Whit Andrews
VP Distinguished Analyst 10 years at Gartner 14 years IT industry
Whit Andrews is a vice president and distinguished analyst in Gartner Research. Mr. Andrews covers information access technologies, including enterprise search, and maintains the information access technology Magic Quadrant with Rita Knox. He is also a significant contributor to e-discovery… .Read Full Bio
I’m looking forward to LegalTech next week. They’re looking for video questions for the e-discovery town hall. Here’s the outtake from my question:
I’ll have the rest of the question at the YouTube site. Those of you who can’t wait can see my video on YouTube before it gets added…but why would you? There’s better in the group, I’m sure.
Your assistant is great, and the question is a good one too. The wide disparity in performance among the tests you cite cuts to the heart of what’s wrong with TREC and the Cranfield model. Judging the performance of a retrieval strategy based on a one-shot query model, applied uniformly across all queries, is absurd. Of course, our legal system is often absurd too, and I wouldn’t want to argue about the flaws of the Cranfield model with a judge. That’s one of the reasons we’re holding this year’s HCIR workshop in DC; we’re hoping some of our ideas rub off on the intellectually recharged federal government.
The feds are hot on early case assessment — I’ve seen that in customer references because they have such limited resources that they need to suss out what’s going to go right in a case (and wrong) before they commit scarce resources.
And (didn’t finish the thought) once you start thinking early case assessment, it is a short walk down the boardwalk to sophisticated multi-query models to gain deeper insights.
Whit Andrews
3 responses so far ↓
1 Daniel Tunkelang January 29, 2009 at 12:22 am
Your assistant is great, and the question is a good one too. The wide disparity in performance among the tests you cite cuts to the heart of what’s wrong with TREC and the Cranfield model. Judging the performance of a retrieval strategy based on a one-shot query model, applied uniformly across all queries, is absurd. Of course, our legal system is often absurd too, and I wouldn’t want to argue about the flaws of the Cranfield model with a judge. That’s one of the reasons we’re holding this year’s HCIR workshop in DC; we’re hoping some of our ideas rub off on the intellectually recharged federal government.
2 Whit Andrews January 29, 2009 at 9:52 am
The feds are hot on early case assessment — I’ve seen that in customer references because they have such limited resources that they need to suss out what’s going to go right in a case (and wrong) before they commit scarce resources.
3 Whit Andrews January 29, 2009 at 9:53 am
And (didn’t finish the thought) once you start thinking early case assessment, it is a short walk down the boardwalk to sophisticated multi-query models to gain deeper insights.