Incremental Knowledge Based Question Answering

TL;DR

This paper introduces an incremental learning framework for knowledge-based question answering that effectively handles evolving knowledge bases by overcoming catastrophic forgetting through a novel loss function and exemplar selection.

Contribution

It proposes a new incremental KBQA learning method with a margin-distilled loss and collaborative exemplar selection, addressing the challenge of evolving knowledge bases.

Findings

Effective in handling evolving knowledge bases

Reduces catastrophic forgetting in incremental learning

Demonstrates efficiency on reorganized SimpleQuestion dataset

Abstract

In the past years, Knowledge-Based Question Answering (KBQA), which aims to answer natural language questions using facts in a knowledge base, has been well developed. Existing approaches often assume a static knowledge base. However, the knowledge is evolving over time in the real world. If we directly apply a fine-tuning strategy on an evolving knowledge base, it will suffer from a serious catastrophic forgetting problem. In this paper, we propose a new incremental KBQA learning framework that can progressively expand learning capacity as humans do. Specifically, it comprises a margin-distilled loss and a collaborative exemplar selection method, to overcome the catastrophic forgetting problem by taking advantage of knowledge distillation. We reorganize the SimpleQuestion dataset to evaluate the proposed incremental learning solution to KBQA. The comprehensive experiments demonstrate its effectiveness and efficiency when working with the evolving knowledge base.