INDEA Geocoding

Geocoding is the process of linking data to a geographic location by comparing data representations contained in two or more files. In the INDEA geocoding application developed for EBC, the aim is to identify the Electoral District (ED) and Voter Area (VA) corresponding to an input address, by associating this address with a road segment in the INDEA.

Two types of geocoding are implemented, namely address geocoding and point geocoding. This was necessary to handle the two types of road layers (standard and source) in the INDEA. These differ in that each road segment in the standard layer has an explicit ED/VA code as well as address ranges associated with each side, while those in the source layer have address ranges but no ED/VA codes.

For roads in the standard layer, address geocoding is realized by first identifying the INDEA coverage road segment whose address range contained the input address. Since each road in the standard layer has an associated ED/VA for each side, address geocoding for roads in this layer then simply involves returning the ED/VA code corresponding to the appropriate side of the identified segment. For roads in the source layer, the road segment containing the address is first identified, and interpolation is then used to estimate the location (i.e., latitude and longitude) corresponding to the address. A point-in-polygon function is then used to determine the ED/VA polygon containing this location.

In order to geocode, the input address must first be matched with its counterpart in the database. A major problem here is that street names, street directions, etc., may not always have exactly the same spelling, and so the process for finding matches must allow for "fuzzy" matching. The INDEA geocoding application achieves a high degree of success in the matching, even when the address data is inconsistent or inexact, by using the commercial address-matching product MatchWare/PACE (MatchWare Technologies, Inc., recently acquired by Vality Technology Inc.) to perform address standardization and matching based on probabilistic schemes. (Integration of MatchWare/PACE into the geocoding required the previous generation of two MatchWare databases, each containing standardized addresses.) The matching also involves the use of special-purpose algorithms for using MatchWare/PACE to perform matching under various conditions, and to use its output to identify optimal and/or reliable matches.

The use of MatchWare/PACE, in conjunction with the strategies for matching and the techniques for address and point geocoding, provides an effective system for satisfying the geocoding requirements of EBC.