High-utility sequential pattern mining in incremental database

Previous algorithms designed for efficient mining of sequence patterns have primarily focused on processing static databases. However, in the context of dynamic database mining, where new data are constantly added, rescanning the entire database to update the information becomes necessary. This maintenance and update process consumes significant time and resources, leading to delayed responses. To address this issue, this paper proposes an incremental mining algorithm called Pre-HUSPM, which leverages the concept of pre-large to insert new sequences into the dynamic database while preserving the discovered efficient sequence patterns. Furthermore, a novel threshold, denoted as \(SWU_{max}\) $SW{U}_{\mathit{max}}$ , is introduced to minimize the frequency of database rescans and enhance the algorithm’s speed. The experimental results show that the algorithm greatly reduces computation time and resource consumption, enabling the algorithm to respond faster to data changes and generate new mining results. This algorithm aids manufacturers in designing and producing products that align with customer preferences based on previous products, thereby improving operational efficiency and guiding customers toward wise purchasing decisions, ultimately resulting in higher profits for the company.