Marios Kokkodis
PhD Student
Information Systems Group
44 W 4th Street,
NYU Stern, KMC 8-185
New York -
NY 100012
Phone: +1-(212)-998-0396
Email: mkokkodi[at]stern.nyu.edu
Advisor
Collaborators
Colleagues
Friends
Profile
I am a third year PhD candidate in Information Systems at NYU Stern school of
business.
My research interests include user reputation and quality assessment in online
labor markets, text and data mining and paid
crowdsourcing.
In 2007, I graduated from the
ECE school of National Technical University of Athens (NTUA), Greece.
I further hold an M.Sc. in Computer Science from
the University of California at Riverside.
Research
My research deals with problems that arise in
online labor markets (such as
oDesk).
Specifically, I am more interested in user repuatation and quality assessment in such settings.
I am further working on matching and recommendation problems in these environments.
Recent Publications
-
Kokkodis, Marios and Ipeirotis, Panos. "Have you done anything like that?
Predicting user performance using inter-category reputation." In WSDM 2013
-
Kokkodis, Marios. "Learning from positive and unlabeled amazon reviews: towards identifying trustworthy reviewers." Proceedings of the 21st international conference companion on World Wide Web. ACM, 2012.
Research - Abstract
On-line labor markets such as oDesk and Amazon Mechani- cal
Turk have been growing in importance over the last few years:
employers post tasks on which contractors work on- line. In
parallel, on-line reputation systems (such as Ama- zon product
reviews) have great impact on everyday on-line trading. In these
settings, reputation mechanisms play a very important role, as they
allow workers/reviewers to sig- nal their current quality which is
often predictive of the qual- ity of their future performance. In
this work, we propose a model to predict the performance of a
worker/reviewer based on his category-specific estimated qualities.
In particular, our model assumes that these qualities are
category-specific, latent and not directly observable, but are
reflected into a set of other measurable characteristics such as
ratings, which are often missing for new categories. We apply and
evalu- ate our model on a large data corpus of Amazon reviews. In
this specific setting, our model based predictions show improved
accuracy compared to baseline predictions based on past average
reviewer quality. Our results indicate that Amazon reviewerÕs
reputation is transferable across product categories.
Back
Misc
In my free time I have developed two applications:
- A Simple Transaction Sales Point which you can find here
- Rootfinder: a program estimates the roots of quadratic,
cubic, quartic and quintic equations. You can find it here.
Back in 2003, we had a band (called Delear). Here you can
find two of our songs:
Teaching
Fall 2011
UB.1.1-004.1-006 : Information Technology in Business and
Society.
Spring 2011
Fall 2012
Working Experience
06/12 - 09/12 : Data Scientist Intern at oDesk.
06/11 - 08/11 : Data Scientist Intern at oDesk.
09/08 - 08/10 : Research assistant and member of the
Networking lab at UC Riverside. Research on characterising the
network level behavior of spamming botnets.
01/08 - 07/08 : Military Service. Served as a soldier in
the National Guard of Cyprus.
03/07 - 09/08 : Telecommunications Laboratory, National
Technical University of Athens. Software Developer (Java,JSF,
ADF,Oracle 10g). Participate in the design and development of an
interoperability information system in the field of health.
Contact
Email : {x@stern.nyu.edu | x=mkokkodi}
Phone : +1-(212)-998-0396
Address : Room 8-185, Department of IOMS, Stern School of
Business, 44 W 4th New York - NY 10012
