Decoding the Popularity of TV Series: A Network Analysis Perspective
Melody Yu
TL;DR
This study frames TV episode quality as a function of character interaction structure by constructing weighted episode graphs from segment-level networks and computing a broad set of network metrics. Using Spearman correlations, it demonstrates that certain metrics relate to IMDb review scores, with distinct patterns across Game of Thrones, House of Cards, and Breaking Bad. The findings suggest that episodes featuring cohesive subgroups or controlled character prominence can influence audience reception, offering producers a quantitative lens on narrative dynamics. The methodology provides a data-driven approach to analyze narrative structures in TV and informs potential production adjustments for future seasons.
Abstract
In this paper, we analyze the character networks extracted from three popular television series and explore the relationship between a TV show episode's character network metrics and its review from IMDB. Character networks are graphs created from the plot of a TV show that represents the interactions of characters in scenes, indicating the presence of a connection between them. We calculate various network metrics for each episode, such as node degree and graph density, and use these metrics to explore the potential relationship between network metrics and TV series reviews from IMDB. Our results show that certain network metrics of character interactions in episodes have a strong correlation with the review score of TV series. Our research aims to provide more quantitative information that can help TV producers understand how to adjust the character dynamics of future episodes to appeal to their audience. By understanding the impact of character interactions on audience engagement and enjoyment, producers can make informed decisions about the development of their shows.