CEQA
About this tool
Official code for a NeurIPS 2023 paper on complex event queries over ASER-50K. Answers must satisfy implicit constraints on whether events occur and their ordering. A theorem prover filters contradictory answers for the benchmark. Implements GQE, Q2P, MLP and FuzzQE encoders and memory-enhanced MEQE variants for event and neuro-symbolic reasoning research.
- Feature 1: Complex event queries combine first-order logic with implicit occurrence and ordering constraints.
- Feature 2: ASER-50K benchmark sampled from ASER 2.1.
- Feature 3: A theorem prover filters answers contradicting constraints.
- Feature 4: GQE, Q2P, MLP and FuzzQE implementations with memory-enhanced variants.
Resource information
- Maintainer
- HKUST-KnowComp, Associate Professor Yangqiu Song's group.
- Type
- Tool / library
- Category
- Knowledge graphs & graph computing
- Runs on
- Tool or library
- Who it's for
- Researchers evaluating or improving event reasoning, commonsense reasoning and complex knowledge-graph query encoders.
- License at time of listing
- MIT
- Source
- GitHub
