Formalising cartographic generalisation knowledge in an ontology to support on-demand mapping

Gould, Nicholas Mark (2014) Formalising cartographic generalisation knowledge in an ontology to support on-demand mapping. Doctoral thesis (PhD), Manchester Metropolitan University.

Preview

Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (10MB) | Preview

Abstract

This thesis proposes that on-demand mapping - where the user can choose the geographic features to map and the scale at which to map them - can be supported by formalising, and making explicit, cartographic generalisation knowledge in an ontology. The aim was to capture the semantics of generalisation, in the form of declarative knowledge, in an ontology so that it could be used by an on-demand mapping system to make decisions about what generalisation algorithms are required to resolve a given map condition, such as feature congestion, caused by a change in scale. The lack of a suitable methodology for designing an application ontology was identified and remedied by the development of a new methodology that was a hybrid of existing domain ontology design methodologies. Using this methodology an ontology that described not only the geographic features but also the concepts of generalisation such as geometric conditions, operators and algorithms was built. A key part of the evaluation phase of the methodology was the implementation of the ontology in a prototype on-demand mapping system. The prototype system was used successfully to map road accidents and the underlying road network at three different scales. A major barrier to on-demand mapping is the need to automatically provide parameter values for generalisation algorithms. A set of measure algorithms were developed to identify the geometric conditions in the features, caused by a change in scale. From this a Degree of Generalisation (DoG) is calculated, which represents the “amount” of generalisation required. The DoG is used as an input to a number of bespoke generalisation algorithms. In particular a road network pruning algorithm was developed that respected the relationship between accidents and road segments. The development of bespoke algorithms is not a sustainable solution and a method for employing the DoG concept with existing generalisation algorithms is required. Consideration was given to how the ontology-driven prototype on-demand mapping system could be extended to use cases other than mapping road accidents and a need for collaboration with domain experts on an ontology for generalisation was identified. Although further testing using different uses cases is required, this work has demonstrated that an ontological approach to on-demand mapping has promise.