How Did You Get to Know That? A Traceable Word-of-Mouth Algorithm

Word-of-mouth communication has been shown to play a key rolein a variety of environments such as viral marketing and virusspreading. A family of algorithms, generally known as informationspreading algorithms or word-of-mouth algorithms, has beendeveloped to characterize such behavior. However, they havelimitations, including the inability to: (1) capture when the communicationsor contacts take place and (2) explain where the influencecomes from. These drawbacks have limited the studiesabout how the spreading of influence takes place in social networks.In this paper, we present a new word-of-mouth algorithmthat considers the temporality of the communications and keepstrack of how influence travels over the social network. We validatethe proposed algorithm via simulations of word-of-mouthtraces on call detailed records, in order to model how influencespreads. Our results indicate that (1) static factors of social networksare not enough to model influence and (2) there seems to bestatistical invariants of how influence spreads in a network.