Discovering networks of actors behind the dynamics of opinion
Abstract
Text-mining analysis can describe evolutions of data, even big data, but between two points many curves can join them depending on the dynamics of the process. If statistics approximation can produce error estimation with tendencies, how modelling of dynamics opinion can produce networks estimations from the data ? This paper studies this research issue with the study case dealing with the controversy of abnormal death of bees among French speaking journalists during a period of 13 years. Articles are tagged with three stances to explain the phenomenon, a uni-factor cause, the use of pesticides, a multi-factor cause, including one other factor different than pesticides at least, or the absence of an understanding. On this basis, evolutions of the respective proportions of each category of agents are obtained. Assuming agents are either flexible or inflexible journalists about their respective reports of the facts, their associated networks are extracted from the data applying Galam Unifying Frame (GUF) of opinion dynamics. The variation of inflexible agents explaining the issue either with unifactor reasons or multifactor causes, is a result of the modelling. From those distributions the actual networks of agents can be inferred.