Temporal topics in online news articles: Migration crisis in Venezuela

No Thumbnail Available
Journal Title
Journal ISSN
Volume Title
The migration process of citizens of a country in crisis can be extended by months or years depending on the social, economic, political, institutional situation of the country of origin. There are several articles that approach Venezuelan migration based on information provided by international organizations such as ONU and ACNUR. However, so far there is not a study that analyzes the information media offers about this issue, so we have temporarily analyzed online news to be able to obtain the topics that emerges from this social phenomenon. First, we extract 10K news articles published online in different newspapers across Latin America since 2015 until May 2019. Second, we build a binary classifier to discriminate whether the article is related to migration or not. Finally we apply topic modeling and word embeddings techniques to extract the most important issues discussed each year. The automatic text analysis, in time, reveals how the country moves from an intense migratory flow to an exodus of people of all ages and conditions who are welcomed by the host countries of the region starting with the closest neighbors. Our temporal analysis shows evidence that the migration process continues to increase and it is spreading throughout the continent mainly due to unemployment, insecurity and the lack of medicines and food.
migration crisis, venezuelan migration, text classification, online news, word embedding, topic modeling