Stage-wise Fine-tuning for Graph-to-Text Generation
This work addresses the challenge of fully leveraging graph structure in text generation for natural language processing applications, representing an incremental improvement over existing pre-trained language model approaches.
The paper tackled the problem of graph-to-text generation by proposing a structured model with a two-step fine-tuning mechanism and a novel tree-level embedding method to better utilize graph structure information, resulting in significant improvements in all text generation metrics on the English WebNLG 2017 dataset.
Graph-to-text generation has benefited from pre-trained language models (PLMs) in achieving better performance than structured graph encoders. However, they fail to fully utilize the structure information of the input graph. In this paper, we aim to further improve the performance of the pre-trained language model by proposing a structured graph-to-text model with a two-step fine-tuning mechanism which first fine-tunes the model on Wikipedia before adapting to the graph-to-text generation. In addition to using the traditional token and position embeddings to encode the knowledge graph (KG), we propose a novel tree-level embedding method to capture the inter-dependency structures of the input graph. This new approach has significantly improved the performance of all text generation metrics for the English WebNLG 2017 dataset.