As a new manufacturing paradigm, cloud manufacturing (CMfg) is proposed to realize the added-value and on-demand use of manufacturing resource and ability in the form of manufacturing services. Considering that there always exist correlations among cloud services (CS), which affect the cloud service composition (CSC). Hence, how to mine the correlations among CSs and apply them to CSC is a key issue for realizing the added-value. This paper presents a framework for correlation relationship mining for CSC. Four function modules for mining correlations among CSs are analyzed, and the involving key issues were preliminarily discussed as well.