CrowdChunk

@rrhoover Correct. We're useful to apps that have a lot of feedback. (by proxy a lot of passionate users) Product Hunt has enough users to generate relevant scores. We haven't been able to find any strong correlation in similar feedback. 55% of your app reviews are positive feedback. That's a great back pocket stat.

@rrhoover invite me to beta! The iOS app needs much love! :)

@chrismessina agreed! It's going to be so 🔥 but we're a ways from a beta. I'll keep you posted!

@chrismessina yep, we have a great lexical analysis engine looking at every review ever written about all apps on the iOS App Store and Google Play. What's cool is we look at entire statements, not just keywords, which is huge.

@russ_sm right -- but so, a lot of reviews are gibberish. How do you make sense of those? Also, have you detected any oddities/fake reviews/bots/spam?

@chrismessina At the moment we focus on english reviews. English based reviews are the easies for us to detect precise feedback and emotion. Codeq is a precision over recall shop. We do have some baked in logic to notice when there's noise or a bot involved, but it's a work in progress. So far we've found our signals to be statistically accurate based on our expectations. Our computational linguistics team has spent a massive amount of time doing supervised learning to ensure we're filtering out questionable data.

@drew team appreciates the kind words! :D

@normanliang for now! We're really hoping to expand the feature set to encompass a lot of use cases. Really excited to see how people use it!

@russ_sm it would be great to hear some ways people can use this today to solve problems you have encountered as a developer

@normanliang Sure. One of the biggest use cases is looking at competitive apps, or apps within an entire category, to gauge who is winning with consumers. It's especially effective for app categories with lots of feedback. (EG: Games, Social, etc.) You can also look at data trends for specific apps over long ranges of time to compare when you released a feature and what effect it had on your consumer experience. We scan both Play and the iOS App store daily. That's more text than a human could ever read within 24 hours. We're doing something for developers humans simply can't do fast enough. We're finding insights you could only produce with machine learning at scale.

@dshan agreed!