As a new software design paradigm, microservices structure an application as a collection of services that are independently deployable and loosely coupled. A key step of migrating non‐microservices‐based systems to microservices‐based systems is the identification of microservices in the target application. Traditional approaches to identify microservices, however, usually suffer from lack of full automation and low effectiveness. This paper puts forward an expert system to identify microservices automatically from legacy systems by leveraging the similarity of RESTful APIs. The system consists of three major parts. The first part calculates the candidate topic similarity and the response message similarity of APIs, and the overall similarity is obtained through their combination. Afterwards, the second part constructs a graph of API similarities with API as the node and the overall similarity as the weight. The third part employs a graph‐based clustering algorithm to identify candidate microservices from the API similarity graph. Experiments conducted on open‐source projects demonstrate the effectiveness of our system.