This started as a little data science experiment. I analyzed 15,000 reviews and was surprised by what I found. A lot of reviews contain phrases like, "I took this [probiotic] for my gut health but then I noticed my eczema went away." The top 10 lists I was generating suggested supplements that weren't always suggested by the lists you find online when you google "supplements for eczema". I suspect this is because those lists are generated based on mechanisms we already understand while working backward from reviews also reveals effective supplements whose mechanisms are still not understood.
The ranking algorithm is still flawed and my database is still too small, but please check it out! I will improve both over the next few weeks.
Please let me know your thoughts, feedback, and suggestions. Especially if you have ideas for where I could promote this.
I know how to monetize it, I just don't want to yet.
On the backend, I use supervisor and gunicorn and a web server to manage and serve a python/flask API that uses flat files as a database. The scraper uses scrapy. Part of my motivation for this project was to learn new technologies, and so I built the front-end in Gatsby, which uses React.
1 comment
[ 0.30 ms ] story [ 13.9 ms ] threadThe ranking algorithm is still flawed and my database is still too small, but please check it out! I will improve both over the next few weeks.
Please let me know your thoughts, feedback, and suggestions. Especially if you have ideas for where I could promote this.
I know how to monetize it, I just don't want to yet.
On the backend, I use supervisor and gunicorn and a web server to manage and serve a python/flask API that uses flat files as a database. The scraper uses scrapy. Part of my motivation for this project was to learn new technologies, and so I built the front-end in Gatsby, which uses React.