In this age of information overload, providing useful recommendations to the users of E-commerce websites is an important and challenging problem. Personalized item ranking to recommend top-N items of interest to a user is more challenging in the case of implicit feedback data. Due to the lack of explicit user preferences, a model-based approach like matrix factorization is not effective for implicit user feedback. However, a neighborhood-based technique like Item-based collaborative filtering (IBCF) has shown better performance for implicit user feedback data and is a popular technique due to its scalability property. Although better than matrix factorization based technique for implicit feedback data, IBCF generates low-quality recommendations when data is sparse. In this work, we propose a method to address the problem of IBCF due to the sparseness of data by incorporating the metainformation related to items, and hybridization is done by forming a heterogeneous item information network. The proposed method uses a meta-path based framework for generating personalized item ranking and generates good quality recommendations in real time. The interest of users and popularity of items are leveraged simultaneously to improve the quality of recommendations. To find the intrinsic interest of users from the implicit feedback data, we perform the personalized weight learning to integrate the semantics of various meta-paths in the network. Experimental evaluation of the proposed method using the real-world data shows that the proposed method performs better than IBCF.