back home... |
University of Michigan     (September 2009 - present) I am now an
I'm heading back to grad school to get my PhD.... and I'm done... :)
As of February 2001 I'm at the HP Information Dynamics Laboratory. Among other things, I am working on a system called SHOCK (Social Harvesting of Community Knowledge). You can see slides from a talk I gave here: HTML
or (animated)
PPT Xerox Palo Alto Research Center     (July 1998 - February 2001) I joined PARC in July of 1998 after graduating from MIT. I started my
work in the Quantitative Content Analysis
(QCA) area of the Information Sciences and Technologies Lab. While in QCA I developed a the Pipes toolkit for composing new Information Retrieval systems. The system, written
entirely in Java, allowed you to rapidly prototype new and interesting
IR systems (including pre-processing, indexing, and retrieval).
Laboratory for Computer Science & Artificial Intelligence Lab, MIT Project Haystack     (January 1996 - May 1998)
When I was at MIT I worked on a Personal Information Retrieval system called
Haystack. With Haystack you can index everything on your machine (although
that's probably a little optimistic). You can annotate the resulting data
with your own comments like, "this is a really interesting paper." Soon,
we'll have methods in place to do queries like, "get all postscript files,
that I thought were 'interesting' and that have the word thingamajig that
were written between 10/1/97 and 10/5/97." Well, it might not be as "natural"
as that, but it'll be close. This will be an improvement on both databases
(which have no notion of fuzzy, i.e. they can't do useful things like ranking
responses), and IR systems (which have no notion of exact, i.e. they can't
tell you if something was written between two dates, just if the document
contains a specific date).
The system was intended to learn gradually from your queries,
and provide you with the facility query other people's Haystacks. Learning is provided by doing things like tracing your queries as you search for something (first you search for "apple" and the "apple computers" and then "apple newton pda"). When you find what you are looking for the full trace is added to the object allowing you to jump to the end whenever you use any query in the path.
For my work on Haystack I was awarded the Anna
Pogosyants UROP (Undergraduate Research Opportunities) Award. I am
very honored to have been nominated for the award, let alone winning it.
Laboratory for Computer Science, MIT Library 2000     (October 1993 - November 1995) "Library 2000 is a computer systems research project that is exploring
the implications of large-scale on-line storage using the future electronic
library as an example."
During my time at Library 2000 I worked on a number of projects including automatic indexing and conversion of various bibliographic records through a customized WAIS client.
My main efforts for Library 2000 were centered around an
algorithim for producing relevance matching between two documents (or
a query and a document) using repeated boolean searches. Basically
what we have accomplished is a way to create links on the fly so that
users browsing through a document, upon finding something of interest
can merely click and have our system find the closest match within the
library. In 1995 we presented a paper on this work, and in 1998 I
received a patent on the algorithm.
Non-research work Agency.com     (November 1995 - May 1997)
For a while I consulted for Agency.com, doing various projects for clients such as Metlife, American Express, GTE, Hitachi, and British Airways.
In 1995 I co-founded an on-demand CD publishing company. The idea was
that users could come into the site, deposit a bunch of URLs, we would
retrieve the data, burn it onto a CD, and send it back to the user. I
designed and implemented most of the technical aspects of the system.
It was an interesting, if short, first foray into the world of
start-ups.
I worked as both a consultant and intern for Trilogy/pcOrder where I developed an automated testing framework as well as one of the first prototypes of pcOrder's web interface.
This server which I
created for fun holds a web interface to all the MIT floorplans. It's
still used by a lot of MIT people. The system is pretty neat
(shameless plug). If you're in the MIT network you can take a look at
it here (guessing this is no longer up). |
|