🧪 Test?View on arXiv
When Should Semantic Priors Enter Partially-Identified Causal Discovery?
Not provided
causal discoverysemantic informationobservational data
2611.14736
Builder Relevance
19h ago60%
Abstract
The paper explores the integration of semantic information from large language models in resolving causal directions in observational causal discovery.
Reality Card
Core Claim
Semantic information improves causal recovery but targeting unstable edges does not provide a clear advantage over random allocation.
Method / Result
Semantic information improved recovery over coin-flip orientations in experiments across benchmark networks.
Limitations
Learned references contain the true DAG in only three of 80 planned tasks, indicating significant limitations in reference adequacy.
Paper to code
Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.
No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.