This document summarizes a research paper on using clustering approaches to improve the discovery of semantic web services. It begins by defining semantic web services and semantic similarity measures. It then discusses using clustering to eliminate irrelevant services from a collection before applying semantic algorithms. Specifically, it proposes a clustering probabilistic semantic approach (CPLSA) that filters services based on compatibility with a query before clustering the remaining services into semantically related groups using probabilistic latent semantic analysis (PLSA). The document concludes by discussing applications of approximate semantics and challenges in scaling semantic algorithms.