High Utility Itemset Mining (HUIM) extends Frequent Itemset Mining. HUIM identifies itemsets with significant impacts on metrics like sales or customer satisfaction. With the advent of Big Data, there is a demand for algorithms to extract these High Utility Itemsets (HUIs) efficiently. While the size of the output can be controlled via well-chosen support and confidence, the quality of the results can be altered. To address this challenge, various approaches generate more compact representations (CRs) of high-utility itemsets (HUIs), significantly reducing the output size while maintaining the integrity of the information and ensuring that the derived rules retain their high significance. Among these CRs we find closed, maximal, minimal itemsets, and generators. While CRs-based HUIs are typically compared internally, our study takes a unique approach with an external comparison. We analyze algorithms designed for HUIM without CRs (e.g., FHM, HUI-Miner, UPGrowth) against those incorporating various CRs (e.g., MinFHM, CHUI-Miner, CHUI-Mine, CLS-Miner, GHUI-Miner). Experiments on two large-scale datasets (Cosmetics and Chicago Crimes) evaluate performance based on runtime, memory size, and extracted HUIs. The findings indicate that algorithms utilizing CRs based on closed itemsets surpass those relying on generators; these latter yielding the least efficient results. Specifically, FHM and MinFHM demonstrate efficient extraction of HUIs. Additionally, MinFHM excels in runtime, memory usage, and in the size of extracted HUIs.