I am a PhD student in the Computer Science program at Indiana University. I started in the program in August, 2008. My research interests are in online social networks, tagging systems and web mining. I am a member of the NaN group, led by Professor Fil Menczer. Currently, I am an RA on the GiveALink project.
If you need to contact me, feel free to email me at dnikolov -at- indiana -dot- edu, or see me at my desk in INFO East 400.
Some of my present and past projects are described below.
Social Spam Detection, January 2010 - Present
Supervisor: Fil Menczer
We study ways to efficiently and correctly identify spammers among the users of the GiveALink social bookmarking system. The research involves developing a spam detection system incorporating a variety of features and machine learning techniques such as boosting and cascading classification, clustering and attribute selection.
Feature Extraction from Fingerprint Examination Data, July 2010 - August 2011
Supervisor: Thomas Busey
We analyzed eyetracking fixation data generated in experiments with expert fingerprint examiners and novices. We used independent component analysis (ICA) and information theory to identify salient regions in fingerprints. In addition, we attempted to use a variety of clustering and classification algorithms to discriminate experts from novices accoring to their fixation patterns. For more information on Professor Busey's research on the topic, look here, here, here, here and here.
Approximate Functional Dependency Mining, September 2009 - June 2010
Supervisors: Ed Robertson, Jeremy Engle
We studied algorithms for discovering approximate functional dependencies in databases. Together with Jeremy Engle, we worked on the implementation and evaluation of a boundary following AFD mining algorithm which cuts both space and time costs compared to other state-of-the-art algorithms.
Recommender Systems, January 2009 - May 2009
During my first year at IU, I worked on a couple projects that dealt with recommender systems.
In Fil Menczer's web mining class, I worked together with DongInn Kim on the Netflix Challenge. We used a collaborative filtering approach, combined with content-based classification based on movie data that we collected from IMDB. In the end, our approach proved to be a bit too inefficient for the scale of the problem. You can read about it in our report, or check out the blog we used for status updates during the project. The Netflix Challenge has since been won.
In David Leake's symbolic learning class, I designed a case-based news recommender system enhanced with commonsense knowledge from the Open Mind Commonsense Project. The code currently doesn't interface well with Open Mind, and I've been meaning to revise it or rewrite it in Python whenever I have more free time. Read about the project in my report. Feel free to contact me if you are interested in the code.
Primary instructor: Gregory Rawlins
Professor Rawlins' class is a software engineering class that introduces students to design patterns, good coding and software engineering practices, teamwork, and the use of bug-tracking, code coverage and continuous integration tools. As an AI, I was responsible for introducing the students to the tools (Eclipse, CVS, SVN, Jira, Emma, Bamboo, JUnit, log4j, Confluence), maintaining the tools infrastructure, conducting code review, meeting with students to discuss design questions, helping in administrative matters such as grading and creating team schedules for the class assignments. Enrollment in the class is usually around 20 people, most of them graduate students.
Primary instructor: Charles Pope
A110 is an intro to computing class for non-majors. Students learn about creating web pages with HTML and CSS, and working with data and tools such as Excel, Access and Word. I taught 2-3 labs a week to a total of about 50-70 students depending on semester. My duties included presenting new material in lab, administering tests, grading homework assignments and tests, and holding office hours.
- Q550: Models in Cognitive Science, Michael Jones
- I690: Mathematical Methods in Complex Systems, Alessandro Flammini
- I586: Artificial Life, Larry Yaeger
- B501: Theory of Computing, Daniel Leivant
- B673: Advanced Scientific Computing, Randall Bramley
- B652: Computer Models of Symbolic Learning, David Leake
- B656: Web Mining, Fil Menczer
- Q540: Philosophical Foundations of Cognitive Science, Colin Allen