DolphinSearch's roots go back over a decade, when Herbert L. Roitblat began studying the dolphin's brain under a contract with the U.S. Navy. Today, Roitblat is both a vice president of DolphinSearch and its chief scientist. He says that insights into the mind of the dolphin are helpful in understanding how humans process information and solve problems. He describes DolphinSearch as "biomimetic." He explains: "[DolphinSearch] employs neural networks similar to those used to model the dolphin biosonar performance and language understanding to read and ‘understand' an organization's documents. It is entirely self-organized. It learns the meanings of the words from the context in which they are used."

Tyler Howes, LexisNexis' senior director of product development and planning, explains how DolphinSearch will fit into the LexisNexis portal: "We partnered with Plumtree to develop our legal portal product. We want to provide an array of tools for use within a practice. We see DolphinSearch as a best-of-breed technology that greatly enhances our portal product.

"By adding DolphinSearch, we allow a firm that subscribes to our portal to search their collections of documents stored in whatever document management system they may use (PCDOCS, iManage, etc.)," said Howes. "They also can use DolphinSearch to index and manage internal documents. In many cases, a firm's e-mail repository is bigger than its DMF [document management facility] content."

I asked why a firm might get DolphinSearch from LexisNexis instead of buying directly from the vendor. Howes said: "The LexisNexis taxonomy is integrated into our DolphinSearch part of the portal. Thus if a legal researcher wishes to browse through familiar LexisNexis categories, he or she can do so. LexisNexis supports DolphinSearch as if it were our own product. We offer 24/7/365 support to our customers. They can have confidence that whenever they need support using DolphinSearch or any features of our portal product, they'll get it."

DolphinSearch CEO Andy Kraftsow says that the typical potential customer for DolphinSearch is likely to be a large law firm with a large collection of documents. "A small firm with, say, only a few thousand documents, may feel they don't need DolphinSearch. A large firm with millions of documents may find it hard to live without it. Such law firms may prefer dealing with large technology providers, and many of them already have relationships with LexisNexis."

Asked how the DolphinSearch announcement relates to the recent LexisNexis alliance with iPhrase [see the October 29, 2001 NewsBreak at http://newsbreaks.infotoday.com/nbreader.asp?ArticleID=17472], Howes said that these were really two separate undertakings. iPhrase provides a natural language mechanism for searching a collection of known answers to questions—sort of a searchable FAQ. DolphinSearch provides a way to sift through huge collections of text.

Michael Kraft, an attorney and principal of the legal consulting firm Kraft, Kennedy & Lesser, Inc. (KKL), said, "DolphinSearch is utterly amazing. There's just nothing like it." New York-based KKL provides technical services to some of the largest law firms in the U.S., and has recommended DolphinSearch for some of its clients who are working on major cases.

Kraft said: "When a law firm is working on a big case today, they find that they must work with huge collections of documents. It could be 500,000 documents or it could be millions. Suppose the firm is analyzing that large a collection of evidence in order to build its case. In order to make use of the documents, you have to understand what is in them. Historically, to do that you would hire a roomful of knowledge workers (such as paralegals) and have them ‘code' what each document is about. There is both ‘objective coding' for metadata and more subjective coding in which each document is analyzed and the coder tries to place each document in appropriate categories.

"There are lots of problems with the coding approach: It is slow, it is expensive, it doesn't do a good job of helping you find the most relevant documents to help your case. How a document is coded may not match the way an attorney later does a search. You always wonder what important pieces of evidence you missed entirely.

"DolphinSearch changes all of that. Instead of relying on human coders, you feed the documents to the DolphinSearch database, and it analyzes the documents for meaning. Instead of doing simple Boolean searching, your firm can navigate through the documents for meaning. You can perform highly effective ‘relevancy triage' because you have confidence that your search has brought the most relevant documents to the top of the list."

Kraft notes that during the process of litigating a case, a new issue will often arise, making it necessary to reanalyze the huge collection of documents in light of the new perspective. "When this happens, you are pressed for time to reanalyze your evidence from this new perspective. That is very expensive and labor intensive even with a team of knowledge workers. With DolphinSearch, all you have to do is re-examine the existing database with the new issue in mind."

DolphinSearch's Kraftsow said that often one litigating party will request a particular document, and the opposing side will wonder why. DolphinSearch can help analyze why the request was made in order to prepare for anticipated uses of the evidence.

Bruce S. Markowitz, automated practice support manager for the Washington, D.C., law firm Mayer Brown & Platt, has not yet used DolphinSearch with LexisNexis, but has applied the technology to some cases for his firm. "The product is unique—really different," he said. "It does things that you just can't do with other tools."
 
Markowitz said DolphinSearch is useful for analyzing massive amounts of text. "In one case we had lots of old e-mail messages from old systems. DolphinSearch deciphered the old content and processed it in a way to make it searchable. It helps us effectively code content by issue."

The legal market isn't the only arena for DolphinSearch. The company has teamed with Sun Microsystems on some projects. The two companies are working on applying the technology to match television ads with households that desire particular products. Since September 11, DolphinSearch has been working with federal agencies that need to analyze and sift through massive amounts of textual information.

Kraftsow said: "We read text with more or less the understanding of a librarian. We and our customers have realized for a long time that Boolean is just too simple a technology to accomplish what people need to do." He notes that DolphinSearch is completely language independent. "It doesn't even know it is working with a particular language. We can feed any language that can be expressed in Unicode to DolphinSearch, and with no pre-processing it can analyze documents. The language could be Russian, or Farsi, or whatever."

I asked Howes where he sees linguistic analysis and natural language tools going in the future of LexisNexis products. While declining to give specifics, he said: "We are always looking at ways to improve the searching experience in our products. We will see more breakthroughs beyond Boolean."

A firm can employ DolphinSearch as an application service provider (ASP) model, or can choose to install a turnkey DolphinSearch hardware solution. Kraft said: "DolphinSearch works well either as an ASP model or using their server in-house. You just plug the box into your network, give it an IP address, and tell it what documents it has permission to index. It can read a wide variety of document formats, and it can analyze and index information inside the popular document-management products law firms use today."