Development of Skyline Query Algorithm for Individual Preference Recommendation in Streaming Data

Abstract

The ability of a recommendation system to deliver relevant outcomes is significantly influenced by its adaptability to the dynamic nature of individual user preferences. Data-streaming-based recommendation systems face substantial challenges in aligning recommendations with rapid shifts in user preferences. Previous research on the development of skyline query algorithms has predominantly focused on processing efficiency and parallel performance optimization yet has not addressed the dynamic nature of individual user preferences—an essential factor for generating relevant and responsive recommendations in streaming data environments. This study aims to develop a skyline query algorithm called Distributed Data Skyline (DDSky) to provide recommendations based on dynamic individual user preferences within data-streaming contexts. DDSky leverages the Recency, Frequency, Monetary, and Rating (RFMRT) model to capture real-time changes in user preferences. This model is integrated with parallel skyline computation and structured to enhance the data processing efficiency on a large scale. The parallel processing approach divides tasks into smaller subtasks executed simultaneously across multiple threads. This strategy enables the simultaneous processing of attributes such as price, distance, and individual user preferences, thereby delivering relevant and responsive recommendations to real-time changes in user preferences. The DDSky algorithm was evaluated using a local dataset from the JALITA application and compared with the Eager algorithm. The results demonstrated that DDSky outperformed Eager, achieving an average recall value of 0.45 and an F1-measure of 0.55, compared to Eager's recall value of 0.33 and F1-measure of 0.47. Furthermore, DDSky achieved an average precision of 0.73, which closely approached Eager's precision of 0.82. Additionally, DDSky exhibited optimal throughput performance for datasets containing up to 10,000 items with high flexibility across various data types. With its unique technical approach, DDSky delivers more responsive and relevant recommendations to dynamic user preferences, establishing its superiority in data-streaming-based recommendation systems.