Uncover the Ground-Truth Relations in Distant Supervision: A Neural Expectation-Maximization Framework

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pages 326–336, Hong Kong, China, November 3–7, 2019. c⃝2019 Association for Computational Linguistics 326 Uncover the Ground-Truth Relations in Distant Supervision: A Neural Expectation-Maximization Framework Junfan Chen BDBC and SKLSDE Beihang University, China chenjf@act.buaa.edu.cn Richong Zhang∗ BDBC and SKLSDE Beihang University, China zhangrc@act.buaa.edu.cn Yongyi Mao School of EECS University of Ottawa, Canada ymao@uottawa.ca Hongyu Guo Digital Technologies Research Center National Research Council Canada, Canada hongyu.guo@nrc-cnrc.gc.ca Jie Xu School of Computing University of Leeds, United Kingdom j.xu@leeds.ac.uk Abstract Distant supervision for relation extraction en- ables one to effectively acquire structured re- lations out of very large text corpora with less human efforts. Nevertheless, most of the prior- art models for such tasks assume that the given text can be noisy, but their corresponding la- bels are clean. Such unrealistic assumption is contradictory with the fact that the given la- bels are often noisy as well, thus leading to significant performance degradation of those models on real-world data. To cope with this challenge, we propose a novel label-denoising framework that combines neural network with probabilistic modelling, which naturally takes into account the noisy labels during learning. We empirically demonstrate that our approach significantly improves the current art in uncov- ering the ground-truth relation labels. 1 Introduction Relation extraction aims at automatically extract- ing semantic relationships from a piece of text. Consider the sentence “Larry Page, the chief ex- ecutive officer of Alphabet, Google’s parent com- pany, was born in East Lansing, Michigan.”. The knowledge triple (Larry Page, employed by, Google) can be extracted. Despite various ef- forts in building relation extraction models (Ze- lenko et al., 2002; Zhou et al., 2005; Bunescu and Mooney, 2005; Zeng et al., 2014; dos Santos et al., 2015; Yang et al., 2016; Liu et al., 2015; Xu et al., 2015; Miwa and Bansal, 2016; G´abor et al., 2018), the difficulty of obtaining abundant training data with labelled relations remains a challenge, and thus motivates the development of Distant Su- pervision relation extraction (Mintz et al., 2009; Riedel et al., 2010; Zeng et al., 2015; Lin et al., 2016; Ji et al., 2017; Zeng et al., 2018; Feng et al., ∗Corresponding author 2018; Wang et al., 2018b,a; Hoffmann et al., 2011; Surdeanu et al., 2012; Jiang et al., 2016; Ye et al., 2017; Su et al., 2018; Qu et al., 2018). Distant su- pervision (DS) relation extract methods collect a large dataset with “distant” supervision signal and learn a relation predictor from such data. In detail, each example in the dataset contains a collection, or bag, of sentences all involving the same pair of entities extracted from some corpus (e.g., news reports). Although such a dataset is expected to be very noisy, one hopes that when the dataset is large enough, useful correspondence between the semantics of a sentence and the relation label it implies still reinforces and manifests itself. De- spite their capability of …

Content truncated to 500 words