https://github.com/soldni/wndist

Calculate distance between two lemmas in wordnet
https://github.com/soldni/wndist

Last synced: 6 days ago
JSON representation

Calculate distance between two lemmas in wordnet

• Host: GitHub
• URL: https://github.com/soldni/wndist
• Owner: soldni
• License: mit
• Created: 2015-08-17T16:57:56.000Z (about 9 years ago)
• Default Branch: master
• Last Pushed: 2015-08-18T17:09:27.000Z (about 9 years ago)
• Last Synced: 2023-08-17T13:53:49.637Z (about 1 year ago)
• Language: Python
• Size: 133 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# WNDist

problem: finding the distance (i.e., shortest path length) between two nodes

in the NLTK implementation of WordNet is:

##### A. slow

```python

from nltk.corpus import wordnet as wnnltk

nltk_exec_time = time.time()

dists = []

for s1, s2 in product(wn.synsets(w0), wn.synsets(w1)):

    dists.append(s1.shortest_path_distance(s2))

nltk_exec_time = time.time() - nltk_exec_time

print(nltk_exec_time)

>>> 4.06659889221

```

##### B. not flexible

There is no way to specify a subset of relationships to consider when evaluating the distance.

------

This library solves both issues: it is fast and flexible. It uses the excellent [NetworkX](https://networkx.github.io), plus a combination of pickling and shelving to achieve greater speed. Here's the same task in WNDist:

```python

from wordnet_distance.wndist import WNDist

# specify a cache folder where to store the cached results,

# as well as provide the location of the wordnet dicitonary

wn = WNDist(cachedir='cache', sourcedir='dict')

w0, w1 = 'war', 'peace'

custom_exec_time = time.time()

wncstm.lemmasdist(w0, w1)

custom_exec_time = time.time() - custom_exec_time

print(custom_exec_time)

>>> 0.0069420337677

```

The source dictionary is avaiable from the official [WordNet page on Princeton's website](http://wordnet.princeton.edu/wordnet/download/)