Answering k-Most Promising Products based on Time Intervals on Multi-Dimensional Time Series Data using R-Tree-based Indexing

Many applications, such as recommender systems and business intelligence platforms, rely on temporal attributes in time-series data to enhance recommendation accuracy and relevance. For example, an apparel manufacturer aiming to identify the top k-products that could attract customers and increase sales during a specific period—such as the rainy season from October 2023 to April 2024—would see different results if the time interval was adjusted to the summer. Temporal data plays a crucial role in data analysis, as it helps understand seasonal preferences and identify yearly patterns. However, the existing methods cannot adequately answer time interval-based queries. This paper formulates the problem of product selection within a specific time interval, named k-most promising products based on a time interval (KMPPTI). Given a dataset of products from various manufacturers, a customer dataset, an integer k, and a defined time interval, we identify the top k-products from the entire dataset that achieve the highest market contribution scores within a specified period. We propose an approach that efficiently uses two types of skyline computation—dynamic skyline and reverse skyline—to answer the KMPPTI query. We employ the R-Tree index structure to enable the pruning of large portions of the search space, which is crucial in skyline computation. Our paper demonstrates the effectiveness of the proposed algorithms through comprehensive experiments on both real and synthetic datasets. The experimental results indicate that R-Tree-based indexing significantly reduces the algorithm's execution time by approximately 68%.