Search at Work
On the enterprise side, search continues to be integrated more deeply into technical solutions. As Craig Carpenter, VP of marketing for Recommind, points out, "We used to call ourselves an enterprise search company. Then we began coming out with technical solutions that incorporated search for vertical markets, including e-discovery." The company now considers itself a provider of "search-powered information risk management software," with products such as MindServer Search and MindServer Categorization providing automated concept search solutions for the legal profession, among other vertical markets.
The stakes for accurate and timely search in the legal industry are particularly high; the cost associated with information discovery can be steep, and the failure to find the right piece of content critical to a case can mean the difference between winning and losing a case.
kCura offers an online content review application for the e-discovery market called Relativity that pushes search beyond analytics and straight into workflow. The Relativity platform uses
a combination of concept search, categorization, and clustering based on conceptual similarity, to cut the time it takes between starting a search and finding the right documents. Traditionally, this was done at the beginning of a case using simple keyword search to start sifting through the thousands of documents that might be pertinent to a case, but it was time-consuming and prone to shortcomings. As Adi Elliott, kCura's director of marketing says, "Humans don't necessarily speak in keywords. But technology is finally developing such that concept search is possible."
Relativity search results are presented in such a way that an attorney looking at a results list of emails clustered based on conceptual similarity can quickly determine how germane they are to
the case. "If they're of less relevance," explains Elliott, "the attorney can just right-click to assign a paralegal to do the review and can move on to more relevant documents."
Bruce Furukawa is an attorney with Severson & Werson, a mid-sized Bay Area law firm. In 2008 the firm had a client, a major national bank, which was sued by its recently acquired mortgage lending company. The case hinged on the interpretation of a term in the purchase contract related to the valuation of certain types of subprime loans involved in the securitization process. Furukawa's team needed an efficient means of reviewing 80,000 email messages in a matter of days. "Initially, I budgeted for the case assuming we'd process 400 documents a day, giving each document one of 15 issue codes," Furukawa says.
Using Relativity's concept analytics, with the help of integration partner Evolve Discovery, the process took half the time. "It appears to almost think for us," Furukawa says. "We ended up processing 800 documents a day with 99% accuracy." He also likes that the platform is hosted so that he can scale up or down depending on the work at hand.
Workflow integration of search was the goal when Dow Jones introduced its Media Relations Manager (MRM) platform in November 2009. The product is a news-enabled media database and contact management tool that helps communications professionals pinpoint and engage the journalists and bloggers most likely to be interested in their story. Martin Murtland, VP and managing director for Dow Jones, says, "We were trying to understand how we could get more value of our content by understanding user workflow and addressing their pain points."
MRM users can input basic text search terms, for example, "operating systems" and "Microsoft," for a story they'd like to pitch to generate a targeted list of journalists and bloggers who cover the topics. From that point they can drill down into the journalists' profiles to see articles they've written, bar charts showing companies and industries by coverage frequency, and keyword clouds to help focus the pitch to the individual writer. An email template can be customized or a briefing book for an executive planning a meeting with a reporter can be generated within a few keystrokes from where the original search term was typed.