A type of artificial neural network architecture where information flows in one direction, from the input layer through one or more hidden layers to the output layer. Each layer consists of nodes (neurons), and the connections between nodes have associated weights that are adjusted during training to learn patterns in the data.
Published in Chapter:
TABST: A Next-Gen AI Model Unveiling Personalized Recommendations and Targeted Marketing
Diotima Nag (Technology Campus, University of Calcutta, India) and
Unmesh Mandal (Bidhan Chandra College, Rishra, India)
Copyright: © 2024
|Pages: 18
DOI: 10.4018/979-8-3693-2165-2.ch019
Abstract
Recommendation systems (RS) are indispensable for personalized marketing, shaping user experiences in online shopping and OTT media. RS aims to deliver timely, relevant content with minimal user effort, drawing insights from prior data, including preferences, timing, and contextual choices. This study explores the temporal dimension of RS using the Transformer model. The proposed time aware behavior sequence transformer (TABST) predicts a target item's rating on a specific day and time by analyzing users' sequential timing history. TABST employs transformer encoders to extract latent temporal and sequential information, recommending a target item with an optimal consumption time. Comparative evaluations validate TABST's superiority over state-of-the-art algorithms, positioning it as a cutting-edge solution for personalized recommendations. In summary, this research contributes to the evolution of RS by introducing TABST, addressing personalized choices within the temporal and contextual behaviours of users, and ultimately enriching the user experience.