Inter-sentence Relation Extraction with Document-level Graph Convolutional Neural Network

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 4309–4316 Florence, Italy, July 28 - August 2, 2019. c⃝2019 Association for Computational Linguistics 4309 Inter-sentence Relation Extraction with Document-level Graph Convolutional Neural Network Sunil Kumar Sahu1, Fenia Christopoulou1, Makoto Miwa2,3, Sophia Ananiadou1,* 1National Centre for Text Mining, School of Computer Science, The University of Manchester, United Kingdom 2Toyota Technological Institute, Nagoya, 468-8511, Japan 3Artificial Intelligence Research Center (AIRC), National Institute of Advanced Industrial Science and Technology (AIST), Japan {sunil.sahu, efstathia.christopoulou, sophia.ananiadou}@manchester.ac.uk makoto-miwa@toyota-ti.ac.jp Abstract Inter-sentence relation extraction deals with a number of complex semantic relationships in documents, which require local, non-local, syntactic and semantic dependencies. Exist- ing methods do not fully exploit such depen- dencies. We present a novel inter-sentence relation extraction model that builds a la- belled edge graph convolutional neural net- work model on a document-level graph. The graph is constructed using various inter- and intra-sentence dependencies to capture local and non-local dependency information. In or- der to predict the relation of an entity pair, we utilise multi-instance learning with bi-affine pairwise scoring. Experimental results show that our model achieves comparable perfor- mance to the state-of-the-art neural models on two biochemistry datasets. Our analysis shows that all the types in the graph are effective for inter-sentence relation extraction. 1 Introduction Semantic relationships between named entities of- ten span across multiple sentences. In order to extract inter-sentence relations, most approaches utilise distant supervision to automatically gen- erate document-level corpora (Peng et al., 2017; Song et al., 2018). Recently, Verga et al. (2018) introduced multi-instance learning (MIL) (Riedel et al., 2010; Surdeanu et al., 2012) to treat multiple mentions of target entities in a document. Inter-sentential relations depend not only on local but also on non-local dependencies. De- pendency trees are often used to extract local dependencies of semantic relations (Culotta and Sorensen, 2004; Liu et al., 2015) in intra-sentence ∗Corresponding author. Figure 1: Sentences with non-local dependencies between named entities. The red arrow represents a relation be- tween co-referred entities and yellow arrows represent se- mantically dependent relations. Example adapted from the CDR dataset (Wei et al., 2015). relation extraction (RE). However, such depen- dencies are not adequate for inter-sentence RE, since different sentences have different depen- dency trees. Figure 1 illustrates such a case be- tween Oxytocin and hypotension. To capture their relation, it is essential to connect the co-referring entities Oxytocin and Oxt. RNNs and CNNs, which are often used for intra-sentence RE (Zeng et al., 2014; dos Santos et al., 2015; Zhou et al., 2016b; Lin et al., 2016), are not effective on longer sequences (Sahu and Anand, 2018) thus failing to capture such non-local dependencies. We propose a novel inter-sentence RE model that builds a labelled edge Graph CNN (GCNN) model (Marcheggiani and Titov, 2017) on a document-level graph. The graph nodes corre- spond to words and edges represent local and non- local dependencies among them. The document- level graph is formed by connecting words with local dependencies from syntactic parsing and …

Content truncated to 500 words