How to Use Displacy in Flask For Named Entity Recognition

In this tutorial we will be see how to render the named entities of a text extracted using spacy’s displacy in flask.

First of all the basic requirement is to use the following packages

• Flask
• Spacy
• Flask-Markdown

To install the various packages we will be using pip as below

pip install flask Flask-Markdown spacy

To work with spacy you will need a particular language model for the natural language processing. You can download the english model with the code below.

python -m spacy download en

Let us begin with building our app.

The Basic Structure of Our App (DisplaCify)

• Front -End
• Back-End

Working on the Back-End.

The main logic to be able to render displacy in flask is to parse it via markdown. Hence we will be using Flask-Markdown, a flask extension to enable us to be able to render our named entities in a very nice format in our front-end.

All we need is to import Markdown from the flaskext.markdown and wrap it around our app as below

from flask import Flask,url_for,render_template,requestfrom flaskext.markdown import Markdown
#Init 
app = Flask(__name__)
Markdown(app)

That is all we need to be able to work with the rendering of displacy’s results at the back-end.

We can then set a custom format for our html styling with

HTML_WRAPPER = """<div style="overflow-x: auto; border: 1px solid #e6e9ef; border-radius: 0.25rem; padding: 1rem">{}</div>"""

The full code for the back-end

 

from flask import Flask,url_for,render_template,request
from flaskext.markdown import Markdown

# NLP Pkgs
import spacy
from spacy import displacy
nlp = spacy.load('en')
import json

# Init
app = Flask(__name__)
Markdown(app)


@app.route('/')
def index():
	return render_template('index.html')


@app.route('/extract',methods=["GET","POST"])
def extract():
	if request.method == 'POST':
		raw_text = request.form['rawtext']
		docx = nlp(raw_text)
		html = displacy.render(docx,style="ent")
		html = html.replace("\n\n","\n")
		result = HTML_WRAPPER.format(html)

	return render_template('result.html',rawtext=raw_text,result=result)


@app.route('/previewer')
def previewer():
	return render_template('previewer.html')

@app.route('/preview',methods=["GET","POST"])
def preview():
	if request.method == 'POST':
		newtext = request.form['newtext']
		result = newtext

	return render_template('preview.html',newtext=newtext,result=result)


if __name__ == '__main__':
	app.run(debug=True)

 

For the Front-End

We will be using a basic form that uses POST as it method and a text-area to receive the user input and finally send the result to a route at the back-end.

<form method="POST" action="/extract" >
<textarea rows="5" cols="5" required="true" name="rawtext" class="form-control">
</textarea>
<button type="reset" class="btn btn-primary"><i class="fa fa-eraser"></i> Clear</button>
<button type="submit" class="btn btn-info"><i class="fa fa-database"></i> Submit</button>

</form>

 

We will be using material bootstrap for our UI in the front-end.

You can check the video tutorial here.

 

To get more about how to build machine learning and NLP apps you can check out this awesome course here

Thanks a lot for your time

Jesus Saves

By Jesse E.Agbemabiase (JCharis)