Defining Neighborhoods

Humans think and talk about regions imprecisely in terms of vague concepts (e.g. downtown). While administrative regions such as area codes and land parcels have sharp boundaries imposed on them, other regions concepts used by people are more fuzzy. The use of vague spatial concepts in geospatial communication have been studies for many years (see for instance Daniel Montello’s Where’s downtown?: Behavioral methods for determining referents of vague spatial queries). However, the understanding human’s perspective of the space hardly translates to its digital definition stored geographic information systems. As a consequence, the local search and location-based services industry has invested a large amount of energy and money in obtaining neighborood expertise. Companies such as Urban Mapping sells its extensive user-centered neighborhood datasets to major search engines. New approaches profit from people submitting their own version of the boundaries for the same neighborhood. WikiMapia is an example of this kind of community editing. It defines specific areas (roads, parks) with polygonal entries (see Matt Jones’ Wikimapia Invades Google Earth!. Similarly, the Intelligent Middleware project at the MIT aims at providing a mechanism for accumulating local knowledge about neighborhood-scale land use. This people-generated content helps re-interpreting the administrative datasets and develop customized analyses of neighborhood conditions.

A classic Neighborhood areas & census block groups (2000) by the Seattle City Clerk’s Office neighborhood map atlas (right). People-defined neighborhoods and areas of Barcelona in Wikimapia

Nowadays, there are new sources to implicitly reveal the crowd “mental maps”. This has mainly been done by via tag maps such as the ones generated by WorldExplorer. In Tracing the Visitor’s Eye, I am currently taking this idea a bit further and try to generate the “area of influence” of monuments and points of interest. This work should reveal a spatial relevance of geographic objects as zones (i.e. neighborhoods) instead of administratively-defined points or rectangles (like the ones provided by GeoNames). Inspired by Jonathan Raper’s concept of “Geographic relevance” (match between area of attention and area of influence). For instance, areas defined implicitly by people’s behaviors (mobility, digital traces) and activities (geotagging) could help provide more relevant answers to queries based on monuments and proximity such as “what is close to the Empire State Building?”

A crowd’s mental maps of Barcelona in WorldExplorer

Update: I forgot about the The Neighborhood Project (mentioned previously), as an attempt to map what street addresses people on Craigslist consider to be within certain neighborhoods. One problem with, though, it is that Craigslist provides a certain list of neighborhood names and they aren’t necessarily the same ones that people use in real life.

Neighborhoods defined by people using Craigslist.

Jonathan Raper’s Geographic Relevance presented at LBS2007

Relation to my thesis: Exploring the relationship between people’s perception of neighborhood and the concept of location information granularity.