For video clustering YouTube is using the following
aspects
a) Keyword popularity
b) Viewer Engagement
c) User browsing history
These metrics are used in displaying video content by
relevance i.e. by your choice. The problem definition does not
end here. Finding out the content objects of interests and then
associating them in appropriate clusters for storage. The
information does not necessarily always have to be there.
Based on level of information the algorithm adjusts itself and
its functions.