When I calculated the similar words of careful + brave with word2vec, it felt unexpectedly valid

Introduction

I don't know how many brews it will be, but I used django and word2vec to vectorize words in N dimensions to create a web page that searches for the most similar words in cosine similarity and publishes it on PythonAnywhere. Did.

URL of similar word search page

How to use

-Just enter words in the text box by combining them with + or-and press Extract Synonyms.

Example of use

I made it with great effort, so let's try carefully + brave man.

image.png

Yup. Isn't it so wrong? (confusion)

Strong + brave ... image.png

Perhaps the word brave itself is a bad word! ?? Let's add justice to make it harder ...

image.png

It is useless. Justice lost to the brave ... Do you want to try it with a hero?

image.png

Oh, Oita decent!

Conclusion

The hero is not good. (Rogue) If you rely on it, you should be a hero.

The model I used this time

The learned Japanese model of word2vec will be released

Recommended Posts

When I calculated the similar words of careful + brave with word2vec, it felt unexpectedly valid
What I did when I couldn't find the feature point with the optical flow of opencv and when I lost it
Be careful of the type when making an image mask with Numpy
When I tried to change the root password with ansible, I couldn't access it.
[Python] I introduced Word2Vec and played with it.
[Python Data Frame] When the value is empty, fill it with the value of another column.
It became TLE when I confirmed the operation with the print function in the competition pro
When I crawl the webapi that appears during rendering, it was played with CORS
What I did when I was angry to put it in with the enable-shared option