Improving Distantly-Supervised Named Entity Recognition with Self-Collaborative Denoising Learning

Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 1084–1099 November 7–11, 2021. c⃝2021 Association for Computational Linguistics 1084 ReGen: Reinforcement Learning for Text and Knowledge Base Generation using Pretrained Language Models Pierre L. Dognin IBM Research pdognin@us.ibm.com Inkit Padhi IBM Research inkpad@ibm.com Igor Melnyk IBM Research igor.melnyk@ibm.com Payel Das IBM Research daspa@us.ibm.com Abstract Automatic construction of relevant Knowl- edge Bases (KBs) from text, and generation of semantically meaningful text from KBs are both long-standing goals in Machine Learning. In this paper, we present ReGen, a bidirec- tional generation of text and graph leveraging Reinforcement Learning (RL) to improve per- formance. Graph linearization enables us to re-frame both tasks as a sequence to sequence generation problem regardless of the genera- tive direction, which in turn allows the use of Reinforcement Learning for sequence train- ing where the model itself is employed as its own critic leading to Self-Critical Sequence Training (SCST). We present an extensive in- vestigation demonstrating that the use of RL via SCST benefits graph and text generation on WebNLG+ 2020 and TEKGEN datasets. Our system provides state-of-the-art results on WebNLG+ 2020 by significantly improv- ing upon published results from the WebNLG 2020+ Challenge for both text-to-graph and graph-to-text generation tasks. More details in https://github.com/IBM/regen. 1 Introduction Graph representation of knowledge is a power- ful tool to capture real-world information where complex relationships between node entities can be efficiently encoded. Automatic generation of Knowledge Bases (KBs) from free-form text and its counterpart of generating semantically relevant text from KBs are both active and challenging research topics. Recently, there has been an increased interest in leveraging Pretrained Language Models (PLMs) to improve performance for text generation from graph, or graph-to-text (G2T) task (Ribeiro et al., 2020). Indeed, large PLMs like T5 (Raffel et al., 2020) and BART (Lewis et al., 2020) that have been pretrained on vast amount of diverse and variedly structured data, are particularly good candidates for generating natural looking text from graph data. BART- and T5-related models have been em- ployed by top performers in public challenges such as the WebNLG+ 2020 Challenge (Castro Ferreira et al., 2020b) where both graph-to-text and text- to-graph (T2G) tasks are offered, under the names RDF-to-Text and Text-to-RDF (semantic parsing) respectively; RDF stands for Resource Descrip- tion Framework, a standard for describing web re- sources. One can notice that more teams entered the competition for the G2T task than for T2G as the latter is a much harder task. Best models gen- erally use PLMs and fine-tune them for the target modality at hand (either graph or text). This is pos- sible by re-framing the T2G and G2T generations as a sequence to sequence (Seq2Seq) generation problem, which suits fine-tuning PLMs well. One can therefore hope to leverage the large pretraining of PLMs to improve the overall generation quality. The Seq2Seq formulation requires any input graph to be linearized as a sequence, which is not unique. This creates an opportunity for data aug- mentation where multiple …

Content truncated to 500 words