<![CDATA[Online disaster news contains reported cause information, but the narratives are unstructured and difficult to use for systematic disaster risk analysis. This study develops a context-aware disaster cause mining framework to extract and analyze reported cause-context association patterns from Indonesian online news. The framework integrates text-based cause extraction, contextual enrichment using population density and meteorological variables, GA-optimized Apriori-based Association Rule Mining, and merged rule interpretation. Disaster news records were transformed into transactions containing disaster type, reported causes, population density category, three-day rainfall category, and maximum temperature category. From 742 final transaction records, the Apriori process generated 200 initial association rules. After filtering rules with reported causes and contextual attributes in the antecedent and disaster type in the consequent, 97 target rules were retained. The empirical analysis focused on forest and land fire as a case study, producing 24 rules and 5 merged rule patterns. The strongest merged pattern was related to land burning, with a merged support of 0.1173. The findings show that the framework can organize disaster narratives into interpretable reported cause-context association patterns for disaster risk analytics. However, the results should not be interpreted as verified causal evidence.]]>
Copyrights © 2026
