Resource searching in the current peer-to-peer (P2P) applications is mainly based on the keyword match. However, more and more P2P applications require an efficient semantic searching based on the contents. In this paper, we propose a novel scalable semantic searching algorithm named SemSearch for the unstructured P2P networks. For the consistency and flexibility of semantic analysis, we integrate global and local semantic information to do the semantic analysis. Moreover, SemSearch transfers the searching requests to the peers whose shared resources are more semantic similar to implement semantic searching. We further evaluate the performance of SemSearch through the simulation experiments.