🚨 New Paper Accepted at #NAACL2025 🚨
"Conformalized Answer Set Prediction for Knowledge Graph Embedding"
@yuqichengzhu.bsky.social, Nico Potyka, Jiarong Pan, @boxiong.bsky.social, Yunjie He, Evgeny Kharlamov, @ststaab.bsky.social
✨Check out our paper: https://arxiv.org/pdf/2408.08248v3
"Conformalized Answer Set Prediction for Knowledge Graph Embedding"
@yuqichengzhu.bsky.social, Nico Potyka, Jiarong Pan, @boxiong.bsky.social, Yunjie He, Evgeny Kharlamov, @ststaab.bsky.social
✨Check out our paper: https://arxiv.org/pdf/2408.08248v3
Comments
😲 Do the plausibility scores they return provide this uncertainty information?
Unfortunately, no—these scores are not calibrated and lack a probabilistic interpretation.
"How many entities do we need to guarantee coverage of the true answer at a pre-defined confidence level (e.g. 90%)?"
(The more entities we need, the more uncertain the KGE model is about its predictions.)