4. Consider two product preference networks shown below involving 10 products . If you calculated two sets of unweighted PageRank scores from the two networks A and B, which set would most likely show a higher correlation with sales data ? Why ? Do not actually calculate PageRank scores ; instead answer this question conceptually . (10 points ] Preference Network A Preference Network B if we calculate page rank for Network B, we will observe that many product pages refer to further product pages . If many pages intersect at a product page , that page is important and of higher rank . And when this page in turn refers to another page , the effect of page rank is stronger . Since each page in Network B refers to many other pages in the network , the effect is stronger , and the page rank is of higher value for each product . Hence , the sales data will show a higher correlation for network B. For each page , there is approximately 5-6 referring pages . If all pages had the same number of referring pages , the page rank would have been equal for all of them . We notice that XJ is an in- out node , due to which it has high page rank being linked to a heavy node which in turn flows to all the corresponding nodes . Incase of Network A the overall Page rank is reduced due to high outgoing and transfer of the effect . Hence , the overall page rank at each page decreases