LeSR uses an LLM to propose logic rules from sampled subgraphs, then a trainable reasoner weights those rules against the knowledge base and combines them with RotatE for KBC.
Learn to Explain Efficiently via Neural Logic Inductive Learning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
The capability of making interpretable and self-explanatory decisions is essential for developing responsible machine learning systems. In this work, we study the learning to explain problem in the scope of inductive logic programming (ILP). We propose Neural Logic Inductive Learning (NLIL), an efficient differentiable ILP framework that learns first-order logic rules that can explain the patterns in the data. In experiments, compared with the state-of-the-art methods, we find NLIL can search for rules that are x10 times longer while remaining x3 times faster. We also show that NLIL can scale to large image datasets, i.e. Visual Genome, with 1M entities.
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Large Language Model-Enhanced Symbolic Reasoning for Knowledge Base Completion
LeSR uses an LLM to propose logic rules from sampled subgraphs, then a trainable reasoner weights those rules against the knowledge base and combines them with RotatE for KBC.