To help expedite the process of constructing use case diagrmas, a widely used notation in software engineering, we attempt to develop a generator system that can extract use case diagram automatically from the input of a software requirements specification. The use of natural language processing techniques can greatly assist this process, one of which is to use syntax-driven semantic analysis. Semantic analysis can provide output in the form of semantic representations that can be used to extract appropriate use case elements. A set of rules have been developed to extract information about the elements of use case diagrmas contained in the semantic representation. Our tests show that the system is able to automatically construct use case diagrams for a wide variety of linguistic variations. In a test using real-world cases, an average precision of 0.7375 and recall of 0.691 is obtained.