Categorizing Data Imperfections for Object Matching in Wastewater Networks Using Belief Theory

Nowadays, data on wastewater networks covering the same geographical territory are available from different sources. The fusion of multi-source spatial data provides a new and richer dataset that can serve several purposes such as quality improvement, decision making, or delivery of new services. It has given rise to several research works focused on the visualization, analysis, and fusion of spatial databases. However, the original data is often imperfect: imprecise, uncertain, vague, incomplete, etc. Therefore, it is essential to use formalisms allowing the modeling of imperfections and to propose adapted fusion mechanisms. In this work, we aim to handle data imperfections in a generic way. We first propose a categorization, according to several dimensions, of data imperfections encountered when fusing multi-source spatial data. We then propose to model these imperfections according to the formalism of the belief theory. We consider our conducted experiments that allowed us to match nodes and edges in the different cases of data imperfection, as promising.