Matrix Factorization for Cloud Service Recommendation Based on Social Trust

Recommending trustworthy cloud services is essential to establishing credibility and ensuring better user decision-making based on their specific needs. Traditional recommendation approaches based on collaborative filtering face many challenges, including cold start problem, data sparsity and low accuracy and reliability of recommendations. We propose a social trust-based recommendation approach using matrix factorization which improves the recommendation accuracy and alleviates the problems of traditional recommendation systems. First, users' trust level is inferred from their interactions on social networks. Then, the social trust model is integrated with matrix factorization to generate reliable recommendations for users. Experiments conducted on two public datasets Epinions and WSdream demonstrate that social trust significantly improves recommendation accuracy in terms of error minimization compared to state-of-the-art recommender systems that do not take trust into account.