Analyzing social media data and predict personal and business information from it
Sosyal medya verilerini analiz etmek ve bitten kişisel ve iş bilgilerini tahmin etmek
- Tez No: 796199
- Danışmanlar: Assist. Prof. Dr. MESUT ÇEVİK
- Tez Türü: Yüksek Lisans
- Konular: Elektrik ve Elektronik Mühendisliği, Electrical and Electronics Engineering
- Anahtar Kelimeler: Troll Detection, Deep-Learning, Artificial Intelligence, Social Media, Bots
- Yıl: 2022
- Dil: İngilizce
- Üniversite: Altınbaş Üniversitesi
- Enstitü: Lisansüstü Eğitim Enstitüsü
- Ana Bilim Dalı: Elektrik ve Bilgisayar Mühendisliği Ana Bilim Dalı
- Bilim Dalı: Elektrik ve Bilgisayar Mühendisliği Bilim Dalı
- Sayfa Sayısı: Belirtilmemiş.
Özet
Özet yok.
Özet (Çeviri)
Social networks become an essential part of our daily life it has been used to share personal thoughts, emotions, and life events and it doesn't ends here new trends and challenges emerges daily shaping new generations thoughts, values and even the way of living, all of these social network features makes it a powerful tool that can be used by organizations, government, and individual politicians to drive the society in certain direction using what known as social trolls or bots which are an automated accounts that belongs to certain parties. Efforts have been made to emphasis the effect of these trolls on the society not no serious actions have been made to eliminate their effects. In this thesis we utilized artificial intelligent to detect social networks trolls by analysis tweets textual contents. three deep learning models (CNN Convolutional neural network, FNN Feed forward neural network, RNN recurrent neural network) has been used in order to detect social trolls. A Comprehensive evaluation has been made in order to assess the deep learning models using two data type (tokenized text sequences and PCA data) the evaluation metrices shows promising results the CNN model achieved the highest accuracy of ( 98.33 %), followed by RNN model with accuracy of (97.245 %), and lastly came the FNN model with accuracy (97.533 %) and that was before utilizing the PCA technique which also have been evaluated but it's performance was low with the following accuracies (82.5 %) for the RNN model followed by (82.4%) for the FNN model and lastly (81.6%) for the CNN model. many evaluation matrices have been discussed in this works and presented the ups and down of each used technique.
Benzer Tezler
- Markov zincirleri ile pazar payı tahmini ve renkli televizyon pazarına ilişkin bir uygulama
Market share estimation of colored TV with markov chains for the period of 1990-1995
BÜLENT MENGÜÇ
- Üniversite öğrencilerinin internet bağımlılığı, bilinçli farkındalık ve can sıkıntısı düzeylerinin incelenmesi
Investigation of internet addiction, mindfulness and boredom levels of university students
RABİA ZAFER
Yüksek Lisans
Türkçe
2021
Eğitim ve ÖğretimKırşehir Ahi Evran ÜniversitesiEğitim Bilimleri Ana Bilim Dalı
DR. ÖĞR. ÜYESİ ÖNDER BALTACI
- Mikrobilgisayar kullanarak tarım işletmesi analizi üzerine bir çalışma
Başlık çevirisi yok
BÜLENT MİRAN
- İhracatın sektörel analizine ilişkin fonksiyonel bir araştırma
Functional research on analysis of sectoral export
AYŞENUR TOPÇUOĞLU
- Test sınavlarının güvenirliğinin araştırılması ile ilgili bir çalışma
An Investigation on the reliability of multiple,choice examinations
CANDAN İĞCİ
Yüksek Lisans
Türkçe
1987
Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve KontrolEge ÜniversitesiBilgisayar Mühendisliği Ana Bilim Dalı
DOÇ. DR. M. ÖZEL ERGEN