Advertisement Recommendation System as a Fuzzy Knapsack Problem

Effective advertisement allocation is a critical aspect within recommendation systems, necessitating robust methodologies such as the knapsack problem to accommodate a multitude of constraints. In our study, we delve into the intricacies of addressing the advertisement allocation problem through a fully fuzzy knapsack framework, aiming to provide comprehensive model insights. Our proposed methodology encompasses four key modules meticulously designed to preprocess data for the primary advertisement allocation model. These modules comprise: (i) the utilization of a Fuzzy Inference System (FIS) for detecting advertisement prices, (ii) the segmentation of locations employing Interval Valued Intuitionistic Fuzzy C Means (IVIFCM) clustering, (iii) the segmentation of users through Interval Valued Fuzzy Clustering (IVFC), and (iv) the integration of FastText to estimate the probability of sequential location visits. These modules serve as indispensable components of the fuzzy advertisement allocation problem, which we model as a fully fuzzy knapsack problem. Through our detailed exploration, we aim to offer valuable insights into the operationalization of advertisement allocation amidst imprecise data, thereby facilitating the decision-making process in advertisement display.