Emad Fanous

SEW has an article on a patent awarded to Google:

SUMMARY OF THE INVENTION

One aspect of the invention is directed to a method for generating geographic coordinate information that includes receiving at least one term that specifies an address and accessing a table defining coordinate information for ranges of addresses to find an intersection of sets of rows in the table that correspond to the at least one term. The method further includes reading geographic coordinate information from the table at the intersection of the sets of rows in the table.

This sounds like taking user input and trying to identify the proper location they are searching for. If I type in “pasadena,” I am referring to “Pasadena, CA” or “Pasadena, TX?”

A second aspect of the invention is directed to a system for geocoding postal addresses. The system includes a table including rows that each correspond to a range of one or more addresses. Each of the rows include fields that define the row. A geocoding component generates geographic coordinate information for a received address specified by one or more terms that correspond to the fields by locating at least one row in the table that corresponds to an intersection of a number of sets of rows defined by the terms in the received address.

This second point seems to refer to geocoding addresses for the locations in their location index.

Yet another aspect of the invention is directed to a method for extracting addresses from a document. The method includes identifying possible address terms based on predetermined rules, verifying that the identified possible address terms are address terms by comparing the address terms to a table containing known addresses, and examining a relative position of the verified possible address terms in the document to determine whether the verified possible address terms form a valid address.

This is how they extract addresses from crawled pages in order to get it into the index.

The actual idea of handling things this way is not a new/unique one. Most every decent local product already does it this way…however, it seems like Google might be describing a unique way of technically doing it. However, it seems I may never know due to the strange behavior of the USPTO site. It is embedding tiff images in the HTML pages which aren’t rendering properly for me.

Anyway, it appears the filing was in Sep 2003. If you’ll remember, in 2002, there was a google programming contest which required entries to be submitted by April 2002. I believe they announced the winner (in May 2002) was Dan Egnor’s Google Geocoder.

On the Google Geocoder site:

It includes a geocoder (which uses TIGER/Line data to turn street addresses into latitude/longitude coordinates), a simple indexer that looks for addresses and keywords in documents, and a query engine to search for documents matching certain keywords that also contain addresses within a certain distance of a target location.

This may have been the beginning of Google Local…I remember watching from the front seat as the winner was announced and I was working at citysearch.

(No Ratings Yet)
 

 Loading ...