Mining Interactions between Network Community Structure and Information Diffusion
Abstract
Recent technological advances are opening up unprecedented opportunities to understand the dynamics of society in incredible resolution and scale. Each day mobile phones and online social network services record when, how, what, where, and with whom we communicate and the data reveals fascinating details of human behavior and our society. The understanding of the structure, dynamics, and roles of communities will have huge impacts on both industry and public sectors. The research objective of this research project is to analyze and understand interactions between community structures and information diffusion, and ultimately develop predictive models of information diffusion based on community structure.
Intellectual Merit
The proposed research will advance our understanding of the interplay be- tween network structure and information diffusion by combining two distinct topics — communities and information diffusion — that rarely have been studied together. This research will lead us both to more accurate discovery of highly overlapping communities and to deeper understanding of how network community structure affects information spreading. Such understanding will also allow us to develop predictive models of information diffusion, which leverages the information of network community structure. Furthermore, the intellectual merits are not confined to social network research, as the methods can potentially be broadly applicable to other types of networks, as the PI’s research on biological networks exemplify.
Broader Impact
This project will have strong broader impacts. Because most students actively engage social networking services, it appeals to broader populations. There is a strong demand for network analysis curricula and research opportunities. The PI will leverage this opportunity to initiate a curriculum on network analysis and mining and to provide research opportunities to broad populations. The PI will provide research opportunities and mentoring to students from
HBCUs (Historically Black Colleges and Universities) through A4RC (Alliance for Advancement of African-American Researchers in Computing) and DREU program. The PI will develop online courses on social networks and reach out to underrepresented populations.
Since many social phenomena are related to community structure (e.g. racial and cultural seg- regation, conflict between groups, emergence of political polarization, rumors, misinformation), understanding the structure, dynamics, and roles played by communities will help us to under- stand these phenomena. Furthering our understanding of these phenomena will, in turn, help us to understand and mitigate some of their harmful effects, and will facilitate improved public policy design and campaigns for social good.
Use of FutureGrid
FutureGrid will be used to process and analyze large-scale network datasets. The algorithms will be developed based on map reduce framework and FutureGrid will be used as a testbed for the algorithms.
Scale Of Use
10~100 cores will be used infrequently. Exact amount of cpu time is hard to predict at the current stage.
Futuregrid is a resource provider for