Apply HN: Better descriptions for content

Problem: The amount of digital content worldwide is growing, but the descriptions for this content are subpar, and finding relevant/related content can be difficult. Solution: An search application and API that generates an intuitive 'description cloud' for any text, HTML content, song, etc - and shows closely related content. Background: This is a hackathon project I worked on last year with an incredibly talented team. We ended up putting it on hold due to time constraints, however the beta version of the application worked at a fairly sophisticated level (I will probably deploy from the old repo if we see enough support. backend is scala, frontend is angular, we use R and python as well and also have a separate crawler we use to find new content). These are not simple, old-school word clouds that utilize word density (which have been around for decades). We ended up downloading millions of song lyrics, political speeches, blogs, web pages and more - we wrote scripts to process and analyze the data for every single piece of content we downloaded. We used this data as part of our API that generates descriptions for new content. We also take the content creator/publisher into great consideration, and we can generate descriptions for authors/actors/musicians/politicians/websites if we have more than one document associated with them. We also realized that all content can be connected in webs around the descriptive clouds - it was a very new, fun and "un google" way of browsing content and bubbling up other things I'd be interested in - especially movies and songs. I ended up discovering tons of obscure songs and movies that I now love. I know other people would like it too. This isn't a project that lost steam - it started as a hackathon, we all had a great time and every time we meet up we say "that was a really cool app, why didn't we release it?".