Introduction

“Behold, Google has eliminated the need for metatags, so classifications and their associated software can just go quietly…”

Hold it. Next time you’re at the public library, or Ebay, or Walmart, and see if you can get what you want by demanding…”I want what everyone else wanted before me”.

How could we survive without classifications of medicines, of products and services of all types, of classes of revenues and expenses etc.?

Let’s face it. The ability to search our enterprise using only one or two words, AND get the results we want, EVERY TIME, is search heaven….but we haven’t reached heaven. In fact maybe we are in purgatory, because generally our results stink.

So, to help us improve our “answer finding”, many industry, business and company matters have been reduced to a set of (useful, generally acceptable, organized) terms, and our documents or records were “tagged”; that is, the term-keyterm-keyword-tag was assigned to the material either physically or virtually.

But creating/defining tags and performing tagging has proven to be a substantial ongoing task, requiring a variety of skills, and those skills are not always in abundance in our workgroup or IT department. What’s more, besides the work of tagging, the information system user needs to “see” the tags and their organization, from several perspectives, so s/he can exploit them and find the appropriate material, without having to resort to search. So someone needs to think through the interface.

The enterprise search utility vendors have never picked up the tag-building ball, preferring to shift the responsibility to the end use customer (or claim their results are so good, tags are unnecessary!). Even Verity, whose original search technology enabled leveraging repeatable “topics” for its customers, never delivered the blades with the razor. Classification software vendors such as Autonomy or InXight claim outstanding results AFTER their techniques are “trained” (a taxonomy, or set of tags is developed, and lots of appropriate content samples are found for each term).

“My Google results are sooo good, and my internal search results are soooo bad”

Be prepared for a shock; the Google approach is a metatag approach. It’s just not called that for reasons that only marketeers will understand. The “site popularity contest” technique, which makes Google the current favorite of public Internet engines, virtually “tags” web sites with terms, based on how many other sites containing the same term point to them. Then the exact match between the search you entered and the most popular sites in the group is easy.

Perhaps the “popularity” notion sounds too negative, or self-serving, trivial. After all, the sites that decided to point to another site probably had solid logical reasons. Hopefully no diabolical hackers will put bogus links in, because there’s no way to tell.

The technique can produce excellent results on the Web, where “tagging” can be accomplished because the links are already in place, whatever their motivation. But what does site/page/file popularity have to do with your organization content? Absolutely nothing. Unless of course, you have created (or forced your colleagues to create) a system of links in which each page/file/record points to better/newer/more comprehensive/more official material… or unless you have created a taxonomy and tagged (created a virtual pointer to) your files/records/pages.

Can an organization’s content be “popularity-readied”? Is the organization user base ready to say “this document or record is a great fit for my search XXXXX becasue so many other people have looked at it”? Armed with that data in a database, (eventually) future XXXXX searches will return the content rated by popularity, if the indexing/search tool is adapted to look for the popularity file.

Subject/meta tag definition and auto-assignment of tags to pages/files/records seems trivial by comparison.

Tell you what. Tag your content based on an industry or enterprise taxonomy. In the worst case, EVERY result will be relevant, though it may not be your favorite. Then use existing search experience enhancement software to capture what was read, and build up the “popularity rating” for the group of already relevant set of content. Then order the list by popularity.

Summary

Google and other internet search engines may have declared that taggig is no longer used to determine relevancy, but for your enterprise search to succeed, you need to continue tagging to make sure your critical content is available to users.

[John Lehman is Co-Founder and President of HighClassify Inc., a provider of taxonomies and content classification services and solutions.]