Service recommendation is an essential task in service computing that helps users discover and select the most suitable services from a vast pool of available options. With the rapid growth of service-oriented architectures and cloud computing, the number of services has increased exponentially, making automated service selection crucial. While service descriptions and invocation networks provide valuable preliminary information for service recommendation, current research largely overlooks the anomalous behaviors in these networks, such as invoking dead services or measuring service similarity improperly. These anomalies introduce noise that can lead to lowquality embeddings. Such behaviors are difficult to eliminate and ultimately degrade recommendation performance. To address this challenge, we propose Noise Resistant Service Recommendation via LLM-Augmented Contrastive Learning (LLM-CoSR), a robust model that leverages Large Language Models (LLMs) for service recommendation. Our approach uses graph contrastive learning (GCL) to separate normal and anomalous services in the representation space while clustering normal services together. The invocation network variants in GCL are refined by LLMs through carefully designed prompts, enabling LLMCoSR to resist anomalous behaviors in the invocation network and generate more robust recommendations. Experimental evaluations on the ProgrammableWeb dataset demonstrate that LLMCoSR achieves superior performance compared to state-of-theart (SOTA) methods. The source code and parameter configurations are available at: https://github.com/catwinee/LLM-CoSR.

With the rapid development of new technologies such as cloud, edge and mobile computing, the number and diversity of available services are dramatically exploding and services have become increasingly important to people's daily work and life. As a consequence, using service label recommendation techniques to automatically categorize services plays a crucial role in many service computing tasks, such as service discovery, service composition, and service organization. There have been many service label recommendation studies that have achieved remarkable performance. However, these studies mainly focus on using the text information in service profiles to recommend labels for services while overlooking those social relations that widely exist among services. We argue that such social relations can help to obtain more precise recommendation results. In this paper, we propose a novel Social Relation aware Service Label Recommendation model called SRaSLR, which combines text information in service profiles and social network relations among services. A deep learning based model is constructed based on feature fusion of the two perspectives. We conduct extensive experiments on the real-world Programmable Web dataset, and the experiment results show that SRaSLR yields better performance than existing methods. Additionally, we discuss how service social network affects service label recommendation performance based on the experiment results.