This thesis describes the design and implementation of OKRA, a search engine for the U.S. Forest Services Encyclopedia of Southern Appalachian Oak Ecosystems. Chapter One provides background on the encyclopedia itself, and on the WordNet lexical database, which is used to .nd candidates for additional terms the user can add to his or her query. Chapter Two describes a module for interactively constructing and re.ning queries in the OKRA system. Chapter Three tells how connection strengths are determined between documents and individual query terms. Chapter Four describes how these connection strengths are combined to estimate the relevance of documents in response to a users query, including a novel system for scoring documents releative to Boolean queries. Chapter Five describes the results of tests of the system in actual use.