ilkayO/yildizsezar-turkish-reviews

YıldızSezar是一个土耳其语电子商务评论数据集，包含真实和LLaMA生成的合成数据。它是一个全面的土耳其语客户评论集合，来自电子商务平台，标注了1到5星的评分，旨在训练高精度的多类情感分析模型。为了解决类别不平衡问题（多数为5星或1星评论），作者使用LLaMA-8B-DPO生成了超过90万个针对少数类（2、3、4星）的合成土耳其评论，使数据集高度平衡且形态复杂。数据集总合成样本约90万，任务为5类情感分析/星评级预测，语言为土耳其语。This dataset is a comprehensive collection of Turkish customer reviews from e-commerce platforms, labeled with 1-to-5 star ratings. It was developed to train highly accurate multi-class sentiment analysis models. To solve class imbalance (mostly 5-star or 1-star reviews), it was augmented by generating over 900,000 synthetic Turkish reviews targeting minority classes (2, 3, and 4 stars) using LLaMA-8B-DPO, resulting in a highly balanced and morphologically complex dataset. Total synthetic samples: ~900,000. Task: 5-Class Sentiment Analysis / Star Rating Prediction. Language: Turkish.