Applying machine learning to media analysis improves our understanding of forest conflicts
2024 (English) In: Land use policy, ISSN 0264-8377, E-ISSN 1873-5754, Vol. 144, article id 107254Article in journal (Refereed) Published
Abstract [en]
Conflicts over the management and governance of forests seem to be increasing. Previous media studies in this area have largely focused on analysing the portrayal of specific conflicts. This study aims to review how a broad range of forest conflicts are portrayed in the Swedish media, analysing their temporal, spatial, and relational dimensions. We applied topic modelling, a machine learning approach, to analyse 53,600 articles published in the Swedish daily press between 2012 and 2022. We identified 916 topics, of which 94 were of interest for this study. Our results showed ten areas of forest conflicts: hunting and fishing (35 % of total coverage), energy (24 %), recreation and tourism (11 %), nature conservation (8 %), forest damages (6 %), international issues (5 %), forestry (5 %), reindeer husbandry (4 %), media and politics (2 %), and mining (1 %). The overall coverage of forest conflicts increased significantly over the study period, potentially reflecting an actual increase in forest conflicts. Some of the conflicts were continuously reported upon over time, while the coverage of others exhibited seasonal or event-related patterns. Four conflicts received most of their coverage in specific regions, while others were covered across the whole of Sweden. A relational analysis of the conflicts revealed three clusters of forest conflicts focused respectively on industrial, cultural, and conservation conflicts. Our results emphasise the value of using topic modelling to understand the overall patterns and trends of the media coverage of current land use conflicts, while also highlighting potential areas of emerging conflicts that may be of special interest for planners and policy-makers to monitor and manage.
Place, publisher, year, edition, pages Elsevier, 2024. Vol. 144, article id 107254
Keywords [en]
Forest policy, Agenda-setting power, Daily press, Topic modelling, BERTopic
National Category
History of Science and Ideas Political Science Media and Communications Sociology (excluding Social Work, Social Psychology and Social Anthropology)
Identifiers URN: urn:nbn:se:umu:diva-227604 DOI: 10.1016/j.landusepol.2024.107254 ISI: 001262456900001 Scopus ID: 2-s2.0-85197480879 OAI: oai:DiVA.org:umu-227604 DiVA, id: diva2:1880273
Funder Swedish Research Council Formas, 2017–01956 2024-07-012024-07-012025-02-21 Bibliographically approved