pysentimiento 0.4.0

A Transformer-based library for SocialNLP tasks

pysentimiento: A Python toolkit for Sentiment Analysis and Social NLP tasks

A Transformer-based library for SocialNLP tasks.

Currently supports:

• Sentiment Analysis (Spanish, English)
• Emotion Analysis (Spanish, English)
• Hate Speech Detection (Spanish, English)
• Named Entity Recognition (Spanish + English)
• POS Tagging (Spanish + English)

Just do pip install pysentimiento and start using it:

from pysentimiento import create_analyzer
analyzer = create_analyzer(task="sentiment", lang="es")

analyzer.predict("Qué gran jugador es Messi")
# returns AnalyzerOutput(output=POS, probas={POS: 0.998, NEG: 0.002, NEU: 0.000})
analyzer.predict("Esto es pésimo")
# returns AnalyzerOutput(output=NEG, probas={NEG: 0.999, POS: 0.001, NEU: 0.000})
analyzer.predict("Qué es esto?")
# returns AnalyzerOutput(output=NEU, probas={NEU: 0.993, NEG: 0.005, POS: 0.002})

analyzer.predict("jejeje no te creo mucho")
# AnalyzerOutput(output=NEG, probas={NEG: 0.587, NEU: 0.408, POS: 0.005})
"""
Emotion Analysis in English
"""

analyzer = create_analyzer(task="emotion", lang="en")

emotion_analyzer.predict("yayyy")
# returns AnalyzerOutput(output=joy, probas={joy: 0.723, others: 0.198, surprise: 0.038, disgust: 0.011, sadness: 0.011, fear: 0.010, anger: 0.009})
emotion_analyzer.predict("fuck off")
# returns AnalyzerOutput(output=anger, probas={anger: 0.798, surprise: 0.055, fear: 0.040, disgust: 0.036, joy: 0.028, others: 0.023, sadness: 0.019})

"""
Hate Speech (misogyny & racism)
"""
hate_speech_analyzer = create_analyzer(task="hate_speech", lang="es")

hate_speech_analyzer.predict("Esto es una mierda pero no es odio")
# returns AnalyzerOutput(output=[], probas={hateful: 0.022, targeted: 0.009, aggressive: 0.018})
hate_speech_analyzer.predict("Esto es odio porque los inmigrantes deben ser aniquilados")
# returns AnalyzerOutput(output=['hateful'], probas={hateful: 0.835, targeted: 0.008, aggressive: 0.476})

hate_speech_analyzer.predict("Vaya guarra barata y de poca monta es Miley!")
# returns AnalyzerOutput(output=['hateful', 'targeted', 'aggressive'], probas={hateful: 0.987, targeted: 0.978, aggressive: 0.969})

Also, you might use pretrained models directly with transformers library.

from transformers import AutoTokenizer, AutoModelForSequenceClassification

tokenizer = AutoTokenizer.from_pretrained("pysentimiento/robertuito-sentiment-analysis")

model = AutoModelForSequenceClassification.from_pretrained("pysentimiento/robertuito-sentiment-analysis")

Preprocessing

pysentimiento features a tweet preprocessor specially suited for tweet classification with transformer-based models.

from pysentimiento.preprocessing import preprocess_tweet

# Replaces user handles and URLs by special tokens
preprocess_tweet("@perezjotaeme debería cambiar esto http://bit.ly/sarasa") # "@usuario debería cambiar esto url"

# Shortens repeated characters
preprocess_tweet("no entiendo naaaaaaaadaaaaaaaa", shorten=2) # "no entiendo naadaa"

# Normalizes laughters
preprocess_tweet("jajajajaajjajaajajaja no lo puedo creer ajajaj") # "jaja no lo puedo creer jaja"

# Handles hashtags
preprocess_tweet("esto es #UnaGenialidad")
# "esto es una genialidad"

# Handles emojis
preprocess_tweet("🎉🎉", lang="en")
# 'emoji party popper emoji emoji party popper emoji'

Trained models so far

Check CLASSIFIERS.md for details on the reported performances of each model.

Instructions for developers

0. Clone and install

git clone https://github.com/pysentimiento/pysentimiento
pip install poetry
poetry shell
poetry install

1. Download data TASS 2020 data to data/tass2020 (you have to register here to download the dataset)

Labels must be placed under data/tass2020/test1.1/labels

Open an issue or email us if you are not able to get the data.

2. Run script to train models

Check TRAIN.md for further information on how to train your models

3. Upload models to Huggingface's Model Hub

Check "Model sharing and upload" instructions in huggingface docs.

