3
Arşivlenen Tez
0
DOI Atanmış
0%
DOI Oranı
Anabilim Dalı
Politika değişikliklerinin cinsiyet algıları üzerindeki etkisi: Türkiye'nin İstanbul Sözleşmesi'nden çekilmesinin sosyal medya üzerindeki yansımalarının doğal dil işleme (NLP) yöntemleri ile analizi
In May 2011, Türkiye became the first country to sign the Istanbul Convention, a groundbreaking treaty aimed at combating violence against women and promoting gender equality. However, in March 2021, Türkiye's ruling conservative party decided to withdraw from the Convention, sparking widespread societal debate and reflecting deeper ideological divisions. This study investigates the shifts in public sentiment regarding women's issues in Türkiye following the country's withdrawal from the Istanbul Convention. Using natural language processing (NLP) techniques and BERT-based deep learning models, a large dataset of 55 million Turkish tweets from over 200,000 users is analysed to examine the changes in discourse related to women's issues across different ideologies, genders, and age groups. To enhance the analysis, GPT-based large language models were integrated to annotate tweets based on challenges women face, such as sexual harassment, domestic violence, and gender-based discrimination, revealing how different topics elicited varied emotional tones and ideological responses. The analysis reveals significant emotional and ideological divides, challenging binary assumptions that pro-equality ideologies would become more negative while anti-gender ideologies would shift positively. Instead, unexpected groups—particularly Islamist women and feminist men—demonstrated the sharpest increases in negativity. This suggests a notable integration of new voices into women-related discourse, defying ideological expectations. While feminist discourse remained consistently critical, it did not escalate significantly, indicating a persistently high baseline of concern rather than a drastic spike. This study further reveals complex interaction between religion, politics, and gender discourse in Türkiye, while Conservatism remained sentimentally stable, Islamism underwent the most pronounced emotional transformation. This study sheds light on the complex emotional landscape where gender and ideology interact in unexpected ways, offering more profound insight into cultural backlash and the dynamics of public discourse on gender equality in polarized contexts.
Ideologies, emotions, and anti-immigrant stance in the context of türkiye: a social media analysis through natural language processing (NLP) techniques
In recent years, global migration has been on the rise, leading to intensified debates and growing anti-immigrant sentiment in many parts of the world. Türkiye, as one of the leading host countries for immigrants and refugees, has become a focal point in these discussions. Despite extensive research in Western Europe and the U.S., where right-wing ideologies are often linked with anti-immigrant views, the dynamics in non-Western countries like Türkiye are less explored. This study aims to investigate whether there is an interaction between ideological positions and emotional responses in shaping public stances toward immigration in Türkiye, incorporating Affective Intelligence Theory (AIT), which emphasizes the role of emotions—particularly fear, anxiety, and anger—in shaping political judgment and behavior. The research utilizes social media data sourced from Twitter to analyze the interaction between ideologies and emotions in shaping anti-immigrant sentiment. Natural Language Processing (NLP) techniques, including fine-tuned pre-trained BERT-based models like BERTurk and TurkishBERTweet, are employed to classify immigration-related tweets and detect pro- and anti-immigrant stances. The findings challenge the conventional association between right-wing ideologies and anti-immigrant sentiment commonly observed in Western Europe and the U.S., revealing a more complex range of views across both right- and left-wing ideologies in Türkiye. The study further highlights how emotions, as explained by Affective Intelligence Theory, can either reinforce or shift ideological stances on immigration. This interaction between emotions and ideologies provides deeper insights into the complex dynamics shaping immigration discourse in Türkiye, offering a nuanced understanding of public sentiment.
Mahalle düzeyinde kentsel çevre kalitesine yönelik memnuniyette mekansal ve zamansal değişimler: Kadıköy örneği
Urbanization has transformed cities into dynamic hubs while introducing challenges that impact residents' quality of life (QoL). This study aims to develop a localized measure of satisfaction with environmental quality in Kadıköy, Istanbul, to support local governments in improving urban living conditions. Employing a multi-method approach utilizing computational social science methods, we integrated social media data from Ekşi Sözlük with social listening data from the Kadıköy Municipality call center. Over a multi-year period, we applied sentiment analysis and topic modeling (BERTopic) to analyze public opinions and complaints. The findings reveal key issues affecting QoL, including concerns related to public spaces, walkability, noise pollution, and environmental factors. Spatial and temporal analyses highlighted specific areas and time periods with heightened dissatisfaction, influenced by factors such as population density and seasonal variations. By identifying the dimensions of QoL and their spatiotemporal patterns, this study provides a comprehensive understanding of urban environmental satisfaction in Kadıköy. These insights can inform targeted interventions by local authorities. Integrating both bottom-up (subjective perceptions) and top-down (objective indicators) approaches, this research contributes to the QoL assessment literature and demonstrates the value of combining diverse data sources in urban studies.