Inferring Social Ties And Identification Of Triadic Closure In Homophilic Social Networks

The stochastic communication behavior of the progressing world is tremendously impacting Social Networks(SNs).Rapidly growing SNs encourage individuals to com-municate with new users and to form new ties.In the past two decades,the telecom-munication sector and the internet's communication content have shown exponential growth.Analyzing the Big Data over social platforms and extracting accurate,useful information has remained one of the researchers' key challenges.Besides Analysis,SNs Big Data mining,and solving issues like narrative building,religious denominations,in-fluencing communities,predicting the new ties,and exploring the hidden relationships are some of the associated problems.Social Network Analysis(SNA)is one of the most researched areas;hence extensive theories have been developed to identify and resolve the special issue related to user privacy,an influencer in SNs,rumor propagation ori-gins,and link prediction.Social ties,social triads,and homophily are the fundamental evaluation measures of SNA,which assist several social theories which construct sys-tems like recommender systems and E-Commerce.In the literature,there are numerous applications available based on these measures.However,the relationship between so-cial triads and homophily is one of the less-discussed researched areas.In order to fill the stated research gap,we examined the behavior of social communications and in-vestigated the reasons for the association between the measures using Location-Based Social Networks(LBSNs).Analysis of any SNs generically has three research directions,i.e.,Analysis of user in SNs,Analysis of the social connections,and Analysis of the content in the SN.During the research study,our focus was to explore the social connections within a network.In the study,we used one close source LBSN and two openly available data sets.We experimented with and mined these data sets and proposed an inference framework and an algorithm.Social tie plays a vital role in building a social community.Exploring the variation in ties' strength,such as weak and strong,and estimating the dynamism related to time are the most discussed research issues.The research scope is limited to the inference of social ties and building connections between social triads and homophily.The main contributions are summarized as follows;1.We developed a model and algorithm to solve the problem of inferring social ties based on multi co-existence within time-frame.We explored and investigated the social connections based on direct communication.Initially,we examine the pat-terns of user communication by mining the Caller Data Records(CDR).CDR is a kind of LBSNs,and the key objectives to explore any LBSNs are to understand the User,understand the location,or understand the content of the communication.We find the co-occurrence of people based on the same base station connectivity in a certain time-frame during the research.Furthermore,we related the existence of two users with direct communication.We develop an algorithm that infers the social ties depending upon Co-occurrence count and Gap time threshold values.Furthermore,it gathers the direct call details and cross relates with the inferred results using precision and recall evaluation measures.2.We proposed a framework and algorithm to solve the problem of inferring sus-picious hidden ties by conducting location-based activity and simulation.Users who were not having any social relationship among them but were inferred as social friends were the second research part's baseline.The framework takes the false-positive results of the social tie inference module and evaluates the number of mutual friends and co-occurrences.In the literature,such non-existing social ties are referred to as missing ties.The proposed inference framework further classifies the sub-set of missing ties as suspicious social connections.A pipe and filter architecture-based inference framework was developed that highlights users'hidden relationship,which does not exist in real data.Comparison and evaluation of framework are conducted using real-time human-based activity and computer simulation.3.We developed a framework and proposed a mathematical theory to solve the prob-lem of identifying triadic closure in homophilic social networks.We first study the user communication and mobility patterns by observing the location-based communications.We developed a SN graph of users and identified home loca-tion using Home Detection Algorithm from the datasets.We also proposed a correlation statistical model and formulated triadic closure to solve a problem of triad classification in homophilic social networks and evaluated the throughput.We investigated the origins of social triads in detail and examined the formation of triads.Based on the analysis,we categorized social triads and compared their behaviors within the homophilic SN.We initially identified and then classified the existing class of triads into three classes based on the location.We found positive correlations between the homophily coefficient and the number of specific social triads.We structured homophilic triads into transitive and intransitive groups.We quantified the communication relative throughput for the homophilic triadic classes and correlated to the homophily coefficient.