|
|
Illini Mine Project is lead by Prof. Jiawei Han, the advisor of the data mining group at University of Illinois, Urbana-Champaign. For years, the data mining group at UI designed and published numerous new algorithm in the data mining community. However, these new algorithms are not well-adopted and known by larger crowd because of its complexity for programming. This project is an initiative for making the programs that are developed inside the group to be open-source.
Illini Mine is a package other than a library. This means that most programs in Ilini Mine are individually executable. This makes it easy to extract a single algorithm, and do test/improvement to the code. However, there are still things that are common to almost all algorithms. This include the uniform directory struct, the uniform compiling/testing commands, the uniform document format, the uniform input file format, the shared manual testing GUI and so on.
This project is to perform a systematic investigation of the principles, algorithms, and applications of scalable sequential and structured pattern mining, which covers the following issues: (1) development of highly scalable mining algorithms, including mining max-patterns, closed patterns and top-k patterns; (2) investigation of highly flexible mining methodologies, including mining of multi-dimensional multi-level sequential and structured patterns and constraint-based mining; (3) extension of the scope to cover sequential or structured pattern-based clustering; and (4) application of multi-dimensional, multi-level sequential or structured pattern mining for intrusion detection, Web mining, and other important applications. This will lead to a set of efficient, scalable, and flexible sequential and structured pattern mining methods for scientific and industrial applications.
The proposed project is based on our long-term active research, strong research record, fruitful research results, and rich experience on data mining, data warehousing, database systems, and data mining applications, as well as our preliminary work on multi-dimensional stream data analysis. Due to the fast generation of huge volumes of stream data in many applications, such as computer network traffic, video surveillance, telecommunication, Web clicking stream, stock market, and so on, stream data mining has become an active theme of research in data mining, with broad applications in industry and deep implications in other related research. This project will perform a systematic and in-depth investigation on the principles, methods, and implementation techniques for mining data stream dynamics, which will lead to deep understanding of the issues and effective solutions for mining the dynamics of data streams in multi-dimensional space. The project will further advance the knowledge of the principles, methods, and applications of stream data mining in particular and data mining in general.
The MAIDS (Mining Alarming Incidents in Data Streams) project is aimed to perform a systematic investigation of stream data mining principles and algorithms, develop effective, efficient, and scalable methods for mining the dynamics of data streams, and implement a system prototype for online multi-dimensional stream data mining applications. This project will develop and implement some new and existing algorithms to discover changes, trends and evolution characteristics in data streams, construct clusters and classification models from data streams, and explore frequent patterns and similarities among data streams. The methods developed by this project will be applied to network intrusion detection, telecommunication data flow analysis, credit card fraud prevention, Web click streams analysis, financial data trend prediction, and other applications.
Database and Information Systems Laboratory (DAIS), Department of Computer Science, University of Illinois at Urbana-Champaign