With the prevalence of Social Networks (SNs) and services, plenty of trust models for Trustworthy Service Recommendation (TSR) in Service-oriented SNs (S-SNs) have been proposed. The reputation-based schemes usually do not contain user preferences and are vulnerable to unfair rating attacks. Meanwhile, the local trust-based schemes generally have low reliability or even fail to work when the trust path is too long or does not exist. Thus it is beneficial to integrate them for TSR in S-SNs. This work improves the state-of-the-art Combining Global and Local Trust (CGLT) scheme and proposes a novel Integrating Reputation and Local Trust (IRLT) model which mainly includes four modules, namely Service Recommendation Interface (SRI) module, Local Trust-based Trust Evaluation (LTTE) module, Reputation-based Trust Evaluation (RTE) module and Aggregation Trust Evaluation (ATE) module. Besides, a synthetic S-SN based on the famous Advogato dataset is deployed and the well-known Discount Cumulative Gain (DCG) metric is employed to measure the service recommendation performance of our IRLT model with comparing to that of the excellent CGLT model. The results illustrate that our IRLT model is slightly superior to the CGLT model in honest environment and significantly outperforms the CGLT model in terms of the robustness against unfair rating attacks.