Build your own summariser
Building a summarizer might seem like a very hectic task that would require you to have a
mastery in Machine Learning , NLP and deep learning but thanks to our dev community
for constantly making libraries and wrappers for others to use these extremly complex
algorithmis so easily.
The implementation of this summarizer is done in python so its as easy to understand as english.
So today we are using a library called BERT Extractive summariser Link
but first lets understand what is the difference between Extractive and Abstractive summarizer.
Extractive summarizers:
Instead of computer re-writing the whole summary, extractive summarizers pick out the important
points from the article and alter them to form meaningful summaries. Pros of this are
the summaries generated match alot with human written formats.
Abstractive summarizers:
In these usually computer re-writes the entire summary because of which the tone and
writing style you text might get totally changed.
Now, coming to coding part :
pip install bert-extractive-summariser
you will need to install other libraries like spacy , trasformers and neuralcoref
instantiate the model
from summarizer import Summarizer
body = 'Text body that you want to summarize with BERT'
model = Summarizer()
result = model(body, ratio=0.2) # Specified with ratio
result = model(body, num_sentences=3) # Will return 3 sentences
You can explore more options like num_sentences which is responsible for deciding the length of you summary, but if you set too less number sentences then you might not end up with a very meaningful summary.
You can checkout the live link here this is implemented in Django.
You don't really need alot of knowledge in terms of ML to built this simple yet useful app.