A likelihood-based framework for the analysis of discussion threads

Abstract

Online discussion threads are conversational cascades in the form of posted messages that can be generally found in social systems that comprise many-to-many interaction such as blogs, news aggregators or bulletin board systems. We propose a framework based on generative models of growing trees to analyse the structure and evolution of discussion threads. We consider the growth of a discussion to be determined by an interplay between popularity, novelty and a trend (or bias) to reply to the thread originator. The relevance of these features is estimated using a full likelihood approach and allows to characterise the habits and communication patterns of a given platform and/or community. We apply the proposed framework on four popular websites: Slashdot, Barrapunto (a Spanish version of Slashdot), Meneame (a Spanish Digg-clone) and the article discussion pages of the English Wikipedia. Our results provide significant insight into understanding how discussion cascades grow and have potential applications in broader contexts such as community management or design of communication platforms.

Generative models of online discussion threads: state of the art and research challenges

“…• Kumar et al [83] • Wang et al [84], • Gómez et al [85] • Backstrom et al [86] • Nishi et al [87] • Lumbreras [88] • Aragón et al [89] The selection of these models is based on the consideration of Kumar et al [83] as the first generative model for online discussion threads. The rest of the models in the survey were selected after examining the publications citing this work in Scopus (52 papers) and Google Scholar (92 papers) 2 , and including the studies which proposed a generative model for the structure and growth of online discussion threads.…”

“…(1) and (3). In Gómez et al [85], a model comparison was also done to show the relevance of each feature in every dataset. This statistical test was performed by considering the likelihoods of different reduced models that neglect each of the features separately.…”

“…Actually, some of the previous models already denoted long path-like reply trees as 'skinny' in Kumar et al [83] or 'focused' in Backstrom et al [86], and large star-like trees as 'bushy' in Kumar et al [83] or 'expansionary' Backstrom et al [86]. Because many of the previous models are based on the branching process model which does not capture appropriately long chains of messages in discussion threads, the depth distribution is often underestimated, as noted in Kumar et al [83] and Gómez et al [85]. This last model proves that the branching process model produces unrealistic fractions of long path-like trees or large irregular trees (combination of star-like and path-like structures).…”

“…It is motivated by observing that the growth of discussion threads in previous generative models, in particular Kumar et al [83] and Gómez et al [85], is irrespective of the user who is writing a new message.…”

“…To this end, the model builds upon Gómez et al [85] and introduces latent types of users or roles. In this model, a role u, u = 1, .…”