Хронологічне моделювання тем і аналіз емоційного забарвлення текстів на основі ML.NET

Abstract

Following work is devoted to developing a system for chronological topic representation and sentiment analysis using ML.NET. To structure large-scale text data, the system organizes topics based on keywords, entities, and contextual patterns while analyzing emotional tones. Data is collected from news, blogs, and social media, undergoing preprocessing and feature extraction using named entity recognition and sentiment analysis. Transparency is ensured through SHAP and LIME. The system leverages ML.NET for seamless .NET integration, supporting real-time analysis and scalability.