Ontology-based information retrieval system framework to support oncology drug development planning and regulatory research

A plethora of data on approved drug products is being generated daily and made available by the Food and Drug Administration. Identifying opportunities and gaps in this existing array of information early on can support and expedite drug development planning and regulatory research activities. However, the highly compartmental and unstructured nature of the drug approval packages creates barriers in finding and exploiting such information. Moreover, current techniques used to retrieve information from drug approval packages are largely based on key-word syntax matching. A setback with key-word based information retrieval techniques is that they do not consider domain knowledge, meaning of words, and semantic relationships between concepts, which leads to irrelevant and poor quality search results. This dissertation presents a framework for an ontology-based information retrieval system to retrieve documents relevant to the user's information need, from published drug approval packages.;Two critical challenges were identified while developing the system framework: 1) Creation of a knowledge base for the pharmaceutical regulatory affairs domain to conceptualize and formalize FDA's drug approval process 2) Enabling semantic understanding of user queries to capture and support the information need in a way intuitive to the user.;To address these challenges, the following developments were undertaken: 1) Ontology for Drug Development and Regulatory Research is the first regulatory ontology that captures and structures knowledge of FDA's drug review and approval process 2) Development of a semantic Ontology Information Retrieval module which was composed of: a) Definition of query pattern identifying rules to enable identification of patterns in queries b) Ontology Matching Algorithm c) Pattern Matching Algorithm d) Hybrid String Comparison Algorithm e) Ranking Algorithm. It was the premise of this dissertation that the developed system framework would provide the following benefits: 1) Definition of a common vocabulary to facilitate sharing, adaptation, and extension of information by different applications built for the regulatory affairs domain 2) Improved precision in search results over key-word syntax matching 3) "Richer" user experience in constructing queries. The results from the developed system framework provide evidence of improvement in inference, retrieval, and accuracy of search results; and natural language query processing capability.