How mediafeedr Determines Related Content

mediafeedr doesn't use AI to link related articles together, instead it's a bit of word magic, maths and a sprinkle of fairy dust.

While AI linking of related articles would lead to more positive and relevant matches, we don't feel that adding to the climate crisis should be high on our priority list — plus it's really, really expensive and we don't have the budget for it.

Instead we've built a word matching algorithm that looks at the headlines, attempts to find commonalities between articles and applies a weighting to see how related we think these stories seem to be. If it's a high enough rating, we class the articles at releated.

This of course will naturally lead to false positives if there is enough of a crossover between certain articles, and while we're actively working to better the algoritm to pick this up, because we're looking at hundreds of publications all with different writing styles it is challenging to say the least!

If you have any issues regarding how an article has been linked (or suggestions on how we can improve it) please reach out to me on the support forum to discuss it further.