Recent emergence of advanced technologies such as Web 2.0 and Global Positioning Systems (GPS) has given birth to a revolutionary style of mapping. Map making is no more a sole authority of professional cartographers. Today there are six billion human sensors of geographic information. With a mere access to internet and some readily available geo-referencing mechanism anyone can contribute to the generation of geographic information. Goodchild termed this phenomenon as Volunteered Geographic Information [VGI] and asserts that VGI can lead to hitherto unknown, innovative and cheap production and usage of maps [1, 5]. OpenStreetMap(OSM) is by far the most famous example of VGI. Founded by Steve Coast in London in 2004, it is a collaborative project which aims at the creation of free editable map of the world. A rapidly increasing demand for non-proprietary geodata has made OSM highly popular. Several applications based on OSM data such as route planning and geocoding [2], location based services [3] and 3D [4] are being developed. This plethora of crucial applications necessitates the quality assessment and/or improvement of OSM data. As an outcome of several efforts to study the quality of geographic data, five fundamental dimensions for geospatial data standard have been agreed upon [5]. These include positional accuracy, attribute accuracy, logical consistency, completeness, and lineage. Girres and Touya [6], while assessing the quality of French OSM, also considered semantic accuracy, temporal accuracy and usage as potential data quality indicators. Several studies have been conducted to study the evolution and quality of OSM. OSM quality has been assessed by comparing OSM with official datasets [6-8]. However, from [9] it can be observed that there can be several local significant features which may be missing from official or proprietary datasets. Also, comparison with official datasets usually involves manual matching of features. In this paper, a machine learning based solution for improving the quality of OSM data has been presented. In particular the semantic accuracy of road network with special focus on pedestrian and residential roads is being studied. A graph based representation of OSM transportation network has been developed. Artificial Neural Networks (ANN) based framework is being developed to model relationships between data. The rest of the paper is divided as follows. Section 2 describes the dataset used in the study. Section 3 presents the methodology used for road network quality improvement. In section 4 a discussion on preliminary results is presented. This is …
[1]
Peter Mooney,et al.
Using OSM for LBS – An Analysis of Changes to Attributes of Spatial Objects
,
2012
.
[2]
Haklay.
How good is OpenStreetMap information ? A comparative study of OpenStreetMap and Ordnance Survey datasets for London and the rest of England
,
2008
.
[3]
Alexander Zipf,et al.
Generating web-based 3D City Models from OpenStreetMap: The current situation in Germany
,
2010,
Comput. Environ. Urban Syst..
[4]
Guillaume Touya,et al.
Quality Assessment of the French OpenStreetMap Dataset
,
2010,
Trans. GIS.
[5]
P. Mooney,et al.
Comparison of the accuracy of OpenStreetMap for Ireland with Google Maps and Bing Maps
,
2010
.
[6]
Michael F. Goodchild,et al.
Assuring the quality of volunteered geographic information
,
2012
.
[7]
M. Goodchild.
Citizens as sensors: the world of volunteered geography
,
2007
.
[8]
Claire Ellul,et al.
Assessing Data Completeness of VGI through an Automated Matching Procedure for Linear Data
,
2012,
Trans. GIS.
[9]
Manfred M. Fischer,et al.
ARTIFICIAL NEURAL NETWORKS: A NEW APPROACH TO MODELING INTERREGIONAL TELECOMMUNICATION FLOWS*
,
1994
.