Minimize Exposure Bias of Seq2Seq Models in Joint Entity and Relation Extraction

Joint entity and relation extraction aims to extract relation triplets from\nplain text directly. Prior work leverages Sequence-to-Sequence (Seq2Seq) models\nfor triplet sequence generation. However, Seq2Seq enforces an unnecessary order\non the unordered triplets and involves a large decoding length associated with\nerror accumulation. These introduce exposure bias, which may cause the models\noverfit to the frequent label combination, thus deteriorating the\ngeneralization. We propose a novel Sequence-to-Unordered-Multi-Tree\n(Seq2UMTree) model to minimize the effects of exposure bias by limiting the\ndecoding length to three within a triplet and removing the order among\ntriplets. We evaluate our model on two datasets, DuIE and NYT, and\nsystematically study how exposure bias alters the performance of Seq2Seq\nmodels. Experiments show that the state-of-the-art Seq2Seq model overfits to\nboth datasets while Seq2UMTree shows significantly better generalization. Our\ncode is available at https://github.com/WindChimeRan/OpenJERE .\n