on improving the recommendation accuracy i.e. the accuracy with which the recommender system predicts users ratings for items that are yet to be rated. The diversity of recommendation also plays an important role to be considered, it is important to explore the relationship between the accuracy and diversity and also the recommendation quality. Empirical analysis consistently shows the diversity gains of different recommendation techniques which is being used in several real world rating applications or datasets and uses different rating prediction algorithms. Individual users and online content providers will also benefit from the proposed approaches, where in which each user can find more relevant and personalized items or products from accurate and diverse recommendations provided by these recommender systems. These approaches, ranking techniques and algorithms could potentially lead to increased loyalty and sales in e-commerce application sites, thus benefiting the providers as well. Thus, serving these needs can result in greater success regarding cross-selling of related products, up selling, product affinities, and one-to –one promotions, larger baskets and customer retention.
Keywords: Recommender systems, recommendation accuracy, diverse recommendation, empirical analysis, ranking techniques, collaborative filtering, performance evaluation metrics, aggregate diversity, RMSE, extensions of recommendation approaches.