Similar to product recommendation used in e-commerce sites and other fields, the recommendation of web service often take advantage of the history invocation information of users and services (denotes as QoS, Quality of service) to predict the unknown QoS value and then recommend web services to the active user with the best QoS. In order to adapt to the complex and changeful prediction occasions of QoS value of web service, this paper presents a hybrid collaborative filtering approach for the recommendation of web service by combining user-based collaborative filtering algorithm (UBCF) and item-based collaborative filtering algorithm(IBCF). This hybrid method consider the personalization of invocation information of users and services in the net while using Pearson Correlation Coefficient (PCC) to measure the similarity of two users or two services and adaptively balance the weigh of UBCF and IBCF while recommending services to the active user. Finally, though the experimental data provided by [12], we conduct a set of experiments and the results show that our proposed improved hybrid collaborative filtering algorithm had improved the accuracy of recommendation.