Semantic search techniques have increasingly gained attention in information retrieval literature. Authors are great sources of semantic interpretation for documents, especially in scholarly domains where articles mostly reflect the research interests of the authors. Being able to interpret semantic meanings of documents from their authors would give rise to many interesting applications, especially in academic digital library literature. In this paper, we present taxonomy-based query-dependent schemes for computing author profile similarity. Our schemes have the capability to capture partial similarities, as opposed to traditional topic overlap based approaches. We generalize our schemes so that they can be easily adopted to other application domains. We acquire resources from multiple places such as Wikipedia, CiteseerX, ArnetMiner, and WikipediaMiner as part of our work. We provide encouraging anecdotal results along with suggestions on potential applications of the proposed schemes.