Büyük veri teknikleri kullanarak akıllı şehir modelleme

Smart city classification is a complex problem of semantic topics in city modeling. Urban area characteristics are analyzed through hierarchical tasks where location properties are gathered along with municipality records. The improvement of location-acquisition technologies has enabled individuals to incorporate a location aspect into their current online social networks in numerous ways, resulting in the development of location-based social networks (LBSN). These networks integrate the spatio-temporal attribute and real-time engagement of users. Thanks to the contribution of location dimension, LBSNs bridge the gap between the physical world and online social networking services and bring social networks closer to reality. Urban locations are graded in social networks with their popularity, quality, and customer-based interactions evaluated and shared as city insights in different social platforms. As a result, urban semantics can be combined with urban demographics and citizen sentiments in urban places. In this thesis, we focused on the smart city classification and modeling tasks in some important cities in Türkiye. First, we collected Turkish venue data from over ten categories from Foursquare and Twitter platforms. After preprocessing, we trained more than fifteen language models on our dataset and performed a comprehensive evaluation. Our findings revealed that the BERT model outperformed the other models, achieving a classification success rate of over 93%. In addition, by applying content-based and spatio-temporal analysis techniques to our dataset, we reached results that contributed to the advancement of our research. Our results show that location microblogging will be a promising benchmark in city semantics and modeling.