Closed hailiang-wang closed 6 years ago
From https://lyfat.wordpress.com/2012/05/22/euclidean-vs-chebyshev-vs-manhattan-distance/
Thanks . I have tested on several different similarity measures, and found that Euclidean distance is the best. And I merged some other features such as synonys' word embeddings , unigram overlap, POS tag ...
Commit c0a477555174baa5f5c03243c3bea79aa206b8e9
I just test on the 778 pairs of sentences (link: https://github.com/fssqawj/SentenceSim/blob/master/dev.txt)
and get these curves:
description
use advanced method to compute similarity for two sentence.
related with #4
solution
leverage adv distance measurements.