14
Lehrstuhl Informatik 5 (Informationssy steme) Prof. Dr. M. I5-KPCa-0209- 1 Complex 2009 Ralf Klamma Pham Manh Cuong Yiwei Cao You Never Walk Along: Recommending Academic Events Based on Social Network Analysis Ralf Klamma, Pham Manh Cuong, and Yiwei Cao 曹曹曹 Informatik 5 (Information Systems), RWTH Aachen University 25.02.2009 Shanghai, China The First International Conference on Complex Sciences (Complex‘09)

You Never Walk Along: Recommending Academic Events Based on Social Network Analysis

Embed Size (px)

DESCRIPTION

The First International Conference on Complex Sciences (Complex‘09)

Citation preview

Page 1: You Never Walk Along: Recommending Academic Events Based on Social Network Analysis

Lehrstuhl Informatik 5(Informationssysteme)

Prof. Dr. M. JarkeI5-KPCa-0209-1

Complex 2009

Ralf KlammaPham Manh Cuong

Yiwei Cao

You Never Walk Along:Recommending Academic Events Based on Social Network Analysis

Ralf Klamma, Pham Manh Cuong, and Yiwei Cao 曹怡蔚

Informatik 5 (Information Systems), RWTH Aachen University

25.02.2009Shanghai, China

The First International Conference on Complex Sciences (Complex‘09)

Page 2: You Never Walk Along: Recommending Academic Events Based on Social Network Analysis

Lehrstuhl Informatik 5(Informationssysteme)

Prof. Dr. M. JarkeI5-KPCa-0209-2

Complex 2009

Ralf KlammaPham Manh Cuong

Yiwei Cao

Agenda

Research Questions Methodologies and Data Set Web Services for Academic Communities

– Conceptual Design and Implementation – Results

Conclusions and Outlook

Page 3: You Never Walk Along: Recommending Academic Events Based on Social Network Analysis

Lehrstuhl Informatik 5(Informationssysteme)

Prof. Dr. M. JarkeI5-KPCa-0209-3

Complex 2009

Ralf KlammaPham Manh Cuong

Yiwei Cao

Research Questions Research Communities

– How can you young researchers identify emerging research communities?

– New, evolving, interdisciplinary research communities?– How can research communities attract brilliant newbies with Web 2.0

services (Facebook, Flickr, YouTube, Blogs, Wikis, …)? Data Management

– Digital Libraries trace past events– Event Management systems trace future events– Research community websites trace event media

Web Tools for Computer Scientists– Visualizing academic community networks– Visualizing academic community development– Identifying key community members– Recommending future academic events

Page 4: You Never Walk Along: Recommending Academic Events Based on Social Network Analysis

Lehrstuhl Informatik 5(Informationssysteme)

Prof. Dr. M. JarkeI5-KPCa-0209-4

Complex 2009

Ralf KlammaPham Manh Cuong

Yiwei Cao

Evolution of a Database Research Community

VLDB 1990 VLDB 1995

VLDB 2000 VLDB 2006

Page 5: You Never Walk Along: Recommending Academic Events Based on Social Network Analysis

Lehrstuhl Informatik 5(Informationssysteme)

Prof. Dr. M. JarkeI5-KPCa-0209-5

Complex 2009

Ralf KlammaPham Manh Cuong

Yiwei Cao

Combination of SNA and Recommendation Techniques

Algorithmic Social Network Analysis (SNA)– Centrality measures [Brandes and Erlebach 2005]– Web-based network visualization

Data Management– Event model for academic events– Crawling of web data sources– Spatio-temporal analysis of event series

Recommendation techniques [Adomavicius and Tuzhilin 2005]– Collaborative Filtering (CF) [Breese et al. 1998 ]– Content-based recommendation [Sarwar et al. 2001]– Hybrid approaches [Burke 2002]

Page 6: You Never Walk Along: Recommending Academic Events Based on Social Network Analysis

Lehrstuhl Informatik 5(Informationssysteme)

Prof. Dr. M. JarkeI5-KPCa-0209-6

Complex 2009

Ralf KlammaPham Manh Cuong

Yiwei Cao

A Model for Academic Events and Scientific Communities

Page 7: You Never Walk Along: Recommending Academic Events Based on Social Network Analysis

