Natural Language Processing API - FastAPI, spaCy, SKLearn

api that does sentiment analysis and analyzes text for tokens index, text, tag, POS, Dependency, Lemma, Shape, Alpha, and stop words

model.

class Input(BaseModel):
    sentence: str

endpoints.

POST.

/get_svm_sentiment (EXPERIMENTAL).

gets sentiment using a support vector machine, highly experimental.
request will be:

{
"sentence" : "I hate this book. It's not good."
}

POST.

/analyze_text

gets input text and breaks it down by part of speech, lemma, shape, alpha, and stop word.
request will be:

{
"sentence" : "I am going to play basketball"
}

POST.

/summarize

summarizes paragraphs
request will be:

{
"sentence" : "Lately I have been really interested in language. I’ve been all over polyglot Youtube and started learning French on Duolingo. Communication and how we speak is very interesting to me. Even non verbal, ie: communicating with animals, colors (red bad green good) and so forth. I have a hard time with side projects, I’m in the web space and you can imagine how boring it is making some sort of crud app and plus, I don’t really have any crazy business ideas. I love python and web technologies so I wanted to build something to help me visualize words in a sentence and learn more about how it is all broken down. So I made a small API to break down a few things. The first endpoint analyzes text and will show the words part of speech, lemma, shape, dependency, alpha, and if its a stop word (and an id for indexing sake) and another endpoint for named entity recognition that shows start character, end character, the text itself, and the label that is tied to the text ie Mcdonalds : ORG ."
}

POST.

/entity_recognition

gets input text and displays named entities. where they begin, end, what the entity is, and the label.
request will be:

{
"sentence" : "Bill Gates is an awesome guy. Microsoft is a billion dollar company"
}

POST.

/sentiment_analysis

displays a positive or negative score based on polarity and displays all positive and negative words. handled by spaCy.
request will be:

{
"sentence" : "I love this place. the food is so good and the service is awesome!"
}

POST.

/generate_text

generates random text using a gpt2 model and the transformers library. Its fairly wonky but very fun to use. perhaps if you have gpt3 access or another heavy model you will get better results, but very fun to play with:

{
"sentence" : "My name is Nick and I like"
}

POST.

/translate_french

endpoint that will translate english to french.

{
"sentence" : "Where is the bathroom and nearest cafe?"
}

RUN.

pip install -r requirements.txt
uvicorn main:app --reload

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate.