Getting Lost in Cities: Spatial Patterns of Phonetically Confusing Street Names

Abstract A large street network is likely to contain duplicated or similar sounding street names. These conflicts can cause confusion in communication between people or in machine‐human interaction. Municipal authorities have begun to see the importance of uncovering these existing street name conflicts and mitigating future ones, for improved record keeping, emergency response, etc. However the commonly used Soundex phonetic algorithm is generally considered to produce poor similarity results in terms of uncovering street name conflicts. This study reports on a new fusion algorithm that combines phonetic methods and approximate string matching for street names, weighted by street type suffix ( A venue, B oulevard, C ourt, etc.), to quantitatively measure the collision/confusion potential presented by a pair of streets. This algorithm is then applied to the entire street network in G reater M elbourne, A ustralia, and the pattern of collisions at various spatial scales, and within municipalities, is mapped. The goal of this work is to produce better tools that can aid policy makers, administrators and industries dealing with location‐based services to make better decisions when assigning and disambiguating street names.