The Future of Search

Factiva launched Search 2.0 beta this past week and the feedback from the press, industry analysts and customers has been great - Bloor Research, Shore Communications, and Information Today.

I am very excited about this launch because I believe it is the best example of utilizing leading edge technologies (FAST Search and Transfer) and sophisticated business taxonomy (Factiva Intelligent Indexing) to manage huge volumes of information to deliver the most meaningful results to individual users. 

Like so many innovations in information technology, Search 2.0 is going to raise expectations around information management – external sources like Factiva’s, but internal information as well.   

Comments

[Terry Steichen]

Clearly, adding a browsing interface to your content repositories makes sense. If (and it's a big "IF") your content is accurately categorized (IMHO, precision and recall must be well over 95%), you can deliver some real value. (Unfortunately, it's accomplishing the *categorization* goal that usually defeats those that try.)

However, what puzzles me about all the hoopla emphasizing the browsing taxonomy. Indeed, that's the focus of the patent you've been awarded. (BTW, not to be negative, but I wouldn't depend too much on that patent holding up if challenged. The index of a book or a modestly complex org chart fall under it, and these common kinds of things clearly predate the patent. But, I'm sure the patent made a bunch of lawyers happy, if not rich.)

But the browsing taxonomy is pretty arbitrary, and flexible - you can tune and change it pretty freely. In other words, it's no biggie. The fancy "polyarchical" moniker simply notes that there are different paths to the content (just as in, as mentioned above, the index of a book, where a page happens to have information on more than one topic, or an org chart including people with more than one function, etc.).

Anyway, what I wanted to suggest (in a more positive way) is that you might want to go beyond a browse-augmented search capability (which I agree is vastly superior to a raw query interface), and create some derived content from your (hopefully accurately) categorized content. Then you might have a real generator of profit that zooms right to the bottom line.

Just some thoughts.