Пошук академічних текстів

Інформація × Реєстраційний номер 2124U004702, Матеріали видань та локальних репозитаріїв Категорія Препринт Назва роботи Application of Thompson Sampling In Web Advertisement Filtering Автор Rodzin Taras Дата публікації 01-01-2024 Постачальник інформації Український католицький університет Першоджерело https://hdl.handle.net/20.500.14570/4830 Видання Опис Digital marketing has demonstrated a significant growth in popularity during the last decade due to the huge variety of possibilities and flexibility the online ad- vertisement platforms offer. However, profitability and competitiveness in such an environment require efficient resource management and adaptation to the fast- changing customer behavior and interests. Effective advertisement allocation and optimal spending control form the primary problem this work seeks to resolve. The thesis considers the aforementioned setup as one that can be solved as the Multi- Armed Bandit problem using the Thompson Sampling algorithm. The work utilizes the modified principles of the aforementioned approaches to construct the adver- tisement filtering algorithm and evaluate its performance on real-life data. Додано в НРАТ 2025-05-09 Закрити