Thomson Reuters, Legal, sets new precedent for search performance.
by Tara Swords, May 2010
In 1872 a traveling salesman and his brother started a family business in St. Paul, Minnesota. The goal of West Publishing was humble when viewed through today’s lens but ambitious at the time: to make the latest legal opinions and summaries available to attorneys practicing in Minnesota and Wisconsin. After all, court decisions are guided by established legal precedent, so attorneys who don’t have the latest court opinions could waste their time arguing cases based on outdated precedents. Almost 140 years later, it’s easy to imagine that John and Horatio West would be awed to see their fledgling company as the cornerstone of a US$3.6 billion legal business with more than 14,300 employees and one of the best-known brands in the U.S. legal industry: Westlaw.
Launched in 1975 by West two decades prior to that company’s acquisition by Thomson Reuters, Legal, Westlaw is an online legal research service that contains more than 32,000 databases packed full of cases, statutes, synopses, treatises, best practices, news articles, and public records. Its customers include attorneys and information professionals at firms of all sizes, individual practitioners, corporate counsel, government lawyers, and law schools.
Westlaw and its datacenters are a major part of the Thomson Reuters Professional division, where Rick King is chief technology officer. “This is the database that is the authoritative version of the law in the United States,” says King of Westlaw. “Courts, lawyers, defendants, and prosecutors all rely on it.”
It’s difficult to overstate the industry’s familiarity with Westlaw. As one practicing attorney says, everyone in the legal field knows it.
“Lawyers constantly use it,” says Robert J. Ambrogi, a Rockport, Massachusetts, attorney who uses Westlaw often. “Whenever you’re facing a legal issue, from a new client walking through the door to drafting a new document, you need to know the most current law and what the courts have said about it. It’s a matter of researching not just the historical law but what’s the latest. Have there been any cases in the last day or week or month to change my understanding of this issue?”
A million times a day, Westlaw provides those answers.
A Major Overhaul
About five years ago, the team at Thomson Reuters, Legal, began dreaming up the next iteration of its Westlaw application: WestlawNext. The Web had changed the nature of search and was itself evolving every day. Although Westlaw claims a healthy market share lead on its main competitor, decision-makers at the company knew that it would be foolish to rest on their laurels.
“It really was an excellent opportunity to clean up the stack and optimize around Web 2.0 and where the internet is going in the future,” says Cary Felbab, vice president of new product technology at Thomson Reuters, Legal. “We wanted to deliver the performance and capabilities that our customers are going to demand from us.”
The team at Thomson Reuters, Legal, would eventually face the challenge of designing the right hardware and software architectures to power this massive project. But the first challenge was fundamental: creating a search algorithm that was as close to a mind reader as possible.
With the advent of Google, users have come to expect speedy, accurate searches. But compared to the typical internet search, legal research is a different animal. Felbab offers an example: Imagine that a fan wants to know Madonna’s birthday. Thousands of Websites have that data, and a search engine needs to find only one of them. But legal research, Felbab says, is like trying to put together a puzzle without knowing what the finished picture will look like.
“You don’t know how big the puzzle is, you don’t know what all the pieces are, and there often isn’t a single source that answers the entire question,” Felbab says. “Often the challenge for users is ‘how do I know if I have everything I need?’”
Even before formal work began on architecting WestlawNext, the research and development team was at work creating a highly intelligent, proprietary algorithm—elements of which the company began patenting in 2004. The team enlisted practicing attorneys to spend several hours researching legal issues and identifying the documents that were most relevant. Then the researchers ran their latest algorithm to see whether a user could produce in five minutes the same work the attorneys had previously found in five hours. Research scientists and software developers tweaked the code over and over again, until the desired documents were among the first search results returned.
In addition to mastering the search algorithm, the team created a sleeker, more modern interface that would appeal to today’s users. Other features improve the workflow and help customers keep better track of what they’ve found and how it might affect their work product.
For example, not only does WestlawNext allow users to set up folders where they can drag and store documents relevant to their work, but the system continually monitors those documents and alerts users to changes in the legal status of the underlying cases and whether they still are “good law.”