Graph-based strategies for matching points-of-interests from different VGI sources

Several urban studies have been relying on spatial data provided by Volunteered Geographic Information (VGI) sources. The matching of features from different VGI platforms may serve to assess and improve the reliability and completeness of VGI data. In this paper, we propose two strategies for matching points-of-interests (POIs) based on a graph whose nodes and edges represent the POIs and their possible matching pairs, respectively. We called these strategies Best-Best Match and Combinatorial Match. The former drastically reduces the amount of ambiguous edges in comparison to the baseline method, which matches each POI with its best matching candidate. The later completely rules out ambiguous matches. Both strategies are able to produce 1:0 matchings, thus tackling the issue that sometimes a POI from a reference dataset is not represented in a second dataset. The Combinatorial Matching strategy consists in extracting all possible subsets of edges from the graph in which no node occurs more than once. It then selects the subset with the highest sum of edge weights. As a first evaluation of these strategies, we conducted an experiment for matching POIs from OpenStreetMap and Foursquare. The results show that our two proposed strategies perform as comparatively good as the baseline method.