Subgraph Retrieval Enhanced Model for Multi-hop KBQA
What it is
A PyTorch implementation of a model for multi-hop Knowledge Base Question Answering that uses subgraph retrieval to enhance reasoning. It operates by first training a retriever to extract relevant subgraphs from a knowledge graph, then training a reasoner on those subgraphs, and finally fine-tuning the retriever based on feedback from the reasoner.
The problem it solves
Standard knowledge base question answering models struggle with multi-hop reasoning because they often lack access to the specific, relevant subgraph of the knowledge base needed to answer complex questions.
What you could use it for
- Multi-hop question answering over the WebQuestionsSP dataset
- Complex web question answering over the CWQ dataset
- Researching retrieval-enhanced reasoning architectures for knowledge bases
Read the source ↗Back to the library →
Summary written by an AI model from the source; check the source before you rely on it.