The inundation of biological data has produced enormous amounts of information in the form of research articles. The repositories for these articles have become so abundant and so excessively populated that searching for a timely and relevant article has become near impossible. BioKnOT attempts to address this issue and allows for efficient and effective information retrieval. This system implements an iterative refinement of search building upon semantic relevance, with consideration to citation frequencies. It does this by constructing ontologies from term relationships based on words determined by existing term-frequency-inversedocument- frequency (TFIDF) strategies, while including a means of comparing ontologies using scoring matrices that consider pairs of words in and among sentences. BioKnOT will address the demand for sifting through the copious amounts of present and an ever increasing number of biologically related research articles. In this paper, we discuss the theory and intuitions underlying our system and its implementation.