Generative adversarial networks for open information extraction

Abstract

Open information extraction (Open IE) is a core task of natural language processing (NLP). Even many efforts have been made in this area, and there are still many problems that need to be tackled. Conventional Open IE approaches use a set of handcrafted patterns to extract relational tuples from the corpus. Secondly, many NLP tools are employed in their procedure; therefore, they face error propagation. To address these problems and inspired by the recent success of Generative Adversarial Networks (GANs), we employ an adversarial training architecture and name it Adversarial-OIE. In Adversarial-OIE, the training of the Open IE model is assisted by a discriminator, which is a (Convolutional Neural Network) CNN model. The goal of the discriminator is to differentiate the extraction result generated by the Open IE model from the training data. The goal of the Open IE model is to produce high-quality triples to cheat the discriminator. A policy gradient method is leveraged to co-train the Open IE model and the discriminator. In particular, due to insufficient training, the discriminator usually leads to the instability of GAN training. We use the distant supervision method to generate training data for the Adversarial-OIE model to solve this problem. To demonstrate our approach, an empirical study on two large benchmark dataset shows that our approach significantly outperforms many existing baselines.

Abstractive Open Information Extraction

“…Although not explicitly defined as such, existing neural models often treat OpenIE as a labeling problem, where tokens are labeled as being part of the subject, predicate, or object of a relation (Kolluru et al, 2020a;Vasilkovsky et al, 2022). Even in cases where OpenIE is defined as a generative problem, the generated relations don't contain words outside the vocabulary of the original sentence (Kolluru et al, 2020b) (Han and Wang, 2021). Due to the labeling problem definition, prior neural OpenIE models struggle to extract relations with predicates that don't appear in the original sentence.…”

Semi-supervised generative open information extraction

“…Adversarial-OIE [44] is based on Generative Adversarial Network (GAN) [45]. The model aims to obtain a generator which can generate tuples so similar to the gold annotations that a discriminator cannot distinguish them.…”