What is it about?

Illegal logging has generated recurrent civic mobilization in Romania, but the available evidence is dispersed across semi-structured civic and web archives. We construct a corpus of 278 source records dated between 2007 and 2026 and distinguish source records from unique protest events.

Featured Image

Why is it important?

A blinded primary annotation of all records identified 114 records describing physical actions, six national or online campaigns, 152 non-protest records, and six unclear cases; a second coder annotated a prespecified 84-record sample. We compare gazetteer matching, rule-based natural language processing (NLP), and named entity recognition (NER). At county level, F1 was 35.0% for the gazetteer, 30.5% for rule-based NLP, and 41.1% for NER. At exact-locality level, the gazetteer had the highest F1 (27.0%). After removing false-positive county assignments diagnostically, NER most closely reproduced the reference distribution (total-variation distance 0.133; Spearman rho 0.958).

Perspectives

Documented physical-action records peaked in 2015, and the manually coded geography was concentrated in Ilfov/București, Cluj, Alba, Brașov, and Prahova. The findings show that established extraction methods have different error profiles and should be validated before they are used for substantive spatial inference.

Dr Remus Cretan
west university of Timisoara

Read the Original

This page is a summary of: Where forest protests happen: Validating location extraction in reports of mobilisation against illegal logging, Next Sustainability, December 2026, Elsevier,
DOI: 10.1016/j.nxsust.2026.100561.
You can read the full text:

Read

Contributors

The following have contributed to this page