- cross-posted to:
- technology@lemmy.world
- cross-posted to:
- technology@lemmy.world
The ‘Brand Safety’ and ‘Suitability’ industries have financially crushed the news business by keeping ads away from articles that its ‘sentiment analysis’ algorithms think will make people sad or upset.
Sentiment analysis is basic data mining.
Get access to an API and scan the tweets / posts / whatever and look for key words. Hate, dumb, wtf, love, awesome, etc. You can get 60-80% accuracy just counting the number of words that express a sentiment. Literally that simple, just count positive and negative words/phrases. Across a large dataset you can get a decent idea what people think, and there are easy ways to dig deeper.
And that was the half-assed quick approach, folks who know Markov chains and LLMs could do far, far deeper analysis.
Our world is breaking and we don’t know how to fix it, but we can guess pretty accurately how people feel about it. Turn AI onto fixing our problems people who make the robots.