Question Answering for Biomedicine

  • Author / Creator
    Liu, Yifeng
  • The field of biomedicine is reeling from “information overload”. Indeed, biomedical researchers find it almost impossible to stay current with published literature due to the vast amounts of data being generated and published. As a result, they are turning to text mining. Over the past two decades the field of biomedical text mining has experienced significant advances, such as the development of high quality biomedical knowledge bases and ontologies, the construction of biomedical search engines and the development of biomedical relationship mining tools. However, users still have to manually examine the retrieved documents and connect snippets of information from various databases to find answers to their queries. Ideally what is needed is a “wise” question answering (QA) system. With the advances in QA systems, including the triumph of IBM Watson on Jeopardy!, many biomedical researchers, including myself, believe that now is the time to further advance biomedical text mining by developing a biomedical question answering system. Such a system would be able to answer questions regarding biomedical entities and help researchers better digest existing knowledge and formulate new hypothesis. The task of biomedical question answering is faced with two central challenges: 1) retrieving relevant information from heterogeneous data sources (structured databases and free-text collections), and 2) formulating natural language answers from retrieved concepts and snippets. My research focuses on developing an association mining tool (PolySearch2) and a web-based biomedical question answering system (BioQA), that would provide precise answers with encyclopedia-like commentary to a wide range of biomedical questions. In particular, PolySearch2 mines concept associations from free-text collections based on co-occurrence statistics. BioQA uses PolySearch2 and other tools to decode natural language questions and formulate natural language answers for both descriptive and associative queries. Both PolySearch2 and BioQA offer public web interface to answer questions posed by biomedical researchers, physicians, students and the inquisitive public. PolySearch2 and BioQA represent an integrated solution to the core challenges in biomedical question answering.

  • Subjects / Keywords
  • Graduation date
    2016-06:Fall 2016
  • Type of Item
  • Degree
    Doctor of Philosophy
  • DOI
  • License
    This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.
  • Language
  • Institution
    University of Alberta
  • Degree level
  • Department
    • Department of Computing Science
  • Supervisor / co-supervisor and their department(s)
    • Wishart, David (Computing Science)
  • Examining committee members and their departments
    • Gallin, Warren (Biological Science)
    • Greiner, Russell (Computing Science)
    • Zaiane, Osmar (Computing Science)
    • Pavlidis, Paul (University of British Columbia)
    • Wishart, David (Computing Science)