FRIEND RECOMMENDATION USING GRAPH MINING ON SOCIAL MEDIA

Abstract

Recommendation system is an important type of machine learning algorithm that provide precise suggestions to the users. Recommendation systems are used in innumerable types of areas such as generation of playlists, music and video services like Jio savaan, wynk, amazon prime music etc., and products recommendation for users in e-commerce applications and commercial applications. The recommendations that are provided by various types of applications increases the speed for identifying and makes easier to access the products that users are interested in. For each user, the recommendation system is capable of envisaging the future predilections on a set of items and recommend the top items. In several industries, recommendation systems are very useful as they generate huge amount of income and this type of industries can stand uniquely from competitors. Due to cumbersome number of items that each user can find in the web, the impact of recommendation system has been increased in the internet. Recommendation systems are used for custom-made navigation by getting huge amount of data particularly in social media domain for recommending friends. A recommendation system act as a subclass for the information filtering system that pursue to predict the rating. The similarity measures that are calculated in this research are Jaccard distance and Otsuka-Ochiai coefficient. The feature extractions that are used in this paper are Adar index, PageRank, Katz centrality, Hits score. Now a days many research people are implementing different types of algorithms in various domains for recommendation systems.