Hey there, thank you for sharing your work and thoughts.
Content discovery is a problem I'm really interested in but every time I've tried to find a solution that makes sense I end up with a bunch or pre-requisites that need to solved before I get to it.
Unfortunately when it comes to content discovery I keep going back to some form of tokenization and simple vectorization or even worst word embeddings. This coupled with an overlay dht to allow for nearest neighbour lookups seems to be enough for basic discovery in small scale. There is really interesting literature around p2p multi-dimentional k-nearest search than can even be adapted for use withe existing DHTs. I am by no means implying this is an easy thing but even the harder parts vectorizing images/video/etc are issues which have been tackled with by academia for quite some time so some first-version solutions can be found.
The issue I'm really struggling with trust. Any possible solution I start thinking eventually has the same issue. How do you deal with abuse/spam when the network gets big enough? People have been abusing email for a very long time, imagine what happens if you can target ads based on similarity. Will anyone be ever able to find something legit?
Hey there, thank you for sharing your work and thoughts.
Content discovery is a problem I'm really interested in but every time I've tried to find a solution that makes sense I end up with a bunch or pre-requisites that need to solved before I get to it.
Unfortunately when it comes to content discovery I keep going back to some form of tokenization and simple vectorization or even worst word embeddings. This coupled with an overlay dht to allow for nearest neighbour lookups seems to be enough for basic discovery in small scale. There is really interesting literature around p2p multi-dimentional k-nearest search than can even be adapted for use withe existing DHTs. I am by no means implying this is an easy thing but even the harder parts vectorizing images/video/etc are issues which have been tackled with by academia for quite some time so some first-version solutions can be found.
The issue I'm really struggling with trust. Any possible solution I start thinking eventually has the same issue. How do you deal with abuse/spam when the network gets big enough? People have been abusing email for a very long time, imagine what happens if you can target ads based on similarity. Will anyone be ever able to find something legit?
ps. Sorry for the rant :D