This is an implementation of the stretch goal from part 1 of MP1, namely to compute the correlation between PPMI (using some window of your choice) and human similarities.
Note that the co-occurrence tables can easily grow very large and even exceed system memory.
This is an implementation of the stretch goal from part 1 of MP1, namely to compute the correlation between PPMI (using some window of your choice) and human similarities.
Note that the co-occurrence tables can easily grow very large and even exceed system memory.