Locating desirable resources is very important for a large distributed system. However, the distributed, heterogeneous, and unstructured nature of the system makes this issue very challenging. The discovering mechanism has to be not only semantically rich, in order to cope with complex queries, but also scalable to handle large numbers of information sources. In this paper, we address these problems by proposing OntSum, an efficient peer-to-peer query routing scheme based on concise ontology indexing. Unlike most existing systems, our system does not assume a global ontology but heterogeneous ontologies. Peers in the system use their own ontologies to describe their resource knowledge and the network topology is adjusted according to peers? ontological properties. A novel indexing strategy enables forwarding queries only to semantically related nodes. The architecture improves interoperability among network participants and aids efficient resource discovery through an expressive query language.