İTÜ Bilişim Enstitüsü Coğrafi Bilgi Teknolojileri Lisansüstü Programı öğrencisi İrem ERKUŞ tarafından Doç. Dr. Arif Çağdaş AYDINOĞLU danışmanlığında hazırlanan “Sosyal Medyada Kullanıcı Profillerinin Belirlenmesinde Konumsal İstatistik Tekniklerin Kullanilmasi” konulu yüksek lisans tezi tamamlanmıştır.
ABSTRACT
Social media became hot topic after millennium. Beside communication techniques such e-mail and mobile phone, social media platforms have started to be used in various different social and working discipline, to illustrate this, Facebook for social communication, Twitter for short messages, and Flickr for photo sharing, and so on. Most social media platform has possibility to locate any social media activity spatially. Defining any activity with location or position in social media means that distribution of social media users can be analyzed spatially to determine user profiles.
The purpose of this study is to determine user profiles of social media platforms with the examples of online dating site. The data of users are integrated with in GIS environment using spatial statistics techniques.
In methodology part, spatial statistical methods were used to understand the data relationships. Firstly, the distribution of patterns is analyzed spatially to determine data outliers. Then, user profiles are analyzed by age, gender, income, educational level and occupation to examine socio-demographic relationships. Lastly, target areas are determined by using cluster analysis while relations of the data is analyzed by using regression techniques.
The applications consist of four parts. As working principles of GIS, related spatial data was captured, queried, analyzed and visualized in ArcGIS program. Data about ten thousands of users who often use this online dating site was retrieved from the MySQL database by using spatial data query. The distribution and outliers of the data were determining by using Exploratory Spatial Data Analysis (ESDA) tools which are histogram, semivariogram,voronoi map, and QQ plots. Furthermore, the population of Istanbul was examined with spatial statistical analysis based on the data of Adress based Population Registration System and social media site. The results were interpreted and showed in thematic maps, and presented results in a diverse range of visualization options such as reports, statistical graphics and tables. The third part examine the hot-spot analysis to find the clustered area of social media users by using mapping cluster toolset. The final part demonstrate the OLS and GWR regression methods to identify regression model by using spatial relationships tools. In this way, this study contributes to research about social media a new developing discipline. This gives a perspective of user profile in view of education and occupation by using spatial statistical approaches. All in all, this thesis gives an idea about social media and usable statistical techniques.
Prepared by:
İrem ERKUŞ
Thesis Adviser:
Assoc. Prof. Arif Çağdaş AYDINOĞLU, PhD (İTÜ / GYTE)
Jury Members:
Prof. Taşkın KAVZOĞLU, PhD (GYTE)
Assist. Prof. Himmet KARAMAN, PhD (İTÜ)
Date of Defence:
28 August 2014