I’ve got a conundrum with this post. Somehow, every time it’s posted (every couple of weeks) it lacks the word “Musk” in the title, which bypasses my keyword filter and it shows up on my feed.
Next time this is reposted could whoever does it put Musk in the title so I don’t have to see it?
This is really what we need from things like Google Lens and things that can scan images for text; filtering out stuff based on text visible in an image.
Confirmation bias. You don’t see all the ones that are blocked. Deviations from that norm stand out.
Also, you replied to the post, training the algorithm that you’ll engage with Mu$k spam, and even said his name. You now have your filters duking it out with the algorithm.
Also, you replied to the post, training the algorithm that you’ll engage with Mu$k spam, and even said his name. You now have your filters duking it out with the algorithm.
Fair on the confirmation bias. I guess I should’ve said that I see this post every couple of weeks, even though I have a keyword filter blocking anything with the word “Musk” in the title.
I wasn’t aware that Lemmy had much of a personalized algorithm, beyond a simple upvote/down vote system with older posts falling down the list. Comments might boost it in the Hot sorting maybe? I doubt the content of those comments matters much.
I’ve got a conundrum with this post. Somehow, every time it’s posted (every couple of weeks) it lacks the word “Musk” in the title, which bypasses my keyword filter and it shows up on my feed.
Next time this is reposted could whoever does it put Musk in the title so I don’t have to see it?
This is really what we need from things like Google Lens and things that can scan images for text; filtering out stuff based on text visible in an image.
Fuck Google, just use tesseract
yeh!
Lemmy should do this automatically on the server to help with our filters.
How do we submit a feature request?
That would be a pretty expensive feature request.
Why?
OCR and especially AI image processing takes a lot of computing power, much much more than filtering plain text.
Last time I checked tesseract had pretty good performance, but I may be mistaken.
You’re right. I’ve preferred it in mobile applications where specialized hardware is not guaranteed and battery use matters. Results are consistent.
What do you use that enables filters?
I use Connect on Android primarily.
Oh damn he added filters? I haven’t used Connect in a long time. Thanks!
Confirmation bias. You don’t see all the ones that are blocked. Deviations from that norm stand out.
Also, you replied to the post, training the algorithm that you’ll engage with Mu$k spam, and even said his name. You now have your filters duking it out with the algorithm.
There is no such algorithm on Lemmy
Fair on the confirmation bias. I guess I should’ve said that I see this post every couple of weeks, even though I have a keyword filter blocking anything with the word “Musk” in the title.
I wasn’t aware that Lemmy had much of a personalized algorithm, beyond a simple upvote/down vote system with older posts falling down the list. Comments might boost it in the Hot sorting maybe? I doubt the content of those comments matters much.