A closer look a the pairwise similarity based graphs and ways to capture relationships beyond pairs

This extended abstract aims to investigate pairwise similarity-based matrices and the graphs they imply with examples and elaborates on ways of capturing higher order relationships. At this stage, we articulate several emerging questions rather than presenting conclusive results: Can repetitive patterns be observed in matrices, irrespective of the similarity measure used? How does the resulting graph change with different threshold values? Can higher-order relationships be easily detected and how are they connected to the matrices? The next step involves exploring the impact of these differences on learning outcomes for graph-based tasks and making comparisons to other data representations, such as hypergraphs or combinatorial complexes for different domains.