This study leverages social network analysis to examine the evolution of recommender systems research from 2007 to 2024, focusing on influential research categories and scholarly communities. By constructing a social graph linking highly cited papers to extracted categories, we applied centrality measures such as degree, betweenness, and PageRank along with Louvain community detection to uncover structural patterns and temporal dynamics often missed by traditional methods. Each link represents a paper–category association identified by the IBM NLP Cloud service. The findings suggest that analyzing trends in this field can reveal broader societal and technological shifts. Detected communities demonstrate dynamic, period-specific clustering, reflecting evolving subfields and emerging applications. This network-based perspective offers insights into the shifting knowledge structure in recommender systems and supports anticipating future directions shaped by societal needs.