Enterprise -> Technology
By: Roger Whitehead, Associate Analyst - Collaboration, Bloor Research
Published: 17th January 2008
Copyright Bloor Research © 2008
Modern information management tools should help the user answer these and similar questions about the context as well as the content of pieces of data. So says Dr Jan Puzicha, co-founder and Chief Technology Officer of Recommind. His company's products have been designed to do so from the first.
Recommind is based in San Francisco, USA, with offices in New York, Chicago, Atlanta, London and Bonn. Its software is based on work on machine learning that Dr Puzicha did at the Massachusetts Institute of Technology and the University of California, Berkeley, both in the USA. The result of his research is something imposingly entitled Probabilistic Latent Semantic Analysis (PLSA), a patented technology.
PLSA makes the company's software especially suitable for extracting concepts from large bodies (corpora) of data, says Craig Carpenter, the company's Vice President of Marketing and Business Development. As well as the industry-standard keyword searching, it can examine data for explicit and implicit links to other data, regardless of the language or subject.
Recommind's three main products all use PLSA and other proprietary techniques and are variants of its MindServer software platform. They are, respectively, for enterprise search, automated filing of emails ("Decisiv") and legal discovery ("Axcelerate").
The enterprise search software is the company's most-used and longest-used product. (This is referred to as a knowledge management product in legal markets, for historical reasons.) An optional module called Projects & Expertise links information across repositories to deliver information about relevant people and projects at the same time as data ‘hits'.
The Decisiv filing product works with Microsoft Outlook, automatically classifying and archiving email messages. Recommind says it makes significant savings in disk space needed, while giving speedier and more accurate access to those archives.
Axcelerate is the company's e-discovery product. Discovery is the process of presenting to a court the relevant files and records that a law firm or legal department will offer as evidence when making a case. Lawyers use Axcelerate to track down and retrieve material about a particular topic (‘matter') on which there is litigation.
The use of computers for this task is well advanced in the USA. It is becoming accepted in the United Kingdom, which has a conceptually similar legal system, and is starting to be used in mainland Europe, which has several different legal codes.
‘Discovering' (that is, uncovering) legal documents for a case makes up much of the time and cost involved. Doing so electronically gives greater speed and accuracy to the review process. Like any electronic method, it also allows the work to be done in—and the results sent to—dispersed locations.
Several of Recommind's larger legal clients have hundreds or thousands of lawyers, in offices around the world. Axcelerate enables them to automatically pre-code information and distribute an organised set of documents to those lawyers for review. The sources analysed can include scanned files, databases, Web pages and email messages.
As an adjunct to Axcelerate, Recommind has a product called Litigation Hold. This uses the same automatic analysis and categorisation tools but is applied before litigation begins. Prudence, and legal requirements in the USA and elsewhere, demand that organisations identify potentially actionable documents as they are filed, not afterwards. Litigation Hold helps with that task.
In November 2007, Recommind announced version 5.1 of its MindServer platform. The major change with this release is its ability to handle federated searching. This allows it to integrate the results of its own searches with those conducted on and by other search tools. Many larger organisations have more than one make or model of search software in use, sometimes because of a corporate take-over. Federation allows these tools to work in harness.
MindServer 5.1 can also search external online resources, such as EDGAR Online, European Union services, London Stock Exchange and Google. Users can specify how they want those searches conducted—in one go or asynchronously—and the results presented.
Unsurprisingly, most of Recommind's customers are in the legal market. It also has a strong following in media companies, such as Bertelsmann, and pharmaceutical companies, including Novartis. A typical customer would have between 1,000 and 3,000 users, most of them being heavy users of the system, such as researchers.
Recommind mainly sells its products direct. It has no resellers but is developing a stronger link with systems integrators. About 70 per cent of Recommind's business comes from the USA. Of the rest, some comes from Germany and Austria but the bulk comes from Britain, where the company is growing fast, mainly in the legal sector.
The company is self-financed and overall turnover is growing at between 100 per cent and 150 per cent a year. It employs about 80 people.
The Bloor view
Recommind is, in the jargon, mainly a niche player. The company recognises that much of its success has been because of its narrow focus. However, it is in a large niche. The world always needs lawyers, they increasingly operate multinationally and they need electronic aids for their work.
Whether there is scope in that niche to become a major software house is doubtful. However, the MindServer platform, in its various instantiations, is something any firm or department looking for an electronic discovery system should put on its long list, at the least.
Posted: 17th January 2008 | By Dolin duval :
the idea of federated semantic sourcing is a good one. We have tried to implement federated search using keywords with little success. The amount of data returned because of the lack of decent filteringwith keywords makes things unmanageable. We have looked at recommind and are also looking at a smaller company called PureDiscovery. This will be a growing field as the web continues to expand more broadly and deeply.
Posted: 17th January 2008 | By Roger Whitehead (Author):
I agree, Dolin. As more organisations or parts of them adopt search software, the greater the likelihood of different makes being present, hence the greater need for federating results.
It's not a perfect answer, though, as the integrated results are only as reliable as the worst individual set.
The messages above were all contributed by IT-Director.com readers. Whilst we take care to remove any posts deemed inappropriate, we can take no responsibility for these comments. If you would like a comment removed please contact our editorial team.
We automatically stop accepting comments 180 days after a post is published. If you would like to know more about this subject, please contact us and we'll try to help.
Published by: IT Analysis Communications Ltd.
T: +44 (0)190 888 0760 | F: +44 (0)190 888 0761