Making a Text-Summarizer with Keras
Gur Raunaq Singh
One of the common uses of Machine Learning in a lot of mobile applications is Text-Summarization. It is one of the key techniques companies are using for improving their products, or some even have complete mobile apps based on it (apps like Awesummly).
Most summarization tools in the past were Extractive, which worked well in fields like Finance, Weather forecast generator, and Medicine. In this talk, will be making a summarizer which will be Abstractive, which will be good at understanding the meaning of a piece of article, and create a short summary of it.
The aim of this talk is to provide a introduction to NLP, Semantic Analysis and Syntactic parsing, understand their working principles, and a Jupyter Notebook showcasing a working demo.
The topics covered in this talk are as follows :
- Basics of Semantic Analysis and Syntactic Parsing
- Word Vectors
- GloVe (Global Vectors)
The talk will explain :
- how to use pre-trained word embeddings in a Keras model
- use it to generate an output sequence of words, given an input sequence of words using a Neural Encoder Decoder
- add an attention mechanism to our decoder, which will help it decide what is the most relevant token to focus on when generating new text
Basic understanding of :
I am a final year student, pursuing B.Tech CSE from IPU, New Delhi. I am currently working as a Data Scientist at ABinBev GAC Bangalore, India.
Over the years, I have had the work to dabble in various fields such as Game Developement, Augmented and Virtual Reality Developement and Data Science. When not hitting head-shots in CS:GO, or scoring goals in the football field, I usually like to work on building small projects or participating in various programming competitions.
I am a regular at hackathons, having won over 9 in the past year including the Grand Prize at Angelhack Delhi, 2016.