Week 2 Assignment 2 Exercise 2 - complete_analogy

Course Q&A Deep Learning Specialization Sequence Models
1

When test for the complete_analogy function I find the result is not as expect

italy -> italian :: spain -> spanish
india -> delhi :: japan -> tokyo
man -> woman :: boy -> girl
small -> smaller :: large -> smaller

The last line ‘large’ should be analogized to ‘larger’ rather than ‘smaller’
Also I test for ‘big’ and it is analogized to be ‘competitors’
Even though I passed all the test and got 100/100
I just wonder why and how to solve it?
Thanks

2

This is a shortcoming of the method. A single scalar value (the cosine similarity) cannot 100% accurately represent the association between two words without considering the syntax and other uses of those words.

This assignment uses the smallest of the available GloVe models (400K vocabulary and 50 dimensions).

Much larger sets of word vectors are available. They may may work better.

See this for reference: