Network Analysis Of YouTube's Recomendation System From It's Early Years
Abstract
The purpose of the project is to conduct a network analysis in the field of digital media studies. In particular, we aim to analyze the YouTube video recommendation system of the years 2007 and 2008 as a system of recommendation for non-signed-in users, based on “popularity” and relatedness of content, rather than on the personalization from previous activity on the website.
Our idea is to investigate the non-personalized architectural space of the recommendation system of YouTube, in its early years, so as to be able to infer some general patterns in the way it historically made connections between videos, in terms of their relatedness. As such, our study pertains to a specific historical analysis of YouTube’s algorithm: concentrating on its network of contents, beyond the differentiated personalization of the recommendation algorithm.
Our idea is to investigate the non-personalized architectural space of the recommendation system of YouTube, in its early years, so as to be able to infer some general patterns in the way it historically made connections between videos, in terms of their relatedness. As such, our study pertains to a specific historical analysis of YouTube’s algorithm: concentrating on its network of contents, beyond the differentiated personalization of the recommendation algorithm.
Related Hyper-Links
Key Learnings
Network Building
Measuring Network's Properties
Network Visualisation In Gephi
Network Building from Raw Data
Statistical Analysis of Data
