Comparison—the act of finding similarities and differences—is rooted as a fundamental analysis task. However, this task is non-trivial when analyzing large-network or high-dimensional datasets. In such cases, it is difficult to identify the key contributing factors to the similarities or differences of different groups from all possible relationships or attributes. Representation learning (RL) can potentially amend this challenge by extracting influential factors for a particular aspect within a dataset (e.g., data variance). However, existing RL methods have limited capabilities for comparative analysis due to their inability to cover a wide range of data types and analysis targets. In this talk, I address the challenges of data group comparison by concurrently using interactive visualization and new RL techniques that utilize contrastive learning—a new emerging machine learning scheme that finds salient patterns in one dataset relative to another. I will demonstrate the effectiveness of this combined approach by analyzing real-world network and high-dimensional datasets. This talk discusses how new RL schemes  further facilitate uncovering once-hidden data patterns through the fundamental interaction of visual comparisons.