Paper Title
A Trust-Based Recommendation Approach for Social Network Big Data

Abstract
The popularity of recommendation approach rapidly grows in social network to help the social network users for recommendation of popular big data resources such as subscribing celebrities, trending posts and videos, etc. As higher the social networks are, the higher of the trust in recommender systems becomes. Therefore, recommender system should focus on trust recommendation so as to secure important data or relations of social network users. This paper emphasizes on trust in recommending social resources to the users. In this case, we consider trust degree on big data based on optimized combination of users and similar user groups, and users and items groups. Each group considers recommended scores based on social trust degrees existed the relationships among similar user groups or among users and items groups. Trust degrees are deduced from trust models structured with graphs to understand the trust relations of big data resources. We also consider mixed weighting systems to tune the significant level between different pairs of comparison between user-user or between user-item. We then experiment our approach in trust-emphasized datasets to show the significance improvement of trust in social recommendation system. The results show that our recommendation system is superior to other proficient recommender systems in terms of accuracy rate and MEA (mean absolute error)evaluations. Index terms- recommender system, social network, big data, trust degree, graph model.