You’re talking with a friend about your favorite movies, and you remember one that you saw a few years ago and liked a lot but you just can’t remember the name. It’s right on the tip of your tongue, but you can’t quite place it. You remember that the cast included George Clooney and Catherine Zeta-Jones, but the title escapes you. So you decide to do a Google search. You type "George Clooney Cath" and that’s as far as you get, because before you even hit the Enter key, the 4th "Google Instant" search results is "George Clooney Catherine Zeta Jones Intolerable Cruelty." Intolerable Cruelty, that’s it!
How did Google do that? You didn’t even complete the full search, and the search engine can predict the terms that you’re most likely to use, and delivers the results to you in real time.
Interactive search techniques like Google Instant have changed how we look for information online. We no longer need to know the exact search terms we want to use, in fact there are benefits to using less specific terms and letting the instant search results guide us (think "diners San Francisco.") Sometimes we find comparable results that we weren’t previously aware of. It can even be interesting to play a type of game, by typing your search slowly to see how many letters of the search term the website needs before it can predict your results. Usually it’s surprisingly few.
Problems like this fall into the category of "fuzzy matching," "predictive search," "context-free search," or even "artificial intelligence," and it’s a lot like what we do at Global-Z.
Here’s an example. You go to your favorite website to place an order, and the website asks for your billing and shipping information. You type in your billing address with a 5-digit ZIP code (e.g. 05201), and click the submit button. The website processes your information and returns your address with your full 9-digit ZIP code (e.g. 05201-0915.) Mail and package deliveries are routed more quickly and more cost-effectively using a full 9-digit ZIP code, but most people don’t know their 9-digit ZIP, or even how to formally write their address the way the post office or shipping companies want to see it.
Somebody needs to change the incomplete address that you provide into a complete and accurate address that the post office or delivery company can use. In the case of global address data, that somebody is Global-Z. Global addresses have additional complexity such as foreign characters, addressing formats (e.g. letters in the ZIP code instead of numbers), different alphabets such as Chinese and Japanese, etc., which make this a very difficult problem for most companies.
So what does that have to do with Internet search? It turns out that the techniques used to correct a complex international address are the same as interpreting an Internet search. The user provides input data in free-form with no structure or context, and the data needs to be parsed, analyzed, and matched against a reference data set in order to add context and match the information. To do this we license reference data sets from data suppliers all over the world, and we have very sophisticated algorithms that can take context-free data (without structure), analyze the data, match it to a data set, and add context and meaning to the data.
Similar techniques allow us to provide a variety other high-tech services to our customers, who are all large global marketing firms. Global-Z’s data quality and matching technology allows some of the world’s largest companies to better identify who their customers are and what their interests are, so that they can provide more meaningful and relevant marketing content to those customers, and improve their customer experience. And all of this is done right here in Bennington, Vt.
The next time you submit your contact information to a website and that information is corrected or improved in some way, it may be that Global-Z is behind the scenes improving your data.
Dimitri Garder is executive vice president of Global-Z International Inc.