Beyond I.I.D.: Three Levels of Generalization for Question Answering on Knowledge Bases
This work addresses the generalization challenge in KBQA for large-scale knowledge bases, which is incremental as it builds on existing methods by introducing new evaluation settings and a dataset.
The authors tackled the problem of question answering on knowledge bases (KBQA) by challenging the standard i.i.d. assumption and proposing three levels of generalization: i.i.d., compositional, and zero-shot. They introduced a new dataset, GrailQA, with 64,331 questions and a BERT-based model, demonstrating the role of pre-trained embeddings in KBQA generalization.
Existing studies on question answering on knowledge bases (KBQA) mainly operate with the standard i.i.d assumption, i.e., training distribution over questions is the same as the test distribution. However, i.i.d may be neither reasonably achievable nor desirable on large-scale KBs because 1) true user distribution is hard to capture and 2) randomly sample training examples from the enormous space would be highly data-inefficient. Instead, we suggest that KBQA models should have three levels of built-in generalization: i.i.d, compositional, and zero-shot. To facilitate the development of KBQA models with stronger generalization, we construct and release a new large-scale, high-quality dataset with 64,331 questions, GrailQA, and provide evaluation settings for all three levels of generalization. In addition, we propose a novel BERT-based KBQA model. The combination of our dataset and model enables us to thoroughly examine and demonstrate, for the first time, the key role of pre-trained contextual embeddings like BERT in the generalization of KBQA.