Optimized Conflation of Authoritative and Crowd-Sourced Geographic Data: Creating an Integrated Bike Map

A complete and accurate geographic dataset is critical for relevant analysis and decision-making. This chapter proposes a four-step geographic data-conflation system: preprocessing, automatic conflation, evaluation, and manual adjustments. The automatic-conflation component uses an optimization approach to find matched features and a rubber-sheeting approach to complete spatial transformation. This system was tested using two bikeway datasets in Los Angeles County, California, from an authoritative source (Los Angeles County Metropolitan Transportation Authority) and an open source (OpenStreetMap). While bikeways that are already in both datasets are improved in terms of positional accuracy and attribute completeness, the conflated bikeway dataset also integrates complementary data in either of the input datasets. Experiments demonstrate the advantages of using crowd-sourced data to improve official bikeway data, which is important for building and maintaining high-quality bicycle-infrastructure datasets. The framework described in this chapter can be adapted to conflate other types of data themes.