Incremental identification of topological errors in spatial data

This paper proposes a semi-automatic decision tree approach to identify topological errors within spatial data. In contrast to the identification of incorrect objects using manually defined spatial integrity constraints, a learning process is applied to induce them from an error-free geodata. This data is used to create an initial decision tree. Once new geodata is made available, the tree is augmented in an interactive way in case this data is unclassifiable or misclassified. This approach is semiautomatic due to the user interaction during the process of identification.