Migration trends in the Mediterranean region with text mining techniques: The example of Migration Policy Institute

The Mediterranean Basin has been a dynamic location in terms of migration activity throughout history, and in recent years, this area has been exposed to intensive amounts of migration. With the increase in migration activity, studies on migration have increased as well. Traditional methods are insufficient and complicated for examining and analyzing large numbers of studies. Therefore, text mining techniques are used to analyze studies effectively and quickly. This study aims to analyze the sentimental aspect, prominent subjects, and similarities between articles about migration in the Mediterranean context using the text mining technique. Seven hundred thirty-six articles within the context of Mediterranean countries regarding migration provided by the Migration Policy Institute are included in the study. The related articles are analyzed within three sections. The first section consists of determining the polarity (positive/negative/neutral) and polarity values (0.0–1.0) of the articles using sentiment analysis. The second section consists of determining the prominent subjects and predicting prospective topics using latent dirichlet allocation (LDA). The third section is the classification of the articles according to their similarities using text clustering. As a result of the analyses, it was determined that negative sentimental elements were dominant in the articles discussed in the context of the Mediterranean countries. Articles are thematized based on five topics. These topics are international migration and security, participation in working life, immigrant health policies, immigration and human rights, immigration and integration. It is anticipated that future studies will mainly focus on international migration and security. The articles are categorized into five categories according to their similarity status, and 18% of the articles were found to be similar to each other. Keywords: Migration, Migration in the Mediterranean Basin, Text Mining, Sentiment Analysis, LDA, Text Clustering