What lemmatizer can i use for arabic text using python?

Question

How can I get lemmas for Arabic words? I tried the ISRI Arabic Stemmer from NLTK but it returns roots of words:

from nltk.stem.isri import ISRIStemmer
st = ISRIStemmer()
print st.stem(u'اعلاميون')

It returns the root علم and i want the lemma اعلامي

Answer 1

The state-of-the-art is Farasa Lemmatizer.

Farasa Lemmatizer outperforms MADAMIRA Lemmatizer based on accuracy. It gives +7% relative gain in accuracy above MADAMIRA in lemmatization task.

You can read more about Farasa Lemmatizer from the following link: https://arxiv.org/pdf/1710.06700.pdf