Assessing the Quality of Spatio-Textual Datasets in the Absence of Ground Truth

Abstract

The increasing availability of enriched geospatial data has opened up a new domain and enables the development of more sophisticated location-based services and applications. However, this development has also given rise to various data quality problems as it is very hard to verify the data for all real-world entities contained in a dataset. In this paper, we propose ARCI, a relative quality indicator which exploits the vast availability of spatio-textual datasets, to indicate how confident a user can be in the correctness of a given dataset. ARCI operates in the absence of ground truth and aims at computing the relative quality of an input dataset by cross-referencing its entries among various similar datasets. We also present an algorithm for computing ARCI and we evaluate its performance in a preliminary experimental evaluation using real-world datasets.