Efficient and accurate personalized product recommendations through frequent item set mining fusion algorithm

Abstract

In the realm of personalized product recommendation, addressing the challenges of sparse data and “cold start” has been the primary focus. However, filtering invalid information amidst the overwhelming data on e-commerce platforms remains an underexplored issue. This paper proposes a fusion recommendation algorithm based on frequent item set mining to tackle this problem by compressing the commodity data set and identifying the frequent commodity set. The algorithm not only improves time efficiency by reducing the number of candidate frequent item sets but also generates more accurate recommendations by calculating user-commodity interest rankings and recommending similar products. We first present the existing problems in fusion recommendation algorithms based on frequent item set mining, such as redundant rules, low recommendation accuracy, and the inability to explore deep connections between users and products. Next, we introduce our proposed algorithm, which involves filtering the commodity data set, calculating user-commodity interest rankings, and defining similar product recommendation rules. The algorithm's effectiveness is demonstrated by its ability to adapt to users' dynamic preferences and capture their changing interests in real-time. A comparative analysis using our algorithm and other data mining algorithms reveals a reduction in the number of frequent commodity data sets and weighted frequent item sets, leading to decreased algorithm operation time. This research contributes to the development of more efficient and accurate personalized product recommendation algorithms, enhancing user experience on e-commerce platforms.