Embedding Informational Ecosystems in Ideological Spaces and Tracking Group Dynamics
- Pedro Ramaciotti Morales (SciencesPo, Paris, France)
Abstract
Recent years have seen numerous efforts to use the structure of social networks (to the exclusion of textual content) to infer ideological positions of users. In this work, we use ideological inference from topological data to estimate and track the ideology of collectives. Representing heterogeneous informational ecosystems at nation-wide scale as Knowledge Graph data, we use ideological positions inferred on social networks and project them to other parts of these ecosystems. This methodology allows us to embed different web entities, such as websites of news outlets or specific media content, in ideological spaces where dimensions stand for issues of public debate, and positions signal favorable or opposed attitudes. In particular, these projection operations allow us to embed web entities representing collectives, such as Facebook groups associated with social movements. We show that tracking the trajectories of social movements on ideological spaces is a valuable tool to characterize the evolution of their ideological and attitudinal stances.