Dissertation Information for Laura Slaughter NAME: - Laura Slaughter
DEGREE:
- Ph.D.
DISCIPLINE:
- Library and Information Science
SCHOOL:
- University of Maryland (USA) (2002)
ADVISORS: - Dagobert Soergel - Marilyn Domas White
COMMITTEE MEMBERS: - Kent L. Norman - Thomas Craig Rindflesch - Keith Wilson Cogdill
MPACT Status: Fully Complete
Title: Semantic relationships in health consumer questions and physicians' answers: A basis for representing medical knowledge and for concept exploration interfaces
Abstract: This dissertation is concerned with the representation of medical knowledge in health consumers' questions and physician-provided answers. It has implications for the design of user interfaces that improve health consumers' ability to define their information problem and understand retrieved results. The underlying semantics of health consumers' questions and physicians' answers were analyzed as a basis for (1) understanding questioning behavior and answering patterns and (2) constructing a frame-based representation system for expressing semantic relationships in health-related questions and answers (or "potential" answers).
The primary research question involved the identification and analysis of semantic relationships in medically-oriented consumer questions and answers from Ask-an-Expert Web sites. Using the semantic relationships within the Unified Medical Language System (UMLS) Semantic Network as a starting point, this work used an iterative approach of analyzing and comparing semantic relationships in question-answer pairs as a basis for constructing an inventory of semantic relationships which evolved from instances in actual text rather than introspection. The results suggested important modifications to the UMLS Semantic Network, including clarifying the definitions and structure of the original UMLS relationships and introducing new semantic relationships.
A total of 509 semantic relationship instances were identified in twelve question-answer pairs (97 in the questions, 334 in the answers, 78 implied between question and answer concepts). Certain relationship categories were seen more often in questions than in answers and vice versa. Health consumers, for example, often use temporal semantic relationships in explaining their information need. A taxonomy of question types (Graesser's) was used to get a broad overview of the questions asked by consumers. The semantic relationships from questions were analyzed with respect to question type to appraise how the two might be used in conjunction with one another. Regarding the answering strategy of physicians, a direct match between the semantic relationships in the question and those in the answer occurred in only half of the question-answer pairs analyzed. All answers contained expansions of one or more question concepts. These concept expansions used many relationship types such as TREATS and RESULTS_IN.
|