Lehrstuhl Informatik 5(Informationssysteme)

Prof. Dr. M. JarkeI5-KPCa-0209-7

Complex 2009

Ralf KlammaPham Manh Cuong

Yiwei Cao

Web Data Sources of Academic Events

Eventseer.net (up-to-date)– 9,759 events,

5,220 topics – 700,285 people – 3,119 organizations

DBLP– 1,048,576 records– 220 > 1 million publications– Over 12000 person records

Mashed-up data set– 16821 events in 2099 series– 522938 authors with 4910 topics– 1282796 links about co-authorship of events

Page 8: You Never Walk Along: Recommending Academic Events Based on Social Network Analysis

Lehrstuhl Informatik 5(Informationssysteme)

Prof. Dr. M. JarkeI5-KPCa-0209-8

Complex 2009

Ralf KlammaPham Manh Cuong

Yiwei Cao

Event Recommendation Tools for Researchers: Algorithm

Input: set of events E = (e1, e2, ..., eN) set of researchers U = (u1, u2, ..., uM)

Output: top k most recommended events to target researcher ut

Building the model- Constructing the participation matrix

Computing similarities: the cosine-based approach

Generating recommendations: the aggregation function

Page 9: You Never Walk Along: Recommending Academic Events Based on Social Network Analysis

Lehrstuhl Informatik 5(Informationssysteme)

Prof. Dr. M. JarkeI5-KPCa-0209-9

Complex 2009

Ralf KlammaPham Manh Cuong

Yiwei Cao

Event Recommendation Services for Researchers

Based on three-tier architecture- Data backend: Oracle database- Middle tier: Oracle standalone OC4J

Implementation: using Java- Client side: AJAX (yFiles AJAX, Google

Maps, Yahoo chart)- Server side: JSPs, Java Servlet, YouTube

APIs, Flickr APIs

Implementation IDEs- Oracle JDeveloper, Eclipse

Data for evaluation- DBLP XML records, the DBLP Web site

and EventSeer.net

Page 10: You Never Walk Along: Recommending Academic Events Based on Social Network Analysis

Lehrstuhl Informatik 5(Informationssysteme)

Prof. Dr. M. JarkeI5-KPCa-0209-10

Complex 2009

Ralf KlammaPham Manh Cuong

Yiwei Cao

Spatio-Temporal Analysis and Visualization

ACM SIGMOD conference location distribution

Page 11: You Never Walk Along: Recommending Academic Events Based on Social Network Analysis

Lehrstuhl Informatik 5(Informationssysteme)

Prof. Dr. M. JarkeI5-KPCa-0209-11

Complex 2009

Ralf KlammaPham Manh Cuong

Yiwei Cao

Indentify Key Members in a Research Community

ACM SIGMOD

ACM SIGMOD

Page 12: You Never Walk Along: Recommending Academic Events Based on Social Network Analysis

Lehrstuhl Informatik 5(Informationssysteme)

Prof. Dr. M. JarkeI5-KPCa-0209-12

Complex 2009

Ralf KlammaPham Manh Cuong

Yiwei Cao

Academic Community Development Development of the

community: number of participants over years

Continuity: participants by number of events attended

ACM SIGMOD

Page 13: You Never Walk Along: Recommending Academic Events Based on Social Network Analysis

Lehrstuhl Informatik 5(Informationssysteme)

Prof. Dr. M. JarkeI5-KPCa-0209-13

Complex 2009

Ralf KlammaPham Manh Cuong

Yiwei Cao

We Use It!

Page 14: You Never Walk Along: Recommending Academic Events Based on Social Network Analysis

Lehrstuhl Informatik 5(Informationssysteme)

Prof. Dr. M. JarkeI5-KPCa-0209-14

Complex 2009

Ralf KlammaPham Manh Cuong

Yiwei Cao

Conclusions and Outlook

Web Service available online– Try it out at:

http://bosch.informatik.rwth-aachen.de:5080/AERCS– Web 2.0 media aggregation for events– Recommendation engine on request

Outlook– Plug-in for conference websites– Mobile Facebook application for scientists– Integrate recommender systems with

multi-agent systems to simulate academic events