Sunday, March 13, 2011

Pat Kerr, the CFO of SHMsoft

There is a lot that can be said about Pat, that he is tenacious, persistent, fearless, down-to-earth or that he flies in the sky, but I think the most important is that he is so sure of himself that he is not afraid when people make fun of him.

As a case in point, he could not find any serous photo of himself. Here is his latest attempt.

Friday, March 11, 2011

New open source software for eDiscovery

I have created and published open source software for eDiscovery, called FreeEed. The project is hosted on GitHub here. The discussion group for it is here.

The software is in working state, but it is an early release, which follows the common open source "approach of commit early, commit often." At this time, I am looking for feedback on what the next incremental improvement steps can be.

The software has been tested in Ubuntu, but it may work in Windows. It works in local mode or on a cluster and is scalable: the same code will work on a cluster without any change.

Tuesday, March 8, 2011

Open source eDiscovery restarted

A couple weeks ago I restarted the FreeEed, an open source eDiscovery project. The initial announcement can be found here, but that was two years ago. Now I have changed everything, except for the project name.

The project now is on GitHub, which is the new black for developers. It combines complete processing and has no resemblance to the old version, which relied on Google Desktop search.

Moreover, it is intended to run locally, on a private Hadoop cluster, or on EC2 Amazon cloud. Thus, it can be as small or as big as you want.

Really, it is not at all hard to do this. Look, for example, at this company, talking about open source eDiscovery solutions, and you will see that the choice of components - Hadoop, Lucene, Mahout, Tika, Solr, Nutch - is pretty standard. With the road thus outlined, it was indeed strange that no one has done it so far - so there you, I have done it now.

In the past two weeks I was already able to accomplish about half the work. It is just a matter of combining the right libraries in the right way. Truth be told, the 90-10 rule applies here too: the most fun and usefulness are in the first 90% of the functionality, which takes 10% of the time. The rest are the details.

One can follow the project by joining the Google Group here, or on GitHub, where developers can have free accounts for easy cooperation.

Sunday, March 6, 2011

Armies of Expensive Lawyers, Replaced by Cheaper Software?

The article in the New York Times claims just that, software replacing lawyers. But, as many newspaper articles, it mixes a few things up.

Anyone who is reading this probably knows, the eDiscovery market is about 4B, and it is not shrinking. Of the usual costs of eDiscovery, lawyers get 75% and processing costs constitute only 25%. As this blog is frequent to point out, lawyers are creatures of habit more than others, and this habit is good law, so it is not going away any time soon, even in the presence of statistics that are pro-computer.

Then where are the authors of the article correct? In their praise of the new tools for analytical eDiscovery, even if not in their assessment of their impact. For internal use, these tools provide lawyers with definite advantages and help them find important leads. Whether the judge will accept document classification as privileged, if it is done by a computer review - this will require the change in attitude, the improvement in technology, and perhaps a review of applicable laws.

However, with the driving force coming from customers not willing to pay the high bills of what essentially is human search where automation would be welcome - all sides may eventually be forced to accept this. Why do we trust Google on search results? Because it is not humanly possible to read all relevant information by ourselves.

Art: Pierre Auguste Renoir - The Thinker Aka Seated Young Woman

Friday, February 25, 2011

View & Comment on Any Document

What does this sound like? - To me, sounds like legal review. And indeed, you can get a free account, upload your PDF or Word documents, and redact it to your heart's content.

You can also have some business license, with groups, security, and privileges, and, what's more, you can do your own application using their API - although this is still shaping. I think they have done the redacting part just right!

What is the history and popularity of the application?

Founded by three MIT graduates, and one MIT attendee who dropped out in order to help launch successful internet startups, Crocodoc's creators are the same people who brought WebNotes, a tool used by professionals and educators across the globe to manage research, to the web. Y Combinator, and other investors provide funding for Crocodoc. The company is based in Silicon Valley California, and also has offices in Boston... (from this review)

Art: Paul Cezanne - Uncle Dominique As A Lawyer

Sunday, November 14, 2010

Speaking in tongues


Giuseppe Caspar Mezzofanti (1774-1849) was keeper of the Vatican library and later a cardinal, but he’s best remembered for being a hyperpolyglot, a speaker of many languages.
How many? Estimates range from 24 (in 1805) to 114 (judged after his death). The true number probably lies somewhere in between, but it’s considerable–Byron called Mezzofanti “a monster of languages, the Briareus of parts of speech.”
A Russian traveler once asked Mezzofanti for a list of the dialects he had mastered, and the cardinal sent him the name of God in 56 languages. And Gregory XVI once arranged to have a polyglot group of students waylay him in the Vatican gardens: “[O]n a sudden, at a given signal, these youths grouped themselves for a moment on their knees before his Holiness, and then, quickly rising, addressed themselves to Mezzofanti, each in his own tongue, with such an abundance of words and such a volubility of tone, that, in the jargon of dialects, it was almost impossible to hear, much less to understand, them. But Mezzofanti did not shrink from the conflict. With the promptness and address which were peculiar to him, he took them up singly, and replied to each in his own language, with such spirit and elegance as to amaze them all.”
For another prodigious librarian, see Book Lover.

Thursday, September 2, 2010

An idea for a legal service

This service would create a social portrait of the defendant (or plaintiff, for that matter) and present it to the lawyer.

Why would this service be better than a paralegal, whom the lawyer can task with doing the research? At the paralegal's rate of $100/hour, a computerized service can compete on price, and outsourcing this may be too sensitive. In addition, the service could
  • Provide nicely-formatted, standard reports;
  • Search thoroughly, and not omit any of the hundreds of social sites;
  • Keep a history in the changes in the profile, which may be telling.
I remember a lawyer who found what he needed about the opposing side's defendant by discovering his profile on a Greek singles site. This was a unique achievement by that individual, but a service would make such feats dependable and reproducible. It can also collect defensible evidence in the process.

The suggested name for the service is "Watson."

Why is it OK to discuss this idea in the open? Because anyone can come up with this or a similar idea, and it is the implementation of it that would be a competitive secret. It needs to include
  • Scalable, cloud-based architecture, that can expand on demand and provide pay-only-for-resources-used charges;
  • Rules-based engine with dozens or hundreds of rules of research, based on the case and individual's profile;
  • Specialized text- and image-analytics,  such as image search, blog and articles understanding, etc.;
  • Of course, crawling that avoids being invasive and being blocked by search engines and the social sites.
Art: Sidney Paget - Nothing could be better, said Holmes, illustration from 'The Stockbrokers Clerk by Arthur Conan Doyle 1859-1930, published in Strand Magazine, March 1893