Applied Intelligence and Computing
Enhancing Multi-Criteria Recommendation Systems with Hot Deck Imputation and User-Specific Similarity Measures
Authors: Nithish Chouti, Sakshi Kusale, Sanket Mishra, Shashank Rajora, Bam Bahadur Sinha, Prabhu Prasad B M and Manjunath K V
Publishing Date: 02-01-2025
ISBN: 978-81-955020-9-7
Abstract
Recommender systems are pivotal in the modern digital landscape, where consumers are inundated with choices, leading to decision fatigue. These systems offer personalized suggestions, alleviating the burden of choice by tailoring recommendations to user preferences. One of the challenges faced by recommender systems is dealing with sparse user-item matrices, often resulting from limited user preferences. To overcome this, we adopt a novel approach by categorizing users into three groups based on their rating activities: cold start users, middle users, and heavy users. This categorization enables us to provide more accurate and personalized recommendations. Furthermore, we utilize multi-criteria similarity measures to identify user preferences. Unlike many existing recommender systems that focus on average or overall criteria, our approach considers individual user behaviors, leading to more precise recommendations. It is worth noting that while multi-criteria recommender systems exist, they do not often group users together to enhance results. By categorizing users, we can better understand their preferences and tailor recommendations accordingly. To address sparse matrices, we employ hot deck imputation, filling in missing values by borrowing information from similar records. This strategy ensures that our recommendations are based on a more complete dataset, improving the overall accuracy and effectiveness of our recommender system.
Keywords
Multi-criteria Recommender systems, Collaborative filtering, Hot deck imputation, Similarity measures.
Cite as
Nithish Chouti, Sakshi Kusale, Sanket Mishra, Shashank Rajora, Bam Bahadur Sinha, Prabhu Prasad B M and Manjunath K V, "Enhancing Multi-Criteria Recommendation Systems with Hot Deck Imputation and User-Specific Similarity Measures", In: Mukesh Saraswat and Rajani Kumari (eds), Applied Intelligence and Computing, SCRS, India, 2025, pp. 313-324. https://doi.org/10.56155/978-81-955020-9-7-30