Explainable Deep Learning Architectures for Product Recommendations
Boominathan Perumal, Swathi Jamjala Narayanan, Sangeetha Saman · 2024
Due to the recent pandemic and the convenience offered, online shopping platforms and websites have surged in popularity. These platforms give customers the privilege to stay at home and shop in a comfortable way. Such websites and platforms can be aided with support systems that can assist customers in many shapes and forms. Explainable recommendations are gaining popularity in both the research and industrial worlds. Many times, models that lack explanations may affect user satisfaction and trust in the product recommendation system. With that specific goal, this chapter presents a brief overview of existing approaches such as machine learning and deep learning-based systems, along with highlighting the position of an explainable Product Recommendation System that will assist users in purchasing from and using the platform. To distinguish existing explainable recommendation studies, we propose a two-dimensional taxonomy: One dimension is model-intrinsic, whereas the other is model-agnostic. As a result, it contributes to the visibility, trustworthiness, effectiveness, and customer satisfaction of recommendation systems. It also makes it easier for system designers to examine, debug, and enhance the recommendation algorithm.