Chair of Network Architectures and Services Department of Informatics Technical University of Munich Analysis of Social Networks in IETF Mailing Lists Boonyakorn Jantaranuson Final Talk of Master’s Thesis June 25, 2018 Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Outlines Introduction Approach Analysis Conclusion Bibliography B. Jantaranuson – SNA in IETF Mailing Lists 2
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Motivation • IETF use mailing lists to communicate inside 100+ working groups • Social interactions happen from communications via a large num- ber of emails • Different WGs have different characteristics • SNA using IETF mailing lists is still new B. Jantaranuson – SNA in IETF Mailing Lists 3
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Dataset • Archived e-mails are available online • Database available from Bachelor’s Thesis and IDP of N. Schwell- nus [9] in PostgreSQL database format • DB relation of discussion threads is done by the bachelor’s thesis of S. Klimek [7]. B. Jantaranuson – SNA in IETF Mailing Lists 4
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Relationships between People • Replying emails • Appearing in same discussion threads B. Jantaranuson – SNA in IETF Mailing Lists 5
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Example Network (bmwg) B. Jantaranuson – SNA in IETF Mailing Lists 6
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Analysis Workflow B. Jantaranuson – SNA in IETF Mailing Lists 7
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Example 1: Centrality • Used a lot in studies using SNA to identify important individuals • In several papers [10, 5, 8], betweenness centrality is a common measure to identify important people in the Enron company • Evaluate the importance using peoples’ roles: CEO, manager, head of departments B. Jantaranuson – SNA in IETF Mailing Lists 8
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Centrality: IETF • Some people have specific roles in the WGs: chairs, area direc- tors, secretary, reviewers, tech advisors • Question: Can we use centrality as a measure to identify important people in IETF context? B. Jantaranuson – SNA in IETF Mailing Lists 9
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Centrality: Roles Analysis B. Jantaranuson – SNA in IETF Mailing Lists 10
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Centrality: What makes them important? • Send more emails or start more discussion threads B. Jantaranuson – SNA in IETF Mailing Lists 11
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Centrality: Changes in Centrality • Use the dynamic social network for the analysis • Data is from 2007-2017 • Moving window of size 1 year is used to generated a network at a point of time with 4 months step size • How stable is the importance of a person? • Example: v6ops (large WG) and rtgwg (small WG) B. Jantaranuson – SNA in IETF Mailing Lists 12
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Centrality: Changes in Centrality (v6ops) B. Jantaranuson – SNA in IETF Mailing Lists 13
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Centrality: Changes in Centrality (rtgwg) B. Jantaranuson – SNA in IETF Mailing Lists 14
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Example 2: Comparing IETF to Real-World Networks • Applying SNA to mailing list of Open-Source Software mailing lists from Bird et al. [1, 2] and conference mailing list from Chen et al. [3] • They found that those the degree distributions follow power-law distribution, which is the same as other real-world networks • Are networks generated from IETF mailing lists the same as real- world networks? B. Jantaranuson – SNA in IETF Mailing Lists 15
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Comparing IETF to Real-World Networks B. Jantaranuson – SNA in IETF Mailing Lists 16
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Example 3: Potential Events Analysis • Diesner et al. [4] and Juszczyszyn et al [6]. generated dynamic social networks and tracked the changes of e.g. # of active users, network density, average centrality • In Enron case [4], they spotted significant rise in the period before company crisis (bankrupt) • In university emails case [6], they found that the significant drop in the changes related to semester break • There are several events occurred in IETF community • Can we detect potential events from the change of social net- works? B. Jantaranuson – SNA in IETF Mailing Lists 17
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Example: httpbis • We analyze from the changes of the number of active users and the number of newcomers • In January 2012, the initiative for HTTP/2 was announced • In July 2014, the WG obsoleted the RFC of HTTP/1.1 • In May 2015, the WG released the RFC of HTTP/2 B. Jantaranuson – SNA in IETF Mailing Lists 18
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Example: httpbis (# of Active Users) B. Jantaranuson – SNA in IETF Mailing Lists 19
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Example: httpbis (# of Newcomers) B. Jantaranuson – SNA in IETF Mailing Lists 20
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Limitations • Very large number of working groups (101 WGs) • Lack of background knowledge in each studied WG • Hard to evaluate the results of possible analysis • The analyses are quite broad and exploratory B. Jantaranuson – SNA in IETF Mailing Lists 21
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Conclusion • Analysis on IETF mailing lists using SNA, statistics, and visualiza- tion techniques • The analyses are done in Jupyter Notebook and they are easy to follow and reproducible • We can use centrality to identify important people in the social network • We found that sending more emails and starting more threads make people in the WG more important • The centrality in larger WG are more stable • We can detect important events e.g. proposing new standards from the changes of social networks B. Jantaranuson – SNA in IETF Mailing Lists 22
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Thank you B. Jantaranuson – SNA in IETF Mailing Lists 23
Chair of Network Architectures and Services Department of Informatics Technical University of Munich [1] C. Bird, A. Gourley, P . Devanbu, M. Gertz, and A. Swaminathan. Mining email social networks. In Proceedings of the 2006 International Workshop on Mining Software Reposi- tories , MSR ’06, pages 137–143, New York, NY, USA, 2006. ACM. [2] C. Bird, A. Gourley, P . Devanbu, M. Gertz, and A. Swaminathan. Mining email social networks. In Proceedings of the 2006 International Workshop on Mining Software Reposi- tories , MSR ’06, pages 137–143, New York, NY, USA, 2006. ACM. [3] H. Chen, H. Shen, J. Xiong, S. Tan, and X. Cheng. Social network structure behind the mailing lists: Ict-iiis at trec 2006 expert find- ing track. The Fifteenth Text REtrieval Conference (TREC 2006) Proceedings , 2006. B. Jantaranuson – SNA in IETF Mailing Lists 24
Chair of Network Architectures and Services Department of Informatics Technical University of Munich [4] J. Diesner, T. L. Frantz, and K. M. Carley. Communication networks from the enron email corpus “it’s always about the peo- ple. enron is no different”. Computational & Mathematical Organization Theory , 11(3):201–228, Oct 2005. [5] J. Hardin, G. Sarkis, and P . C. Urc. Network Analysis with the Enron Email Corpus. ArXiv e-prints , Oct. 2014. [6] K. Juszczyszyn, K. Musiał, P . Kazienko, and B. Gabrys. Temporal changes in local topology of an email-based social network. COMPUTING AND INFORMATICS , 28(6), 2012. [7] S. Klimek. mail_threads.sql · GitHub. https://gist.github.com/simkli/8dffa3ee4ce456d9197404f20d40d4a8 . Accessed: 2018-02-24. B. Jantaranuson – SNA in IETF Mailing Lists 25
Recommend
More recommend
