MPEDS: A Semi-automated Approach for the Generation of Protest Event Data

• In Event: 3453 - From Classic Questions to New Frontiers: Current Work in Mathematical Sociology

Abstract

This article introduces the Machine-learning Protest Event Data System (MPEDS), a system for the semi-automated coding of protest event data. MPEDS uses natural language processing and machine learning tools to
automate parts of protest event data creation process, is attentive to theoretical concerns of social movement scholarship and is readily available to movement scholars. The system reduces the labor required to generate protest event data in order to maximize temporal and spatial coverage of protest event data while minimizing the biases associated with coding only one or two national-level news sources. Data produced with the MPEDS system is as reliable, if not more so, than that produced by human coders. MPEDS is open, available for replication, and extendable by social movement researchers and computational social scientists.

Author

• Alex Hanna, University of Toronto