A couple of weeks ago, I attended the ComputerWorld BI Perspectives conference.  At the conference, we heard several BI users and vendors discuss their version of BI mashups, such as combining a mapping widget showing weather patterns with reports about a retail company’s regional performance.  This entails doing nothing more than combining some widgets in a web page in the presentation layer of an application.  The audience was asked several times over the course of the conference whether or not they are using mashups today, and not surprisingly, pretty much everyone attending said that they aren’t because the value to their business is currently unclear at best.

The issue with BI mashups as currently defined (or if you want to get fancy, you could say “bashups” per Mark Whitehorn’s article in searchdatamanagement.com back in April 2008, as “data plus business intelligence”), is that they’re generally just pulling content from other sources – whether a list of articles with hit highlighting from a search engine or an image tagged as related to the data such as a map of territories or regions – and placing this alongside the data in one page view.  I guess it’s better than having to navigate between windows to see these separate islands of content, but in general, it doesn’t really add much benefit.

So I want to propose a new approach to mashups.  How about thinking of a mashup as what happens behind the scenes in the engine driving the analysis through a portal, BI application interface, or MS Office front end?  More importantly, how about defining a mashup to include more than the ability to analyze transactional data with some pretty pictures or a search bar alongside numbers, and instead defining a mashup as an engine that provides a way to query and analyze across all types of content to make better business decisions?

As far as I can tell, all of the BI vendors’ mashups simply offer numerical analysis through an OLAP cube and plug a search bar alongside to search the report tags on a report or to search a separate store of textual content.  Attivio’s Active Intelligence Engine (AIE) indexes both unstructured content as well as transactional data and preserves the normalized form of the transactional data in the index.  In addition, our patent-pending technology allows you to join across text and data so that you can mashup numbers and text and discover patterns, correlations, and important new insights about your organization’s performance using whatever presentation layer you want to connect to AIE.

With all of this content indexed in one place, you can query and analyze in new and exciting ways.  Think about searching across all content from a search box and bringing back both numbers and text.  Imagine the search engine recommending the navigation you should follow based on the patterns it discovers across all of the content in your company (I’m starting to call this “guided ad hoc analysis” for lack of a better way to describe the convergence of guided analysis through facets or navigators plus the ability to wander through the data in an ad hoc, exploratory way since you don’t have to understand the navigation path before you ask the question). 

I challenge the Attivio community to help me figure out what we should call this category of mashup and to let me know if you think anyone else is doing anything smart with mashups in the BI / information access worlds today.  As far as I see it, no one else seems to be doing anything except window-dressing their silo of either transactional data or text content with a pretty widget from the other camp’s solution.  This doesn’t seem to be providing business value to anyone.  We at Attivio hope to change the approach and show some value in querying and analyzing all of your content and data in the same engine.

If you'd like to comment on this article, feel free to This e-mail address is being protected from spambots. You need JavaScript enabled to view it .