I want to write a program where the user decides whether or not two sentences are similar.
I already made some comparisons with word embeddings and cosine similarity. Now, the user should look through the results of the cosine similarity and decide whether these sentences are really so similar like the automatic comparison suggests.
I cannot find any tool or library that is designed for this and standard input
doesn't let me do what I want.
This is what I want to do:
potential_similars = [(sent1, sent2), (sent3, sent4), (sent5, sent6), (sent6, sent8)]
approved_simil = []
not_approved_simil = []
for tup in potential_similars:
query = input("should " , tup[0], " and ", tup[1], " be regarded as similar? Type Y if yes and N otherwise")
if query == "Y":
approved_simil.append(tup)
elif query == "N":
not_approved_simil.append(tup)