Inspiration

We recently got interested in Natural Language Processing after collectively working on some projects utilizing machine learning in my own time, and thought that it would be a cool idea to have automation techniques which could help deliver content succinctly & accurately to millions of people.

What it does

It analyzes an essay through a natural language algorithm and extracts important keywords out of the essay. Then, the keywords are visualized in an interactive mind map using d3.js to visualize the data.

How we built it

We used Python to do parse the text, assign stop words, and perform data analysis on moderate frequency keywords. Textblob was used to analyze the sentiment of the words and assign tags. Javascript is used to visualize the data & apply animated mind-maps to keywords.

Challenges we ran into

Enhancing the NLP algorithm to produce accurate results whilst still maintaining an acceptable runtime. Finding a way to coherently visualize the large set of data. Utilizing the Target API and Ebay API to find a large set of consumer reviews to analyze.

Accomplishments that we’re proud of

The speed & accuracy of our analysis. The mindmap is also something that is quite barebones, but we’re proud of how it can functionally (and interactively) display the interconnections between the keywords in each essay. In addition, we originally intended to use the Target API to summarize consumer reviews, but find out in the last minute that their API could not supply the consumer reviews. It was quite challenging to find a suitable idea to pivot to within the time we had left.

What we learned

We learned how to use NLP techniques to analyze large text databases. In the process of doing so, we also learned how to apply NLP algorithms to achieve a balance between runtime speed and semantic accuracy.

What's next for easy essay

Finding a suitable API to analyze consumer reviews on a storefront or personal profiles on a social media website. Easy Essay’s algorithms could be used to give a short review of each blurb of text, allowing users to more easily browse products that they’re interested in, or meet people that they’re interested in.

Built With

Share this project:
×

Updates