GPU implementation of Friend recommendation system using CUDA for social networking services
K G Srinivasa , G M Siddesh , Srinidhi Hiriyannaiah , and 3 more authors
In Advances in Systems Analysis, Software Engineering, and High Performance Computing , 2016
Nowadays hybrid recommender systems are used, which utilize both collaborative and content based filtering techniques unlike the FoF system that have been presented in the chapter. Social networking services (SNSs) provide a platform where likeminded people interact and express opinions. The trends of socializing have changed drastically and the general population is turning to these services to socialize and network with new people. Massive infrastructure compliments uninterrupted usage of these services. Owing to the rapidly growing user base of SNSs, there is always a need to improve upon the existing infrastructure to keep cost and performance in tune with one another. GPUs are proving to be a viable solution to bridge the gap between the two. In this chapter, we describe GPU implementation of a Friend recommender system which is based on content-based filtering mechanism. It has given significant speed up from its previous counterparts, thus making the whole process more efficient.