Hi All:
In RowSimilairtyJob, I see the main functionality of countObservations
job is to aggregate the number of users for each item, is that right? if
so, why not directly calculate the counts, just like the WordCount logic.
In current implementations, we initialized a RandomAccessSparseVector,
which may cause the OutOfMemory issue when the number of users is large. Am
I understanding correctly? thanks.Best Wei
