Generalized Nets: A New Approach to Model a Hashtag Linguistic Network on Twitter

Abstract

In the last few years the micro-blogging platform Twitter has played a significant role in the communication of civil uprisings, political events or natural disasters. One of the reasons is the adoption of the hashtag, which represents a short word or phrase that follows the hash sign ($\#$). These semantic elements captured the topics behind the tweets and allowed the information flow to bypass traditional social network structure. The hashtags provide a way for users to embed metadata in their posts achieving several important communicative functions: they can indicate the specific semantic domain of the post, link the post to an existing topic, or provide a range of complex meanings in social media texts. In this paper, Generalized nets are applied as a tool to model the structural characteristics of a hashtag linguistic network through which possible communities of interests emerge, and to investigate the information propagation patterns resulting from the uncoordinated actions of users in the underlying semantic hashtag space. Generalized nets (GN) are extensions of the Petri nets by providing functional and topological aspects unavailable in Petri nets. The study of hashtag networks from a generalized nets perspective enables us to investigate in a deeper manner each element of the GN, substituting it with another, more detailed network in order to be examined in depth. The result is an improved understanding of topological connections of the data and the ability to dynamically add new details to expand the network and as a result discover underlying structural complexities unable to be discovered through traditional network analysis tool due to the prohibitive computational cost. Analysis is performed on a collection of Tweets and results are presented.