I ran pagerank on a multi-directed graph. Each node in my graph has an integer value indicating my prior knowledge regarding its importance. I use this value as personalization value and run nx.pagerank_numpy. However, I have problem understanding the resulting scores of nodes and how I should interpret them.
For example, I see the following situation: 'lag' has a higher pagerank score than 'software' while I believe it shouldn't. My reasons are the following:
1)the personalization values for 'lag' and 'software' are 8 and 32 respectively. 2)the number of inlinks and outlinks are higher for 'software' (in=12) than 'lag' (in=8). 3)the pagerank scores of nodes that are linking to 'software' and the weights of the edges are higher than those linking to 'lag'.
You can download the graph file and the code I am running. How can I interpret the fact that 'lag' is getting higher PageRank score than 'software'? Is there something that I am missing?
Thanks