In this paper, we propose a generic Context Aware Recommendation framework for M-commerce applications. The core of the framework is a Multidimensional Contextual graph which can effectively represent various real-world objects and their relations. Once the graph is created, a Random Walk algorithm is applied to obtain a top-k recommendation list. Both products and services can be flexibly modeled into the framework, in addition to their related information such as ratings and QoS measures. As part of our ongoing work, the framework is aimed at presenting a comprehensive recommendation framework for Mobile users.