As Service Oriented Architecture (SOA) matures, service consumption demand leads to an urgent requirement to service discovery. Unlike web documents, services are intended to be executed to achieve objectives and/or desired goals of users, which means to realize application requirements. This leads to the notion that service discovery should take into account the "application requirement" of service with service content (descriptions) which have been well explored. Content is defined by service developers, e.g. WSDL file and context is defined by service users, which is service usages to application requirement. We find context(application) information is more useful for query generation, especially for non-expert users. So in this paper, we propose to do context-sensitive query processing to resolve application-oriented queries for web service search engine. Context is modeled by a bipartite graph model to represent the mapping relationship between application space and service space. Application-oriented queries are resolved by query expansion based on the topic sensitive bipartite graph. The experiments verify the efficiency of our idea.