We take one second to search for some product on Amazon, two seconds to see the list, three seconds to click in - and then 20 minutes to read the reviews. Reviewers are inborn writers. They tell the story from their work to their social life. And you don't really care. All you care about is the PRODUCT itself!!! We have come up with a website to identify "meaningful" reviews and display them in visual format.
What it does
Given an Amazon product link, the website will show a word cloud of the "meaningful" reviews. A radar graph of basic emotions seen from the reviews will also be displayed. From the graph, users are able to understand the advantages and disadvantages product in a few seconds. They can also compare similar products using the radar graph, and arrive at a quick decision.
How we built it
We first process all the reviews by Natural Language Processing into meaning phrases, ranked by its frequency and importance. We will then map the meaningful phrases into a word cloud. At the same time, we analyze the emotional components from the reviews and plot a radar graph of basic emotions reveals.
Challenges we ran into
- Incorporating Natural Language Processing algorithm
- Grabbing Amazon's reviews and bypassing its anti-robot program.
- Data Visualization
What we learned