About PegasusN

Project History

Pegasus 2.0

Pegasus, a former version of PegasusN, was developed by CMU, and released as an open-source graph mining library. The library is based on Hadoop, which is a representative distributed computing framework. It contains various graph mining algorithms like PageRank, random walk with restart, radius estimation, and so on.

PegasusN

PegasusN improves the usability and scalability of Pegasus by adding graph mining algorithms, extending its underlying framework, and improving the performance of existing algorithms. PegasusN supports Hadoop and Spark, while Pegasus supports Hadoop only. PegasusN contains addtional graph mining algorithms such as triangle and subgraph enumeration, shortest path computation, and so on. The algorithms in PegasusN reduce the running time by distributing subgraphs (or subproblems) into multiple workers with consideration of I/O.

Contributors

PegasusN is developed in Data Mining Lab @ Seoul National University. The main contributors are:
  • Ha-Myung Park
    Ph.D. Student
    School of Computing, KAIST
    Email: hamyung.park@kaist.ac.kr
  • Chiwan Park
    M.S. Student
    Department of Computer Science and Engineering, Seoul National University
    Email: chiwanpark@snu.ac.kr
  • U Kang
    Associate Professor
    Department of Computer Science and Engineering, Seoul National University
    Email: ukang at snu.ac.kr

Acknoledgements

We thank YoungJin Kim for lending us her voice to narrate the demo video.