License

pysentimiento is an open-source library. However, please be aware that models are trained with third-party datasets and are subject to their respective licenses, many of which are for non-commercial use

1. TASS Dataset license (License for Sentiment Analysis in Spanish, Emotion Analysis in Spanish & English)

2. SEMEval 2017 Dataset license (Sentiment Analysis in English)

3. LinCE Datasets (License for NER & POS tagging)

Suggestions and bugfixes

Please use the repository issue tracker to point out bugs and make suggestions (new models, use another datasets, some other languages, etc)

Citation

If you use pysentimiento in your work, please cite this paper

@misc{perez2021pysentimiento,
      title={pysentimiento: A Python Toolkit for Sentiment Analysis and SocialNLP tasks},
      author={Juan Manuel Pérez and Juan Carlos Giudici and Franco Luque},
      year={2021},
      eprint={2106.09462},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}

Also, pleace cite related pre-trained models and datasets for the specific models you use:

%%%%%%%%%%%%%%%%%%%%%%%%%%
% Pretrained models      %
%%%%%%%%%%%%%%%%%%%%%%%%%%
% RoBERTuito
@article{perez2021robertuito,
  title={RoBERTuito: a pre-trained language model for social media text in Spanish},
  author={P{\'e}rez, Juan Manuel and Furman, Dami{\'a}n A and Alemany, Laura Alonso and Luque, Franco},
  journal={arXiv preprint arXiv:2111.09453},
  year={2021}
}
% BETO
@article{canete2020spanish,
  title={Spanish pre-trained bert model and evaluation data},
  author={Canete, Jos{\'e} and Chaperon, Gabriel and Fuentes, Rodrigo and Ho, Jou-Hui and Kang, Hojin and P{\'e}rez, Jorge},
  journal={Pml4dc at iclr},
  volume={2020},
  pages={2020},
  year={2020}
}
% BERTweet
@inproceedings{nguyen2020bertweet,
  title={BERTweet: A pre-trained language model for English Tweets},
  author={Nguyen, Dat Quoc and Vu, Thanh and Nguyen, Anh Tuan},
  booktitle={Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations},
  pages={9--14},
  year={2020}
}
%%%%%%%%%%%%%%%%%%%%%%%%%%
% Datasets               %
%%%%%%%%%%%%%%%%%%%%%%%%%%
% TASS 2020 (sentiment in Spanish)

@article{garcia2020overview,
  title={Overview of TASS 2020: introducing emotion detection},
  author={Garc{\'\i}a-Vegaa, Manuel and D{\'\i}az-Galianoa, Manuel Carlos and Garc{\'\i}a-Cumbrerasa, Miguel {\'A} and del Arcoa, Flor Miriam Plaza and Montejo-R{\'a}eza, Arturo and Jim{\'e}nez-Zafraa, Salud Mar{\'\i}a and C{\'a}marab, Eugenio Mart{\'\i}nez and Aguilarc, C{\'e}sar Antonio and Antonio, Marco and Cabezudod, Sobrevilla and others},
  year={2020}
}

% EmoEvent (Emotion Analysis Spanish & English)

@inproceedings{del2020emoevent,
  title={EmoEvent: A multilingual emotion corpus based on different events},
  author={del Arco, Flor Miriam Plaza and Strapparava, Carlo and Lopez, L Alfonso Urena and Mart{\'\i}n-Valdivia, M Teresa},
  booktitle={Proceedings of the 12th Language Resources and Evaluation Conference},
  pages={1492--1498},
  year={2020}
}

% Hate Speech Detection (Spanish & English)


@inproceedings{hateval2019semeval,
  title={SemEval-2019 Task 5: Multilingual Detection of Hate Speech Against Immigrants and Women in Twitter},
  author={Basile, Valerio and Bosco, Cristina and Fersini, Elisabetta and Nozza, Debora and Patti, Viviana and Rangel, Francisco and Rosso, Paolo and Sanguinetti, Manuela},
  booktitle={Proceedings of the 13th International Workshop on Semantic Evaluation (SemEval-2019)},
  year={2019},
  publisher= {Association for Computational Linguistics}
}
% Sentiment Analysis in English

@article{nakov2019semeval,
  title={SemEval-2016 task 4: Sentiment analysis in Twitter},
  author={Nakov, Preslav and Ritter, Alan and Rosenthal, Sara and Sebastiani, Fabrizio and Stoyanov, Veselin},
  journal={arXiv preprint arXiv:1912.01973},
  year={2019}
}

% LinCE (NER & POS Tagging)

@inproceedings{aguilar2020lince,
  title={LinCE: A Centralized Benchmark for Linguistic Code-switching Evaluation},
  author={Aguilar, Gustavo and Kar, Sudipta and Solorio, Thamar},
  booktitle={Proceedings of the 12th Language Resources and Evaluation Conference},
  pages={1803--1813},
  year={2020}
}