Multi-task BERTurk ile Türkçe e-ticaret yorumlarından duygu + departman analizi ve 0-100 marka itibar skoru.
Fine-tuning BERTurk (encoder), Turkish GPT-2 (decoder) and mBART-50 (encoder-decoder) head-to-head on Turkish movie-review sentiment — 9 experiments, Streamlit demo.
Istanbul Sentiment Analysis v2. Upgraded from generic savasy/bert to custom fine-tuned Emirhan41/bert-base-turkish-128k-istanbul-sentiment model. Features irony detection, semantic duplicate filtering, and Flask dashboard.
End-to-end Turkish sentiment analysis on bank app reviews — TF-IDF baseline vs fine-tuned BERTurk, human-verified label quality audit, live Streamlit demo.
🇹🇷 Türkçe'ye duyarlı kalıcı ajan hafıza katmanı: dedup, getirme, supersede, unut. | Turkish-aware agent memory engine (dedup, retrieval, supersede, forget).
🇹🇷 Türkçe hibrit arama RAG: dense embeddings + BM25 + Reciprocal Rank Fusion. | Turkish hybrid-search RAG (dense + BM25 + RRF).
Yapay Zeka (BERTurk) destekli marka itibar skoru
An end-to-end Turkish e-commerce sentiment analysis project using the BERTurk model, featuring a Streamlit-based interactive customer insights dashboard.
NLP-powered hybrid book recommendation system built with React Native, Flask, TF-IDF, BERTürk and Collaborative Filtering.
BERTurk fine-tuning ile Türkçe duygu analizi (Negative/Notr/Positive). 490K yorum, test macro-F1 0.956. Uçtan uca pipeline: EDA → tokenizasyon → eğitim → dokunulmamış test setinde tek atış değerlendirme + kaynak bazlı hata analizi.
🇹🇷 Türkçe'ye özel, kullanıma hazır RAG araç seti (BERTurk + eklemeli dile uygun parçalama). | Turkish-tuned RAG toolkit.
Turkish 8-class text emotion classifier fine-tuned from BERTurk: dataset building, ablations, statistical evaluation and a Streamlit demo.
Transformer-based Turkish movie review sentiment analysis using BERTurk, Turkish GPT-2, and mT5.
To associate your repository with the berturk topic, visit your repo's landing page and select "manage topics."