I would approach this as follows:
from collections import defaultdict # using defaultdict makes the sums easier
correlations = defaultdict(int) # default to int (i.e. 0)
for i1, i2, correl in strScoresDict: # loop through data
correlations[i1] += correl # add score for first item
correlations[i2] += correl # and second item
output = sorted(correlations,
key=lambda x: correlations[x],
reverse=True) # sort keys by value
Note, however, that the output is
output == ['item2', 'item1', 'item4', 'item3']
As the total correlations are
{'item1': 220, 'item3': 100, 'item2': 240, 'item4': 200}
You can read about defaultdict
here.
solved Sorting a list based on associated scores [closed]