Text Summarization

 An AI that creates a summary from an article you input

Try our app here





Summary:

Motivations

We wanted to create something beneficial and a text summarizer would be the perfect option as it would allow people to cut reading time on articles and give them the main points, making it much more efficent and time saving. 

Pre-processing our data

So using our Amazon product reviews dataset, we cleaned and tokenized our data by getting rid of stopw‚Äčords, punctuation, HTML tags, special characters, short words, etc.

This shows the distribution of sentence lengths in the articles and summaries.

This shows 94% of summary lengths fall below the length of 8 words.

Setting maximum text length to 30 words and maximum summary length to 8 words.

Executing the rest of our code

Then we split our data into training and test data. We import a tokenizer and get rid of words that fall below and threshold frequency and use the LSTM to execute the rest of our project.

LSTM

Training and Evaluation

Our algorithm calculates a rouge score

What we learned from this project

Jeremy Kintana

Product Manager
Goes to Mountain View High School, CA

Tharune Kanagasabai

Web Designer
Goes to Olentangy Berlin High School, OH

Aaron Shih

Mathematician 
Goes to Leland High School, CA

Pradyoth Kalluri

ML Engineer
Goes to Leland High School in San Jose, CA

James Park

Data Scientist
Goes to Mountain View High School, CA
Enjoys Origami, Minecraft, and Guitar

Abhijay Movva

Vivrd Prasanna

Web Designer
Goes to American High School

Teaching Assistant