Please use this identifier to cite or link to this item: http://hdl.handle.net/10603/171866
Title: DIFFUSION MINING FOR SOCIAL NETWORK ANALYTICS
Researcher: Jaitly, Vanita
Guide(s): Sumeet Dua
Keywords: 
University: Chitkara University
Completed Date: 2016
Abstract: Social network or social graph can be defined as a graphical data structure which newlineconsists of social actors or users in a social network interacting with each other newlineresulting in the creation of complex ties between each other. A community in a newlinesocial network, as a corollary can be treated as the logical grouping of social newlineactors that share common interests, ideas, or beliefs. newlineRelated works in the area of social network mining have largely focused newlineon extracting and measuring both the density and sign of the communities, to newlinebetter understand both social and behavioral characteristic of social actors on the newlinecreation and dissemination of communities (also referred to as social influence). newlineThe techniques associated with community identification largely rely on structural newlinecharacteristics of social networks and treat both structural density (between newlinegroups) and type of interactions (sign) as independent variables. In this work we newlinehypothesize that both density and sign in social networks are related to each other, newlinebelieve that we can combine these two variables using higher ordered features of newlinedensity namely betweeness, closeness, centrality, eccentricity, and stress . Our newlineanalysis attempts to show that amalgam of these features is effective in detecting newlineadhoc communities. newlineIn this work, we examine our results of community detection pertinent to newlinesupport our premise of social influence. In order to achieve this, online social newlinenetworks are explored to find out the set of influential and vital users, also known newlineas seed set, who spreads any idea, product or behavior and thus influence the other newlineneighborhood inactive nodes, influential nodes. Here, we focus on building a newlinemodel that will help in increasing the possibility and adoption of a product newlinethrough a small set of influential nodes or users. For this, we use features of newlinemaximum flow to establish the estimation of adhoc communities. We believe our newlineapproach is novel in choosing our influencers (seeds) and thus by using these newlineseeds positive and negative edges are establish
Pagination: 94p.
URI: http://hdl.handle.net/10603/171866
Appears in Departments:Department of Computer Science and Engineering

Files in This Item:
File Description SizeFormat 
final_dissertation_draft vanita.pdfAttached File4.98 MBAdobe PDFView/Open


Items in Shodhganga are protected by copyright, with all rights reserved, unless otherwise indicated.