Research

We have been working on forum spam mitigation. We have investigated its prevalence on the Web and examined various aspects of the forum spam. Based on the analysis, we have propsed our forum spam mitigation. Our paper, "Prevalence and Mitigation of Forum Spamming" would be appeared in IEEE INFOCOM 2011.

We also have been analyzing spam URLs to extract important features which can be used for their detection. In particular, we explore the strucrual characteristics of URL-based link graphs. Its preliminary results are under submission.

In addition, we have surveyed forum spammers' modi operandi. One of their important tools is a forum spamming automator. Its result, "The Nuts and Bolts of a Forum Spam Automator" will be appeared in USENIX LEET 2011.


As a forum spam prevention effort, we have just started developing a new technique. It's still in a brain storming phase. However, we're excited at this new direction.

We have been working on IPsec mobility and power saving. We have focused mainly on saving energy on WiFi mobile devices using a security protocol such as IPsec. The result has been presented at SecureComm 2010.

In this work, Asprox infection dynamics have been extensively investigated. It is pretty interesting how long it takes for administrators to remedy their web servers infected by Asprox SQL injectin attacks. Its result has been published at DIMVA 2009 and is available here.

We have proposed a countering spam scheme based on history-based prioritization and token-based authentication. The result has been published in SRUTI '06.

We have conducted the detailed overhead evaluation of SSL on PDAs to investigate how well security protocols, in particular SSL perform on small devices. Its result was presented at GI 2009.

We have evaluated the performance on IPsec-based VPNs, identifying the performance bottleneck and characteristics of IPsec protocol. In particular OpenSwan, an open source IPsec implementation for Linux, was used for this purpose. Furthermore, we analyzed the snapshotted logs of the VPN server at Indiana University - Bloomington to characterize the real VPN workload. The result has been published in NPSec 2005.

This project was to translate a productive and powerful high level script language such as a MatLab script into more fast and efficient C/C++ or Fortran language; futheremore, it would apply automatically sequencial MATLAB script codes parallelism. I focused especially on type inferencing issue. This work had been performed with Prof. Arun Chauhan.

I worked in Condor Team as a Research Assistant. I was involved into this project to contribute to them with database technologies and perspectives. I had been looking for various ways to integrate DB technologies into a Grid computing world. I mainly worked on how to manage efficiently ClassAds, Semi-structured data in RDBMS; thus, I did some experiments with DBMS and a XML native storage system, and designed and built jqmond (Job Queue MONitor Daemon). Here is a brief report about that. For this project, I did a database research under an advisor, Prof. Jeffrey F. Naughton's supervision, and also worked with Prof. Miron Livny for Condor system details.

I worked on Scalable Search Engines Via Adaptive Topic-Driven Crawlers under Prof. Filippo Menczer's supervision. In particular I explored a data fusion issue, which was a problem happening when each peer should merge search results from other neighbor peers in a P2P distributed web search system. I also looked into utilizing a centralized web search service such as Google, Yahoo, MSN search as peer in 6Search, which is a P2P distributd web search system.

Mobile CyberMap seeks to build a GIS software for PDA (Window CE-based). My responsibility was to design a software architecture and to write codes for display, networking, and GPS functions.

CyberMap builds a Web-based GIS service using Java, a Vector Map, Spatial Indexing, Memory DB, and fast communication on TCP/IP technologies.

MapWorld seeked to build a client-side interoperable concurrent updating tool so that many client can concurrently update spatial data of various GIS servers through cooperation. My responsibility was to design the control protocol of concurrency and consistency of cached spatial data, and to build the server module for the protocol.

The Development of the Open Desktop GIS S/W on Windows 95/98 Platform developed the open GIS compliant client S/W whcih could be used as the client tool interoperable to differenct GIS servers or as the stand-alone desktop GIS S/W.

• The Development of Interoperable OpenGIS Data Provider for MGE based on OLE DB presented the standard interfaces of the Data Source Access Interface which could be used to uniformly connect and then access different data sources independent of the types of GIS servers although it mainly dealt with the issues of implementing the OpenGIS data provider for MGE.
• The Design and Implementation of Interoperable OpenGIS API based on CORBA implemented the standard interfaces of OpenGIS by using CORBA which is a distributed object technology.

The Research and Development of Manipulation Tool for Handling of Geographic Information designed and implemented a spatial query language, called GSQL, on the Gothing S/W. The standard SQL was extended for allowing of spatial operations in the expression of the query language.