Presentation Title

Supporting Human Data Interaction in the Context of Ontology Mapping – Visualizing Ontological Relationships using Matrix

Faculty Mentor

Dr. Bo Fu

Start Date

18-11-2017 12:30 PM

End Date

18-11-2017 1:30 PM

Location

BSC-Ursa Minor 117

Session

Poster 2

Type of Presentation

Poster

Subject Area

engineering_computer_science

Abstract

In recent years, ontologies have become a key platform to structure data in various disciplines. Given the heterogeneous nature of ontological data, it is a pressing issue to achieve data interoperability across ontologies that are typically developed by different people with diverse backgrounds and expertise. Ontology mapping has emerged as a main mechanism in overcoming data heterogeneity. Data visualizations are designed to facilitate human interaction with complex datasets, as they have been shown to be helpful in identifying patterns and relationships. Though automated mapping algorithms provide suggestions, these techniques are error prone if the semantical context is not taken into account. In the scenarios where mapping algorithms create results with high confidence given ontological entities with similar spelling and distinct semantics, human intervention is required to evaluate the correctness or completeness of a mapping set. In order to better assist the human with such tasks, this research presents an interactive matrix visualization to illustrate the mappings between a pair of ontologies. In addition, the matrix visualization is evaluated in controlled user studies against node-link diagrams. The goal is to determine if one visualization technique is more effective than the other. The evaluation will measure the usability of the two in terms of effectiveness, measured as how successful the users will be at completing given tasks, and efficiency, measured as the time it takes to complete tasks, in assisting users in the process of generating correct mappings. We anticipate the results arising from this research will identify usability issues of the proposed matrix visualization as well as the commonly used node-link diagrams in the context of ontology mapping. The findings will generate new knowledge specific to the two visualization techniques and inform designers in the process of building more usable interactive visual support to better facilitate human data interaction.

This document is currently not available here.

Share

COinS
 
Nov 18th, 12:30 PM Nov 18th, 1:30 PM

Supporting Human Data Interaction in the Context of Ontology Mapping – Visualizing Ontological Relationships using Matrix

BSC-Ursa Minor 117

In recent years, ontologies have become a key platform to structure data in various disciplines. Given the heterogeneous nature of ontological data, it is a pressing issue to achieve data interoperability across ontologies that are typically developed by different people with diverse backgrounds and expertise. Ontology mapping has emerged as a main mechanism in overcoming data heterogeneity. Data visualizations are designed to facilitate human interaction with complex datasets, as they have been shown to be helpful in identifying patterns and relationships. Though automated mapping algorithms provide suggestions, these techniques are error prone if the semantical context is not taken into account. In the scenarios where mapping algorithms create results with high confidence given ontological entities with similar spelling and distinct semantics, human intervention is required to evaluate the correctness or completeness of a mapping set. In order to better assist the human with such tasks, this research presents an interactive matrix visualization to illustrate the mappings between a pair of ontologies. In addition, the matrix visualization is evaluated in controlled user studies against node-link diagrams. The goal is to determine if one visualization technique is more effective than the other. The evaluation will measure the usability of the two in terms of effectiveness, measured as how successful the users will be at completing given tasks, and efficiency, measured as the time it takes to complete tasks, in assisting users in the process of generating correct mappings. We anticipate the results arising from this research will identify usability issues of the proposed matrix visualization as well as the commonly used node-link diagrams in the context of ontology mapping. The findings will generate new knowledge specific to the two visualization techniques and inform designers in the process of building more usable interactive visual support to better facilitate human data interaction.