Discovering skill synergy patterns from employment data using bitwise inverted frequent itemset mining

Frequent itemset mining (FIM) plays an important role in analysing large-scale employment data to identify job-skill associations, especially as higher education institutions seek to align academic programmes with evolving labour market demands; however, conventional FIM algorithms such as FP-Growth and Eclat often incur substantial computational and memory overhead when applied to high-dimensional and sparse datasets. To address these challenges, this study proposes BI-FIM (Bitwise Inverted Frequent Itemset Mining), which leverages inverted index structures and bitwise AND operations to compute itemset support efficiently without repeatedly scanning the database. Combined with a depth-first search (DFS) framework and dynamic intersection ordering (DIO), BI-FIM reduces redundant computations and improves pruning efficiency, yielding favourable performance relative to selected classical and recent baselines, including FP-Growth, Q-VIPER, the NegFIN-related vertical closed itemset method NEclatClosed, and the CLA-based FP-tree method CAFP, across multiple benchmark datasets (Kosarak, Retail, Chess, and Connect) and a real-world employment dataset from Guizhou Province (2014–2024). Building on the core design of BI-FIM, the extended SW-BIFIM (Synergy-Weighted BI-FIM) model incorporates skill synergy weighting to highlight both frequent and complementary skill sets, thereby providing a more application-oriented view of skill combinations. The results suggest that the proposed framework can serve as a useful analytical tool for employment data mining and may support downstream analysis in curriculum design, talent recruitment, and related policy evaluation.