Bipartite Frame Networks in the Analysis of Film: a Case Study Utilizing Commercial Computer Vision APIS

Nabeel Siddiqui, Susquehanna University (USA)

This paper overviews a novel methodology that enhances traditional count-based analysis by leveraging, what I call, bipartite frame networks. Bipartite frame networks consist of two types of nodes. The first node type represents the frames of a film while the second refers to the measurable elements within them (e.g., characters, objects). These nodes are then linked through edges that reflect their cooccurrence within a frame. By applying techniques from network science, bipartite frame networks make it possible for scholars to identify significant frame-to-element relationships, revealing patterns of composition and narrative structure that are not easily discernible through traditional count-based methods.

Nabeel Siddiqui is an Assistant Professor of Digital Media at Susquehanna University’s Communications Department where he also serves as Associate Director for their Center for Teaching and Learning. He specializes in data science, cultural analytics, the digital humanities, the history of information science, communication, new media rhetoric, and science and technology studies.