What is the best way to do french text analysis in Python?

Question

I want to do text analysis on a french texts to visualise similarities between those texts, the possible class depending on the words that are used. I ask for your help because I just started working with Python and I would like to know the best way to do text analysis in Python taking into account that my texts are in French ?

Are there libraries specially designed for french texts ? The uses would be to clean the data, and further to analyse the data.

I can already :

• Clean: Removed all special characters, put every word in lowercase and suppressed STOPWORDS
• Tag : It's not optimal on french words (Libraries tested : StanfordPostagger, Tag.pos de NLTK).
• Lemmatize : Not optimal (--) with french words(FrenchLefffLemmatizer, WordNetLemmatizer)
• Stemming: FrenchStemmer snowball

What I can't do with French words : pass to singular, pass verbs to the infinitive form...

Answer 1

Spacy library and Treetagger tool (that you can use through treetaggerwrapper library) have good french support.

Example using spacy :

import spacy
nlp_fr = spacy.load('fr_core_news_sm')
text = "J'ai mangé des pommes hier"
tokens = nlp_fr(text)
for token in tokens:
    print(token.lemma_)

Prints :

je
avoir
manger
un
pomme
hier

Treetagger is more difficult to install but this can help you and here is the documentation of the python wrapper.