VOL. 10, NO. 10, JUNE 2015 ISSN 1819-6608 ARPN Journal of Engineering and Applied Sciences ©2006-2015 Asian Research Publishing Network (ARPN). All rights reserved. www.arpnjournals.com 4683 SERVICE RECOMMENDATION SYSTEM IN SOCIAL NETWORKS Sandra Elizabeth Salim and R. Jebakumar Department of Computer Science and Engineering, SRM University, Chennai, India E-Mail: sandraelizabeth.05@hotmail.com ABSTRACT Social networks have become an inevitable part of today’s life. The content from social media can be used for a number of purposes; one of the main being recommendation systems (RS). Traditional recommendation systems ignored the concept of social media and its influence on people. There have been a lot of RS in industry since the last decade. In this paper, we have proposed a new system called Service Recommendation in Social Networks (SRSN) based on a keyword approach which overcomes the drawbacks of current generation of RS. It utilizes the concept of user based collaborative filtering algorithm (UCF) in generating recommendations. SRSN is designed to work in big data environments as it is implemented in Hadoop, a map reduce paradigm. Keywords: recommendation system, big data, social network, collaborative filtering. 1. INTRODUCTION Social networks have provided a new platform for people to interact and share information. Social networks are now the best medium to create public awareness and spread useful information to the people. Nowadays, numerous sources deliver information in different forms and this data is mostly produced from social network site users through their reviews and comments [1]. People find this type of content more useful than those produced by professional writers. Before the era of internet, it was difficult for people to take decisions regarding what service to go for by evaluating the pros and cons of these services. The only option was to get opinions and suggestions from their friends and relatives. With the advent of internet and especially social media, this has become an easy task. People can easily get any amount of information about a particular service through internet and social media [2]. In some cases, users find it difficult to exactly judge what service to go for or may be what to buy; they may not get the exact result what they are looking for. Some people resort to get help from multiple blogs, news articles or events which are related to the item they are searching for. But still in this case, users need to visit various sources and go through them to differentiate between the content required by them and content irrelevant to them. In such cases, the best way is to provide recommendations to the users regarding their exact requirement. Users feel more comfortable and satisfied to get recommendations from their friends, relatives and other known people than getting from strangers and people unfamiliar to them. This is accomplished through social networks [3]. There are a number of social networking sites such as Facebook, Twitter, Quora, LinkedIn, etc. Through these networks, we get a system where information is integrated from multiple sources to provide content and data to users sharing common interests and ideas. Users can give feedbacks and suggestions about various services which are useful for all other members in that particular social media or communities within the network. However, the current generation of RS have many drawbacks and requires further improvements to make recommendation methods more effective. These improvements include better methods for representing user characteristics and the information about the items to be recommended, incorporation of various user given contextual information into the recommendation process, utilization of multicriteria ratings, development of less intrusive and more efficient recommendation methods for the big data environment. In this paper, we propose a new technique which overcomes the shortcomings of existing recommendation systems in social networks. In Section 2 and 3, we provide a small introduction about social media and big data along with the current recommendation systems. In Section 4, we propose a new system, SRSN which uses a keyword based approach for generating recommendations. 2. SOCIAL MEDIA Social networks have provided an advanced method for people to communicate with each other, that it has become a part and parcel of today’s modern era [4]. Nowadays, there are hundreds of sources delivering content in various forms and more content is generated from users through social networks and the reviews and comments in these networks than through professional writers. Users are often faced with information overflow. In the days before Internet, people found it hard to determine what book to read, what movie to watch, or which place to visit, etc. They were often guided with the recommendations and feedbacks of their friends. Social media enhances social cooperation and communication among people where they exchange information and share ideas [5]. Social media has become an inevitable aspect of today’s life. We hardly find anyone who is not aware or who is not a part of any social network. Social media plays an important role in bridging the communication gap between organizations and individuals. Social media utilizes mobile and web-based features to form advanced interactive environments through which people communicate, share and create their