Skip to main content

Extract/Replaces keywords in sentences.

Project description

This code is not ready yet.
Please use till I finish this.

This code is successor to


Flash Text is a python library that is loosely based on `Aho-Corasick algorithm

The idea is that say we have a corpus of terms/keywords. We want to extract any of the term from the corpus present in a sentence by making on pass on the sentence.

Basically say I have a vocabulary of 10K words and I want to get all the words from that set present in a sentence. A simple regex match will take a lot of time to loop over the 10K words.

Hence we use a simpler yet much faster algorithm to get the desired result.


pip install flashtext


# import module
from synonym.extractor import SynonymExtractor

# Create an object of SynonymExtractor
synonym_extractor = SynonymExtractor()

# add synonyms
synonym_names = ['NY', 'new-york', 'SF']
clean_names = ['new york', 'new york', 'san francisco']

for synonym_name, clean_name in zip(synonym_names, clean_names):
synonym_extractor.add_to_synonym(synonym_name, clean_name)

synonyms_found = synonym_extractor.get_synonyms_from_sentence('I love SF and NY. new-york is the best.')

>> ['san francisco', 'new york', 'new york']


synonym-extractor is based on `Aho-Corasick algorithm


Documentation can be found at `Read the Docs



Say you have a corpus where similar words appear frequently.

eg: Last weekened I was in NY.
I am traveling to new york next weekend.

If you train a word2vec model on this or do any sort of NLP it will treat NY and new york as 2 different words.

Instead if you create a synonym dictionary like:

eg: NY=>new york
new york=>new york

Then you can extract NY and new york as the same text.

To do the same with regex it will take a lot of time:

============ ========== = ========= ============
Docs count # Synonyms : Regex synonym-extractor
============ ========== = ========= ============
1.5 million 2K : 16 hours NA
2.5 million 10K : 15 days 15 mins
============ ========== = ========= ============

The idea for this library came from the following `StackOverflow question


The project is licensed under the MIT license.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for flashtext, version 1.0
Filename, size File type Python version Upload date Hashes
Filename, size flashtext-1.0.tar.gz (5.9 kB) File type Source Python version None Upload date Hashes View

Supported by

Pingdom Pingdom Monitoring Google Google Object Storage and Download Analytics Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page