Hello,

I have decided to work on a little side project. I want to create a web application that utilizes web focus to combine sales rep data with Google maps. I would like to find a way to convert several thousand address records to geocode coordinates into an XML or JSON(Google) format to create a clustered marker system throughout the US.

Has anyone done something even remotely similar to this? Trying to figure out if there is a way to do a mass geocode conversion.

So many sub-topics here, I'll try to cover as many as are reasonable.

* Mass geocoding - - This is possible in several ways. Google and Yahoo are 2 of the
geocoding methods available for free (mostly) and expose the methods through their respective
toolkits. I know Yahoo has a 50,000 hit/ IP address/ day limit. I would surmise that Google
has a similiar limit but have never seen it stated.

You could use WebFOCUS to create a web page and XML doc reference. An onload javascript could
create a DOM from the XML and iterate through it calling the appropriate geocoding API and
updating the nodes with the LAT LON return values. An HTML POST request should offer enough to get the result set back to the server.

*** Note: Be aware that when something is free you may get what you pay for. If you have
strange/ new addresses or slightly mispelled street names or any number of possible geocoding
issues then the free geocoding services may return a high number of non-locatable addresses.

There are commercial geocoding services that offer a much more accurate street dataset for
use which will give you better accuracy. ESRI ArcWeb Services is one such service. $1,250 US
for 100,000 service credits. Navteq is another but I don't have any details.

* Google Maps KML - This is a nice means to display data but be aware of the extra
effort to create the KML with a TABLE request. There are some nice features to the google
mapping service but they can require XML that is several nodes deep which is a text concatenation challenge.

One possible means to achieve this is to use a nice, but undocumented feature in 7.6 WebFOCUS
that allows you to specify an XSLT stylesheet that can be used to transform ON TABLE PCHOLD FORMAT XML (fxf document) output to anything you want. XSLT takes a bit of noodling to get your head around it but is pretty powerful once you do. Be aware that the library WebFOCUS uses supports XSLT 1.0 only.

There are 2 variables to include on the URL for a WFServlet request to invoke the transformation:
* IBIF_report_xsl_location = the directory path where the .xsl document is stored.
* IBIF_report_xsl = the actual XSLT stylesheet document.

I tested this feature with a simple transformation request but it did not use any GIS data or produce Google KML.

*** Note: JSON - JavaScript Object Notation is not a Google format. It is a language
independent subset of the JavaScript language and was created at the end of 1999. Check
out http://www.json.org/ for more info.

* BEWARE: Shameless product pitch, WebFOCUS GIS Adapter - I'm sorry, it's my job.
The WebFOCUS GIS Adapter has been available for 4 years and offers a bidirectional interface
between ESRI GIS products and your data.
Google and Yahoo only provide "location context" to your data.
"Here's your data. Here's your data on a map. No more questions!".

When you have more questions and you need "location intelligence" than a true GIS is the only answer. In the 7.6 WebFOCUS release we completed the support for "Synthetic Map Layers".
This support allows you to read stored spatial data (LAT-LON) and create points, lines
and polygons on the map. You can then select from those synthetic layer features to filter
subsequent report or graph requests.

This has been a long winded response so if you have more questions feel free to post them
or email me directly.

neuro
whats the level of granularity of your data?
you mention 'several thousand'? there are 40k zipcodes, so 'several thousand' would be a smaller subset, if zipcode granularity is all you're after. ... and are you wanting to display the entire map of the us (in which case zip3 granularity is all that is reasonably visible)? or much smaller bits? If zipcode gran is all you want, do you have already the latlon for them? or are you looking for a cheap source?