[–] PaulHoule 1y ago ↗ This is what the analysis was 10 years ago. What has changed? [–] rkwz 1y ago ↗ Data still needs to be cleaned ;)ML based linkage has evolved a lot over the past years, and a few people are trying out LLM based ones - https://arxiv.org/pdf/2403.06434v1 [–] PaulHoule 1y ago ↗ I'd assume you could train a cross encoder network likehttps://sbert.net/docs/cross_encoder/usage/usage.htmlto "predict" if two entities are the same?
[–] rkwz 1y ago ↗ Data still needs to be cleaned ;)ML based linkage has evolved a lot over the past years, and a few people are trying out LLM based ones - https://arxiv.org/pdf/2403.06434v1 [–] PaulHoule 1y ago ↗ I'd assume you could train a cross encoder network likehttps://sbert.net/docs/cross_encoder/usage/usage.htmlto "predict" if two entities are the same?
[–] PaulHoule 1y ago ↗ I'd assume you could train a cross encoder network likehttps://sbert.net/docs/cross_encoder/usage/usage.htmlto "predict" if two entities are the same?
3 comments
[ 4.4 ms ] story [ 14.7 ms ] threadML based linkage has evolved a lot over the past years, and a few people are trying out LLM based ones - https://arxiv.org/pdf/2403.06434v1
https://sbert.net/docs/cross_encoder/usage/usage.html
to "predict" if two entities are the same?