After trying to book hotels in Shanghai, Beijing, and Tokyo, I understand firsthand the frustrations surrounding failed attempts at globalization. Though it sounds academic, metadata is key to providing a personalized, language-independent user experience. It improves how products are stored and displayed on a Web site, and helps ensure product information is accurate and up to date across all languages.
My previous metadata columns have illustrated how to enhance personalization through the use of metadata, create metadata-driven configurators, better understand users, and simplify product categorization through a well-thought out attribute-based taxonomy. There are two other dimensions to all these ideas: language and regionalism.
Localization Vs. Language
The first step to creating localized versions of a Web site is to understand the difference between the content on the site and how that content is displayed.
Most mobile phone companies operate globally, for example. Their products, however, aren’t always globally available, and sometimes they differ slightly between regions. On a macro level, a phone may be available in Europe but not in North America. On a micro level, a phone might be available with one carrier but not another in the same area code.
In addition, there’s language complexity. Regional information must be presented in whatever language the customer understands, yet language is independent of region. A U.S. resident may speak Spanish, Japanese, or Italian rather than English. Similarly, someone residing in Germany might speak English rather than German.
The challenges for such sites are both on the front end (what the user sees) and the back end (what the database sees). On the front end, every product must be described in every language. On the back end, database and programming logic shouldn’t have to deal with every language.
Using Metadata in a Language-Independent World
As I’ve written, metadata can be used to categorize products. With mobile phones, every phone would have specific attributes attached to it: technology (GSM, CDMA, PCS, etc.), features (camera, video, text messaging, etc.), regions (North America, Europe, etc.), carriers (T-Mobile, Verizon, Cingular, etc.), and many others. Users should be able to search and filter results by any of these attributes.
On the back end, these attributes are language-independent. A search on camera phones will always be the same on the back end, because in a metadata environment, “camera phone” equates with some database ID.
Say “camera phone” is ID 91. No matter the language used on the front end, the system searches for products attributed with ID 91. The trick is ID 91 has various display fields in the database, each representing a different language. So when displaying a list of possible search criteria, the front-end system could display ID 91 as “camera phone” or “Kameratelefon.”
Similarly, the site’s search engine must understand all these terms are synonyms for the attribute stored in the database as ID 91. The effects of using metadata this way are profound. Without this functionality, you couldn’t enable any kind of filtering or search in a language-independent way.
Data can become dirty if it must to be input several times. If each phone requires its technical specs to be rewritten in every language, the likelihood is high one language will say the phone is “dual band,” while another will claim it’s “tri-band.” Simple human error can hugely affect whether someone buys the phone or not.
By attaching the “camera phone” attribute (ID 91, in my example), the front-end can translate it into any language without further human intervention. Data are input only once, which means automatic consistency across all languages.
Who Does It Well?
Take a look at Siemens’ Web site. Siemens is a huge player in the mobile phone marketplace, especially in Europe. Its phones are really cool (I have two), and its Web site puts into practice all the issues addressed in this column.
At the top of the page is a menu that allows the user to select a country and region. The user can select, for example, to view products available in Belgium in Dutch or French, in Luxembourg in French or German, and in China in Chinese or English.
Once users select a region and language, they can view each phone’s technical information. The metadata (e.g., whether it’s a camera phone) is in the appropriate language, but you can bet in the back end all the information is stored in one language. Translation occurs in the front end, and the content management system (CMS) decides to display ID 91 as “Kameratelefon” instead of “camera phone.”
Are You International?
Does your company implement metadata and attributes in a language-independent fashion? Do you have globalization horror stories? Let me know, and I may use them in a follow-up column.
Until next time…