6 Automated Geographical Information Fusion
127
The approach holds considerable promise for application to Web-based environments. The increasing availability of geographical information from Web-based
sources is only of limited use unless it is accompanied by concomitant ability to
combine those information sources in a meaningful way. Non-expert users cannot
be expected to do this unaided, so automation is an essential step in extending the
usability of Web-based GIS into a range of new applications and domains.
However, there are several research issues to be addressed before practical automated geographical information fusion systems become a reality, including the
following.
• Inclusion of Human Expert Domain Knowledge: Although the rosetta approach
aims to enable fully automated information fusion, it is also important to allow
the inclusion of partial human expert domain knowledge where it already exists, and integrate this knowledge with automatically inferred knowledge. Some
initial techniques for dealing with this issue are presented in [22].
• Integration with Existing Mediator Architectures: The extensive work on existing mediator architectures and ontology-based GIS (cf. Chap. 7) is complementary to the goals of a rosetta system. Future work aims to integrate both in an
“intelligent geomediator architecture,” which provides the integration capabilities of a mediator with the alignment capabilities of a rosetta system.
• Regions with Broad Boundaries: A high-priority goal of current research is to
extend the existing formal rosetta systems with the ability to operate with vague
categories, where the extents of those categories have broad boundaries.
• Automated Thresholding: In addition to developing new techniques for dealing with imperfection, current research is investigating developing automated
thresholds for reasoning in the presence of inaccuracy and imperfection, as discussed in Sect. 6.5.4.
• Further Spatial Relationships: The inferences discussed in this chapter all concern containment or overlap between extensions of categories. However, given
the rich variety of spatial relationships embedded within spatial data, it is expected that many more types of spatial relationships might be useful as a basis for
inductive inference, including topological and metric relationships (cf. Chap. 8).
• Spatially Varying Alignment: The approach presented in this chapter aims to
infer alignments that are non-spatial, in that they hold for all locations in space.
Developing rosetta systems that can infer spatially varying ontology alignments
(i.e. semantic relationships that hold only in specific regions of geographical
space) will potentially provide much greater flexibility in defining future fusion
systems.
Acknowledgments
Matt Duckham is supported by the Australian Research Council under ARC Discovery Grant DP0662906, entitled “Automatic fusion of geoinformation: The intelligent
geomediator architecture (iGMA).” Collaboration between Matt Duckham and Mike
Worboys has been partially supported by funding from the Australian Academy of
127
The approach holds considerable promise for application to Web-based environments. The increasing availability of geographical information from Web-based
sources is only of limited use unless it is accompanied by concomitant ability to
combine those information sources in a meaningful way. Non-expert users cannot
be expected to do this unaided, so automation is an essential step in extending the
usability of Web-based GIS into a range of new applications and domains.
However, there are several research issues to be addressed before practical automated geographical information fusion systems become a reality, including the
following.
• Inclusion of Human Expert Domain Knowledge: Although the rosetta approach
aims to enable fully automated information fusion, it is also important to allow
the inclusion of partial human expert domain knowledge where it already exists, and integrate this knowledge with automatically inferred knowledge. Some
initial techniques for dealing with this issue are presented in [22].
• Integration with Existing Mediator Architectures: The extensive work on existing mediator architectures and ontology-based GIS (cf. Chap. 7) is complementary to the goals of a rosetta system. Future work aims to integrate both in an
“intelligent geomediator architecture,” which provides the integration capabilities of a mediator with the alignment capabilities of a rosetta system.
• Regions with Broad Boundaries: A high-priority goal of current research is to
extend the existing formal rosetta systems with the ability to operate with vague
categories, where the extents of those categories have broad boundaries.
• Automated Thresholding: In addition to developing new techniques for dealing with imperfection, current research is investigating developing automated
thresholds for reasoning in the presence of inaccuracy and imperfection, as discussed in Sect. 6.5.4.
• Further Spatial Relationships: The inferences discussed in this chapter all concern containment or overlap between extensions of categories. However, given
the rich variety of spatial relationships embedded within spatial data, it is expected that many more types of spatial relationships might be useful as a basis for
inductive inference, including topological and metric relationships (cf. Chap. 8).
• Spatially Varying Alignment: The approach presented in this chapter aims to
infer alignments that are non-spatial, in that they hold for all locations in space.
Developing rosetta systems that can infer spatially varying ontology alignments
(i.e. semantic relationships that hold only in specific regions of geographical
space) will potentially provide much greater flexibility in defining future fusion
systems.
Acknowledgments
Matt Duckham is supported by the Australian Research Council under ARC Discovery Grant DP0662906, entitled “Automatic fusion of geoinformation: The intelligent
geomediator architecture (iGMA).” Collaboration between Matt Duckham and Mike
Worboys has been partially supported by funding from the Australian Academy of