最先端プログラム講演会 ★5月30日★ Prof. Ling Liu talks about Social Influence based approach to Graph mining

[Miyuki Nakano <miyuki [at] tkl.iis.u-tokyo.ac.jp>]

日本データベース学会の皆様、

以下の講演会が5月30日(火)・東大生研にて行われます。
ふるってご参加ください。
　　　　　　　　　　　　　　　　　東京大学生研
　　　　　　　　　　　　　　　　　中野美由紀

　☆☆☆　5月30日　講演会のご案内　　　☆☆☆

主催　最先端研究開発支援プログラム（FIRST）: 超巨大データベース時代に
　　　向けた最高速データベースエンジンの開発と当該エンジンを核とする
　　　戦略的社会サービスの実証・評価
後援　電子情報通信学会　データ工学研究専門委員会


日時　5月30日（水）　午後6時〜7時
場所　東京大学生産技術研究所　E棟　5階　会議室Ｂ（Ew-502）
      http://www.iis.u-tokyo.ac.jp/map/index.html

Speaker : Prof. Ling Liu (Georgia Institute of Technology)
Title :  Social Influence based Approach to Graph Mining


参加費　無料

皆様のご参加をお待ちしております。

                        FIRST :　研究代表者　　　　　　　喜連川　優
                        電子情報通信学会データ工学研究会　委員長
　　　　　　　　　　　　　　　　　　　　　　　　　　　　 中野　美由紀
　　　　　　　　　　

-----------------------------------------------------------------
To: first_lecture [at] tkl.iis.u-tokyo.ac.jp

　講演会　参加申し込み

5月30日（水）の講演会に参加
・名前　　　
・ご所属
------------------------------------------------------------------
プログラム

Title : Social Influence based Approach to Graph Mining


Abstract:
There is a growing interest in clustering social network graph by
considering both social interactions (such as friendship among people)
within a single collaboration network (self influence) and the
interactions between the collaboration network and other information
networks (co-influence). Social influence analysis has great potential
for understanding the ways in which information, ideas and experiences
are spread or diffused across a social network.

This talk presents an innovative social influence based graph clustering
framework with three unique features. First, we define the concept of
influence based distance in terms of propagating heat diffusion kernel
on both collaboration graph and its associated influence graphs. Second,
we introduce a weight function with an iterative update method to
integrate vertex closeness scores on multiple social influence graphs
through weight assignments. Third, we design an iterative learning
algorithm SI-Cluster for social influence based graph clustering. It
partitions a large collaboration network into k clusters by continuously
quantify and adjusts the weighted contributions from different influence
classes in the co-influence model as we recursively refine the clusters
until reaching convergence. To make the SI-Cluster algorithm converge to
a global maximum as faster as possible, we transformed a sophisticated
nonlinear fractional programming problem of multiple weights into a
straightforward nonlinear parametric programming problem of single
variable. Experimental results on three real graphs demonstrate that our
SI-Cluster approach not only achieves a very good balance between
collaboration and influence similarities but also scales well for
clustering large graphs in terms of time complexity while meeting high
density and low entropy guarantee.


Bio:  Ling Liu is a full Professor in the School of Computer Science at
Georgia Institute of Technology. She directs the research programs in
Distributed Data Intensive Systems Lab (DiSL), examining various aspects
of large scale data intensive systems. Prof. Ling Liu is an
internationally recognized expert in the areas of Database Systems,
Distributed Computing, Internet Systems, and Service oriented computing.
She has published over 300 international journal and conference articles
and is a co-recipient of the best paper award of ICDCS 2003, the best
paper award of WWW 2004, 2005 Pat Goldberg Memorial Best Paper Award,
and the best data engineering paper of 2008 International Conference on
Software Engineering and Data Engineering. In 2012, Prof. Liu received
an IEEE Computer Society Technical Achievement Award and an Outstanding
Doctoral Thesis Advisor award for producing outstanding PhD students and
her service and dedication to Georgia Institute of Technology. Prof. Liu
served as general chair and PC chairs of several IEEE and ACM
conferences in data engineering and distributed computing fields and
served on editorial board of over a dozen international journals.
Currently Prof. Liu is on the editorial board of Distributed and
Parallel Databases (Springer), Journal of Parallel and Distributed
Computing (JPDC), IEEE Transactions on Service Computing (TSC), and ACM
Transactions on Web (TWEB). Dr. Liu’s current research is primarily
sponsored by NSF, IBM, and Intel.



-- 
-----------------------------------------------------------------------
中野　美由紀	東京大学　生産技術研究所　戦略情報融合国際研究センタ
Miyuki NAKANO	Institute of Industrial Science, Univ. of Tokyo
　　　　　　　　Center for Information Fusion
miyuki [at] tkl.iis.u-tokyo.ac.jp