Open Access
Wang, Hongmei
Graduate Program:
Information Sciences and Technology
Doctor of Philosophy
Document Type:
Date of Defense:
November 16, 2006
Committee Members:
  • Guoray Cai, Committee Chair
  • Alan Maceachren, Committee Member
  • Michael Mc Neese, Committee Member
  • John Yen, Committee Member
  • natural spoken language enabled GIS
  • human-GIS communication
  • collaborative dialogue approach
  • agent-based computational model
  • vague spatial concept
  • and context-dependent
Natural multimodal interfaces have been proposed as an alternative interface for geospatial information systems. A fundamental challenge in developing a usable conversational interface for GIS is effective communication of spatial concepts in natural language, which are commonly vague in meaning. This study recognizes and makes distinctions between two sources of vagueness in human-GIS communication: (1) there are multiple contexts within which a spatial concept can be interpreted (i.e. context-dependency); and (2) there are multiple interpretations of the same spatial concept in the same context (i.e. fuzziness). Existing studies have addressed the fuzziness aspect of the vagueness problem to a great extent, but little work has been done to handle the context-dependency sub-problem. This study focuses on the context-dependency nature of vague spatial concepts. The goal is to enable effective communication of vague spatial concepts in spoken language human-GIS interaction through better managing, sharing, and utilizing contextual knowledge. Toward this goal, this study has made two major contributions. Firstly, this study provides a Human Communication Framework (HCF) to facilitate our understanding and handling the vagueness problem in human-GIS communication. The HCF explains the vagueness problem in human-human communication, and human communication principles for handling this problem. It helps our understanding about major origins of vagueness, major types of contextual factors, distributed nature of context knowledge and need for building a shared context involved in human-GIS communication. Secondly, this study also provides a collaborative dialogue approach for the GIS to handle the context-dependency problem through collaborative human-GIS dialogues. This approach is driven by the success of collaborative human-human dialogues and distributed nature of context knowledge in human-GIS communication. An agent-based computational model, the PlanGraph model, has been developed to support this approach. This model enables the GIS: (1) to build and keep track of the shared context involved in human-GIS communication, (2) to understand the meaning of a vague spatial concept under constraints of the shared context, (3) to repair the shared context and further reduce vagueness through collaborative human-GIS dialogues, and (4) to effectively communicate vague spatial concepts with the user in various contexts.