Fine-tuning Multi-hop Question Answering with Hierarchical Graph Network
This work addresses multi-hop reasoning in question answering, which is a challenging problem for NLP systems, though it appears incremental as it combines existing graph and fine-tuning approaches.
The paper tackles multi-hop question answering by proposing a two-stage model that first uses a hierarchical graph network to select and concatenate support sentences, then fine-tunes a language model to extract answer spans, achieving competitive performance on benchmark datasets.
In this paper, we present a two stage model for multi-hop question answering. The first stage is a hierarchical graph network, which is used to reason over multi-hop question and is capable to capture different levels of granularity using the nature structure(i.e., paragraphs, questions, sentences and entities) of documents. The reasoning process is convert to node classify task(i.e., paragraph nodes and sentences nodes). The second stage is a language model fine-tuning task. In a word, stage one use graph neural network to select and concatenate support sentences as one paragraph, and stage two find the answer span in language model fine-tuning paradigm.