Skip to Main Content
A real recommender system can usually make use of more than one type of users' feedbacks, e.g., numerical ratings and binary ratings, in order to learn users' true preferences. In a recent work , a transfer learning algorithm called TCF is proposed to exploit such heterogeneous user feedbacks, which performs well via sharing data-independent knowledge and modeling data-dependent effect simultaneously. However, TCF is a batch algorithm and updates the model parameters only once after scanning the whole data, which may be not efficient enough for real systems. In this paper, we propose a novel and efficient transfer learning algorithm called iTCF (interaction-rich transfer by collective factorization), which extends the efficient CMF  algorithm with more interactions between the user-specific latent features. The assumption under iTCF is that the predictability w.r.t. the same user's rating behaviors in two related data is likely to be similar. Considering the shared predictability, we derive novel update rules for iTCF in a stochastic algorithmic framework. The advantages of iTCF include its efficiency comparing with TCF, and its higher prediction accuracy in comparison with CMF. Experimental results on three real-world data sets show the effectiveness of iTCF over the state-of-the-art methods.