Bipartite Analysis of Blogs and Words

The work proposed in our research aims to analyze a tripartite graph, which consists of authors, blog posts, and keywords within those blog posts. Our research shows that there is promise in constructing a tripartite graph to examine clusters and relationships between authors and ultimately, keywords. We also show that analyzing the second bipartite graph (posts to keywords) reveals correlations between hubs and authorities scores and in-degree for keywords. We also notice that running clustering analysis on the second bipartite graph yields a very low clustering coefficient for words that are used most, similar to blogs that use the most words. In contrast, for words that aren't used as often and for blogs that do not utilize many important words, their clustering coefficient is closer to 1.