VISUALLY ENHANCED ANALYSIS OF REFUGEES’ SOCIAL MEDIA ACTIVITIES IN TURKEY

Abstract

The population of refugees in Turkey is significantly growing since2014 and it has already exceeded three millions. This situation urges to study theirconditions and issues as well as their opinions. In this study, we propose a methodto analyze the dataset collected from social media accounts of refugees in Turkeyin a textual and visual manner. Firstly, we acquire social media activities ofrefugees, and then we make textual analysis for which the results are presentedusing word clouds. Finally, for the most significant words obtained from the textualanalysis, we perform a visual analysis to find out the most representative image ora group of images shared in social networks. The association between textual andvisual results enhances their perceptibility and help decreasing the ambiguity ofinferences over analysis results. We experimented with different scenarios andsuggested several methods to enhance computational and qualitative results.