Not logged in.
Quick Search - Contribution
Contribution Details
| Type | Conference Presentation |
| Scope | Discipline-based scholarship |
| Title | The influence potential. A new approach to identify influential individuals from time-varying social interactions |
| Organization Unit | |
| Authors |
|
| Presentation Type | speech |
| Item Subtype | Original Work |
| Refereed | Yes |
| Status | Published in final form |
| Language |
|
| Event Title | Netsci-X |
| Event Type | conference |
| Event Location | Wrostlaw, Poland |
| Event Start Date | January 11 - 2016 |
| Event End Date | January 13 - 2016 |
| Abstract Text | Understanding how people influence or are influenced by their peers can help us understand the flow of market trends, product adoption and diffusion processes. In the last years an increas- ing attention has been devoted to identifying the most influential individuals and using them as super-spreaders of products or ideas. Most often, from the observed data researchers construct an influence network and identify the influentials as the most central individuals in this network or as the key players in the development of a dynamical process. However, in most practical situations there are at least two potential issues with this approach. First, the construction of the influence network is non-trivial as most often the influence relationships between people are not directly observable but rather aspects of people’s behavior. Furthermore, even if the influence relationships were observable, the static network representation cannot capture their time-dynamical aspect. Second, identifying influential individuals based on their role in a dynamical process is sensitive to the model chosen and to the assumed role influentials should play in the process. In this article we present a model-free approach to identify influential individuals from time varying social in- teractions which does not require constructing the influence network nor modeling social influence as a dynamical process. We start by introducing the influence potential (IP), a novel index that captures the intrinsic ability of individuals to consistently influence others while controlling for the total number of individuals in the process. We validate our results by computing an adapted version of the area under the curve (AUC) as indicator of the in-sample prediction accuracy and further use this to identify how many influential individuals are in a dataset. To illustrate our approach we analyze two real world systems: a news discussion forum extracted from cnn.com and a business review platform extracted from yelp.com. In both datasets we identify a low number of users with high influence potential relative to the entire user base. This implies that if we are interested to steer the user behavior on the platform, we can design intervention campaigns tar- geting influential individuals but with very few reliable targets. Furthermore, we compare the two datasets and find that Yelp has a higher percentage of individuals with high influence potential. This suggests that in the Yelp dataset it is relatively easier to locate potential targets, which might have important implications for comparing the intervention costs on the two platforms. |
| Export |
BibTeX
|