KGR^4: Retrieval, Retrospect, Refine and Rethink for Commonsense Generation

TL;DR

KGR^4 is a four-stage framework inspired by human sentence creation, enhancing commonsense generation by retrieval, editing, refining, and rethinking to produce more plausible and grammatically correct sentences, achieving state-of-the-art results.

Contribution

The paper introduces KGR^4, a novel multi-stage framework that significantly improves commonsense sentence generation by integrating retrieval, editing, and refinement processes.

Findings

Achieves 33.56 SPICE points on CommonGen benchmark.

Outperforms previous best by 2.49 SPICE points.

Demonstrates effectiveness through extensive experiments.

Abstract

Generative commonsense reasoning requires machines to generate sentences describing an everyday scenario given several concepts, which has attracted much attention recently. However, existing models cannot perform as well as humans, since sentences they produce are often implausible and grammatically incorrect. In this paper, inspired by the process of humans creating sentences, we propose a novel Knowledge-enhanced Commonsense Generation framework, termed KGR^4, consisting of four stages: Retrieval, Retrospect, Refine, Rethink. Under this framework, we first perform retrieval to search for relevant sentences from external corpus as the prototypes. Then, we train the generator that either edits or copies these prototypes to generate candidate sentences, of which potential errors will be fixed by an autoencoder-based refiner. Finally, we select the output sentence from candidate sentences produced by generators with different hyper-parameters. Experimental results and in-depth analysis on the CommonGen benchmark strongly demonstrate the effectiveness of our framework. Particularly, KGR^4 obtains 33.56 SPICE points in the official leaderboard, outperforming the previously-reported best result by 2.49 SPICE points and achieving state-of-the-art performance.