Structured Search

Written By:
Content Copyright © 2007 Bloor. All Rights Reserved.

The ability to use search technology against structured data is a rapidly emerging sector. First, what it is it?

Structured Search is the ability to use a search-based approach to analysing structured data. For example, to search against an OLAP cube rather than slice and dice it. The proponents of structured search claim that it is much more intuitive (certainly, we all know how to do it) than the graphical dashboards and other stuff than you get with conventional business intelligence solutions. As a result, they claim (and I am not going to dispute this) that structured search provides the answer to the democratisation of BI: making BI available, and useful, to the masses. Moreover, one of the best things about structured search is that it is very easy (much easier than with conventional BI tools) to combine it with unstructured search.

The first structured search product that I saw was something over a year ago, from CopperEye, with its Greenwich product. This puts a search front-end onto structured (flat file) data that is automatically indexed by CopperEye as it is loaded into the system. The product is targeted at the telco, email and similar markets where large amounts of data need to be stored for archival purposes but which you still need to be able to search against from time to time.

More recently, at the end of January, FAST introduced its Adaptive Information Warehouse. This doesn’t just provide indexing on structured data and structured search but also data cleansing (using a linguistic approach), dashboards and so on. In other words, it is much more than structured search and it is, in fact, intended to represent a part of your data warehousing infrastructure alongside conventional environments such as Oracle, IBM or, potentially, Netezza.

Even more recently, I have had a briefing from Ardentia Search, which is a spin-off from the UK-based Ardentia. This is specifically a structured search tool, designed to run against OLAP cubes (and Lotus Domino databases—an important market, ignored by most other vendors). The way that it works is that you index your cube (actually, deciding what to index is probably the most important part of the installation process) and then you can search, filter and so on against the data therein. As you would expect, there is built-in security, you can use Venn diagrams for segmentation purposes, there are Boolean operators available for selecting data, a plug-in to Microsoft Excel, and so on. In the next release (due next month) there will be charting options, the ability to create a cube on the fly and support for metrics and measures.

None of these products are new: all of the companies mentioned (and there are a number of others) have been working on indexing structured data for some years but it is only now that the market is clarifying (though it still has some way to go). I expect to see significant growth in this market and, while FAST will be addressing the warehousing arena, Ardentia Search is well-placed in the wider market as an add-on to existing systems: it is the preferred partner of Ardentia for healthcare (where the company has a particularly strong reputation), and it offers a software as a service option for users, which should prove attractive to many companies.