Resumen de #iamhappybecause: : Gross National Happiness through Twitter analysis and big data

Ahmet Onur Durahim, Mustafa Coskun

• The prominence of social media has contributed to open information access for the researchers. With voluntary information sharing structure of Twitter, user disposition and sentiment analyses can be performed for determining the emotional well-being of the citizens. In this respect, we adopted a sentiment analysis model to calculate the Gross National Happiness (GNH) of a Middle East country, Turkey. For this purpose, over 35 million tweets, published in 2013 and in the first quarter of 2014, of over 20 thousand users were collected and analyzed. In the proposed model, prior to calculating the GNH by considering the polarities of tweets, first convergent and face validities of sentiment analysis and reliability of dataset were tested. After obtaining satisfactory results, the GNH by province survey results of Turkish Statistical Institute was compared to results of sentiment analysis for 2013 in order to state the difference between the surveying method and the proposed social media analysis method. Also, GNH by province in the first quarter of 2014 was analyzed. Additionally, relationships between users' account properties and happiness levels were investigated. Results showed that two GNH calculation approaches give similar results for the country-wide GNH levels. As a conclusion, GNH levels in the first quarter of 2014 were calculated as 47.4% happy, 28.4% neutral and 24.2% unhappy. Besides, strong correlations were found between users' happiness levels and Twitter characteristics