In the modern era of personalization, recommender systems are expected to tailor item recommendations to meet user (U) preferences, helping them explore and discover relevant items (I). However, the complexity and subjectivity of user preferences pose challenges due to their inherent dynamics. To address this, user interactions and overlaps (U-U, U-I, and I-I), along with submitted feedback such as reviews and ratings, are leveraged to achieve a higher degree of personalization and other low-level tasks like product search, product rating, cold-start problems, and long-tail issues. Among these challenges, the long tail is a significant concern, as popular items (the “long tail”) are often overlooked, severely affecting overall diversity. In this paper, a hybrid recommendation strategy has been proposed by combining LDA and EMD techniques to address the long tail issue in recommendations primarily. In the first phase, LDA generates topic distributions, and EMD estimates similarity among these distributions, i.e., user preferences to item descriptions, to align recommendations with the user’s broader interests, not just exact matches. We empirically evaluate the efficacy of this LDA-EMD hybrid strategy, designed to assist customers in finding products that fulfil their personalization needs. The experimental evaluation was conducted on an Amazon e-commerce dataset comprising 1,466 products spanning 16 genres, with varying levels of aspect information. The improvement in recall for a set of 20 and 50 products is 0.1500 and 0.2500, respectively. Integrating LDA with EMD not only identifies direct matches but also discovers items that are similar in more complex ways, which were missed by previous models, resulting in improved recall. Additionally, other metrics, such as NDCG@20 and NDCG@50, demonstrate the superiority of the proposed strategy over baseline models. The empirical evaluation confirms the feasibility of integrating LDA and EMD to enhance recommendation accuracy and user satisfaction, thereby contributing to the advancement of personalized recommender system.