Migrated from eDJGroupInc.com. Author: Babs Deacon. Published: 2013-03-28 05:00:15Format, images and links may no longer function correctly. I interviewed Dr. Roitblat last month and I started out my questions with, “what’s with all the technology named for residents of SeaWorld®?” But, quickly changed my inquiry to, “what’s a nice bio-psycho-acoustician doing in an industry like eDiscovery?” And I’m glad I started this research as early as I did in order to coincide with the announcement of Dr. Roitblat’s fifth patent: US Patent No. 8,401,841 related to “language modeling”.
Legal Technology veterans have known about Herb since he dove into the Knowledge Management world with DolphinSearch™ in early 2000. And we have followed his move from KM to eDiscovery and the debut of his newest offering, OrcaTec. I wanted to know the how’s and why’s of this creative journey and if I was missing some inner truth about analytics and sea creatures?
Dr. Roitblat started in academia at Columbia University doing psych research on easily acquired NYC subjects: pigeons and rats. At the same time, he worked with a group of Columbia University librarians to take their catalog on line. He was shortly recruited by two different universities with a difficult choice: move to snow-bound Syracuse, New York to continue work in library automation, or do research in paradise at the University of Hawaii with dolphins. It was obviously a tough call, but Herb moved to Hawaii where he worked at both the university’s dolphin lab and at the local Navy base.
He and his colleagues were studying attention in dolphins by counting how many sonar clicks they directed at a particular object. The research required figuring out a way to automate the click counting (they had been doing it manually by listening to audio playback) and building neural net models to analyze the relationship between clicks and how the dolphins remembered objects they had pinged previously.
Dr. Roitblat had expanded his research into second language acquisition when Andy Kraftsow, now Chief Scientist at RenewData, asked him to improve the memorability of presentations given to wholesale travel resellers. Following that project, Andy was trying to build a search engine for Seventeen Magazine readers for the pre-Google Internet. Herb knew he could improve the search capabilities and the idea ultimately became DolphinSearch.
Herb and Andy decided to deploy DolphinSearch in the Knowledge Management arena because they realized that employees needed to run searches on their own organization’s documents. Knowledge Management was the flavor of the month at the turn of the millennium, and in the law firm world “brief banks” were getting a lot of attention.
Law firms were beginning to wake up to the idea that they needed a more streamlined, cost efficient way to find and repurpose their own work product, but many firms weren’t satisfied with the search capabilities of their document management systems or they were bogged down trying to build attorney curated brief banks that required review and categorization of every document in the collection. DolphinSearch™ offered a more natural approach to searching an organization’s knowledge stores by understanding language patterns and prioritizing results.
Unfortunately, DolphinSearch may have been ahead of its time or pointed at the wrong market. Herb and his colleagues realized that eDiscovery might be more of a fit because “you have to do it.” They got pulled into the eDiscovery world when customers started to ask them to tiff documents.
Herb recalls that concept search functionality took about seven years before it became a routine request in RFPs. He believes the economic downturn caused a diminution in the number of new litigations, and a lowering of eDiscovery volumes because parties couldn’t afford mountains of eDiscovery. He also believes that the poor employment market for attorneys kept the cost of human review at affordable levels.
Herb and his colleagues, including Brian Golbére, took on the hurdle of analytics, once again, in 2006 with a completely new technology, the Orcatec Document Decisioning Suite™. Dr. Roitblat believes the market is now more accepting of the use of analytics in eDiscovery because people are becoming very comfortable with analytics across the spectrum of their business and personal lives through the use of SPAM filters, data mining/taste predicting capabilities in sites such as Amazon, Pandora and Netflix and even the success of IBM’s Watson on Jeopardy.
In eDiscovery, specifically, he believes that the work of TREC has shown that the traditional review and culling methods, human review and search terms weren’t as accurate as clients assumed. He also believes that the increasing eDiscovery volumes were surpassing the ability to find enough good document reviewers.
Dr. Roitblat is encouraged by the increasing use of analytics in predictive coding and other types of CAR; and is proud to point to several instances, including published cases where OrcaTec has been deployed. His goal for analytics in the near future is the pro-active use of OrcaTec integrated with archiving tools. Analytics would manage data “on the way in” to an archive, based on whether or not it needs to be retained as a business record or for other reasons. He believes that Information Governance can be the area where analytics will have the most impact in saving money for corporations but that “we are a little bit ahead of the curve.”
eDiscoveryJournal Contributor and Director of Strategic Consulting – Babs Deacon
Don’t miss Babs Deacon’s upcoming Selecting eDiscovery Technology & Service Solutions Boot Camps in New York and Austin. To learn more, click here!