Measuring conflation success

Abstract

We are immersed in the Big Data era, where there is a large amount of heterogeneous data, both in time and spatial scales. This data starts to be streamed in real time from different devices and sensors, well illustrated by the new concept of Smart Cities. Conflation processes play an important role in this scenario, defined as the procedure for the combination and integration of different data sources, improving the level of information of the result. It also allows to update geographical databases (GDB), conflating different kind of sources where one of them is more accurate or updated than the other. Regarding geometric conflation, the procedure involves transforming features from one data source to another, minimizing the geometric discrepancies between them. Accuracy has to be taken into account in these processes, and the results need to be measured and evaluated in order to have a better understanding of product quality. In this paper, conflation evaluation process is described along with the different metrics and approaches to assess its accuracy.

An Aggregated Shape Similarity Index: A Case Study of Comparing the Footprints of OpenStreetMap and INSPIRE Buildings

“…They are based, for example, on the calculation of centroids and distances between them, buffers, the geographic context of objects [16,18], using a statistical approach [19], etc. Measuring the success of conflation using various measures is described, e.g., in [34]. The calculation of similarity measures should make it possible to effectively identify different representations of spatial objects, especially when integrating several heterogeneous data sources, updating one data source with another, or checking the quality of spatial data.…”