Ranking by inspiration: a network science approach
- PDF / 1,248,041 Bytes
- 25 Pages / 439.37 x 666.142 pts Page_size
- 12 Downloads / 185 Views
Ranking by inspiration: a network science approach Livio Bioglio1
· Valentina Rho1 · Ruggero G. Pensa1
Received: 24 June 2018 / Revised: 17 February 2019 / Accepted: 6 July 2019 © The Author(s), under exclusive licence to Springer Science+Business Media LLC, part of Springer Nature 2019
Abstract Contagion processes have been widely studied in epidemiology and life science in general, but their implications are largely tangible in other research areas, such as in network science and computational social science. Contagion models, in particular, have proven helpful in the study of information diffusion, a very topical issue thanks to its applications to social media/network analysis, viral marketing campaigns, influence maximization and prediction. In bibliographic networks, for instance, an information diffusion process takes place when some authors, that publish papers in a given topic, influence some of their neighbors (coauthors, citing authors, collaborators) to publish papers in the same topic, and the latter influence their neighbors in their turn. This well-accepted definition, however, does not consider that influence in bibliographic networks is a complex phenomenon involving several scientific and cultural aspects. In fact, in scientific citation networks, influential topics are usually considered those ones that spread most rapidly in the network. Although this is generally a fact, this semantics does not consider that topics in bibliographic networks evolve continuously. In fact, knowledge, information and ideas are dynamic entities that acquire different meanings when passing from one person to another. Thus, in this paper, we propose a new definition of influence that captures the diffusion of inspiration within the network. We call it inspiration score, and show its effectiveness in detecting the most inspiring topics, authors, papers and venues in a citation network built upon two large bibliographic datasets. We show that the inspiration score can be used as an alternative or complementary bibliographic index in academic ranking applications. Keywords Information diffusion · Bibliographic indexes · Citation networks · Topic modeling
Editors: Takuya Kida, Takeaki Uno, Tetsuji Kuboyama, Akihiro Yamamoto.
B
Livio Bioglio [email protected] Valentina Rho [email protected] Ruggero G. Pensa [email protected]
1
Department of Computer Science, University of Turin, Turin, Italy
123
Machine Learning
1 Introduction Contagion models and, in particular, stochastic epidemic models, have been widely studied in life sciences to understand and predict the spread of infectious diseases (Britton 2010). However, their implications are largely tangible in computational social science, machine learning and network science, due to their ability in explaining the dynamics of many social phenomena, such as the diffusion of ideas (Rogers 2003), the virality of certain posts or memes in social media (Yang and Zha 2013) and social influence (Cialdini and Trost 1998). Information diffusion, in particul
Data Loading...
