Discovering Themes from Online News Articles on the 2018 Mt. Mayon Eruption

In this paper, we discovered themes from 2018 Mayon eruption-related articles through a generative statistical model. The article corpus undergone data cleaning before data processing. Based on the experiments conducted, the topic model which achieved the highest composition was generated using 5 topics, 15 top words, 500 iterations and 10 optimization interval as its parameters. Topic models which consist cluster of words were examined and manually labeled. The generative statistical topic modelling was able to draw relevant clusters of words from set of Mayon eruption-related articles. This automated approach aided in determining topic labels and analyzing the topic content out of the processed model. Themes discovered include, affected population and emergency assistance, volcanic activities, disaster tourism, flight status and volcanic eruption hazards. Presence of volcanic activity updates, calamity assistance from both government and private sectors and the destructive impact of eruption to the communities of Albay were revealed in the set of news articles. The discovered themes also show that in spite of the volcanic wrath, there is one positive impact which is disaster tourism. The lava flows and fountains pulled foreign tourists to visit the province.