Skip to main content

A sentence paraphraser based on dependency syntax and word embeddings

Project description

dependency-paraphraser

A sentence paraphraser based on dependency parsing and word embedding similarity.

How the paraphraser works:

  1. Create a random projection of the dependency tree
  2. Replace several words with similar ones

The basic usage (for Russian language) is based on Natasha library:

pip install dependency-paraphraser natasha
import dependency_paraphraser.natasha
import random
random.seed(42)
text = 'каждый охотник желает знать где сидит фазан'
for i in range(3):
    print(dependency_paraphraser.natasha.paraphrase(text, tree_temperature=2))
# желает знать сидит фазан где каждый охотник
# каждый охотник желает знать где фазан сидит
# знать где фазан сидит каждый охотник желает

You can provide your own w2v model to replace words with similar ones:

import compress_fasttext
small_model = compress_fasttext.models.CompressedFastTextKeyedVectors.load(
    'https://github.com/avidale/compress-fasttext/releases/download/v0.0.1/ft_freqprune_100K_20K_pq_100.bin'
)
random.seed(42)
for i in range(3):
    print(dependency_paraphraser.natasha.paraphrase(text, w2v=small_model, p_rep=0.8, min_sim=0.55))
# стремится каждый охотник знать рябчик где усаживается
# каждый охотник хочет узнать фазан где просиживает
# каждый охотник хочет узнать фазан где восседает

Alternatively, you can expand and use the w2v model from Natasha (aka navec):

navec_model = dependency_paraphraser.natasha.emb.as_gensim
random.seed(42)
for i in range(3):
    print(dependency_paraphraser.natasha.paraphrase(text, w2v=navec_model, p_rep=0.5, min_sim=0.55))
# желает каждый охотник помнить фазан где лежит
# каждый охотник желает знать фазан где сидит
# каждый охотник оставляет понять где фазан лежит

For other languages, one way to use this paraphraser is with the UDPipe library

pip install dependency-paraphraser ufal.udpipe pyconll
import dependency_paraphraser.udpipe
path = 'english-ewt-ud-2.5-191206.udpipe'
pipe = dependency_paraphraser.udpipe.Model(path)
projector = dependency_paraphraser.udpipe.en_udpipe_projectortext = 'in April 2012 they released the videoclip for a new single entitled Giorgio Mastrota'
for i in range(3):
    print(dependency_paraphraser.udpipe.paraphrase(text, pipe, projector=projector, tree_temperature=1))
# they released the videoclip in April 2012 for a new entitled Mastrota single Giorgio
# they released in April 2012 the videoclip for a entitled single new Giorgio Mastrota
# they released the videoclip in April 2012 for a new single Giorgio Mastrota entitled

Projectors (models for projecting dependency trees into a flat sentence) can be trained for any language, if you have a corpus of unlabeled sentences and a syntax parser to label them:

import dependency_paraphraser.udpipe
import dependency_paraphraser.train_projector
parser = dependency_paraphraser.udpipe.Model(path_to_your_model)

sents = dependency_paraphraser.train_projector.label_udpipe_sentences(
    texts=your_corpus,
    model=parser,
)
projector = dependency_paraphraser.train_projector.train_projector(sents)

Project details


Download files

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

Source Distribution

dependency-paraphraser-0.0.3.tar.gz (57.2 kB view details)

Uploaded Source

File details

Details for the file dependency-paraphraser-0.0.3.tar.gz.

File metadata

  • Download URL: dependency-paraphraser-0.0.3.tar.gz
  • Upload date:
  • Size: 57.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.1.1 pkginfo/1.5.0.1 requests/2.23.0 setuptools/46.1.3 requests-toolbelt/0.9.1 tqdm/4.45.0 CPython/3.7.6

File hashes

Hashes for dependency-paraphraser-0.0.3.tar.gz
Algorithm Hash digest
SHA256 d1d46dc081288d8fca48a79260822535d2c219e73a9c40d3a9306282dd97ab7b
MD5 01d374ed0e8ba79089c2d287aa6f8179
BLAKE2b-256 6c33717666cba215cabcd06c6c4514c24429d9e14996a291f0452fc273aeef0b

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page