Social Network Analysis and Information Propagation: A Case Study Using Flickr and YouTube Networks

Abstract

Social media and Social Network Analysis (SNA)acquired a huge popularity and represent one of the mostimportant social and computer science phenomena of recentyears. One of the most studied problems in this research area isinfluence and information propagation. The aim of this paper isto analyze the information diffusion process and predict theinfluence (represented by the rate of infected nodes at the end ofthe diffusion process) of an initial set of nodes in two networks:Flickr user’s contacts and YouTube videos users commentingthese videos. These networks are dissimilar in their structure(size, type, diameter, density, components), and the type of therelationships (explicit relationship represented by the contactslinks, and implicit relationship created by commenting onvideos), they are extracted using NodeXL tool. Three modelsare used for modeling the dissemination process: LinearThreshold Model (LTM), Independent Cascade Model (ICM)and an extension of this last called Weighted Cascade Model(WCM). Networks metrics and visualization were manipulatedby NodeXL as well. Experiments results show that the structureof the network affect the diffusion process directly. Unlikeresults given in the blog world networks, the information canspread farther through explicit connections than throughimplicit relations

Examining Sentiments and Popularity of Pro- and Anti-Vaccination Videos on YouTube

“…Furthermore, comparing the cascade effect between Digg and Twitter, researchers have observed that Digg's dense network leads to faster but limited spread of information, while Twitter's lose network created a slower but a wider spread of information. Thus, confirming that network structures can profoundly influence reach [29,30].…”

Efficient presolving methods for the influence maximization problem

“…It is worthwhile remarking that Theorem 4.5 also sheds a useful insight that for the IMP with a general large and well-connected network  (not necessary to be complete) with high arc probabilities 𝜋 ij , the SCNA and INA can be expected to effectively reduce the sizes of the live-arc graphs and SMCLP formulation (3). Indeed, a well-connected network is likely to contain large complete subgraphs.…”

“…Indeed, to simplify the problem formulation, we can remove variable z 𝜔 1 and its corresponding reachability constraint in (3b) and add the objective coefficient of variable z 𝜔 1 into that of variable z 𝜔 2 . In general, for any given two strongly connected nodes i 1 , i 2 ∈  in a live-arc graph  𝜔 (i.e., there exists a directed path from node i 1 to node i 2 in  𝜔 and vice versa), we can remove one of the two variables and the corresponding reachability constraint in (3b) from formulation (3). Notice that for a given SCC in a live-arc graph, as all of its nodes are strongly connected, we can recursively apply the above argument until there remains only a single variable and a single constraint in (3b) associated with this SCC.…”

“…In this section, we investigate a generalization of the IMP (3), which arises from many existing applications including the IMP and its variants [47,48,60,61], and discuss the extensions of the proposed SCNA and INA to this generalization.…”

“…Similar to the IMP (3), problem (31) is built upon a finite number of live-arc graphs  𝜔 = (,  𝜔 ), 𝜔 ∈ Ω. However, in contrast to the IMP (3), problem (31) can flexibly allow any objective function and any constraint in sets  and . Indeed, the IMP (3) can be seen as a special case of problem (31) where…”