Machine learning based tool for identifying errors in CAD to GIS converted data

Abstract

Recent advances in Geographic Information System (GIS) has made the storage, manipulation, and analysis of spatial data easier than ever. As a result, many public and private agencies have switched from the traditional Computer Aided Design (CAD) format to GIS for storing information about their infrastructure. The existing data stored in CAD therefore needs to be converted to GIS, and this process brings about at least two problems. First, GIS requires geographical coordinates that CAD data do not have, and the accurate projection and alignment of the infrastructure is not straightforward. Second, the original CAD data often possess errors such as overlapping lines, split lines, presence of text, and human errors, in addition to errors introduced during the conversion process. The goal of this study is to develop and apply a tool that can identify these errors and correct them automatically without human intervention. This study focuses specifically on the identification of the errors using decision tree learning. This tool can significantly reduce the time spent identifying errors on the GIS data obtained after the conversion. Finally, as a case study, this tool is applied to the wastewater system of the University of Illinois at Chicago campus.