Developing scalable parallel algorithms for graph analysis remains challenging due to the irregular nature of graphs, which makes load balancing non-trivial. Interaction Nets naturally support irregular parallelism at a low level, making them a promising foundation for graph analysis. Using two Interaction Net implementations, Vine and Inpla, we provided an equivalent set of primitives and functions over quad-tree-represented matrices and vectors, and conducted a preliminary performance evaluation. From this, we derived several observations regarding Vine and Inpla, along with directions for future research.