Context-Aware Representations for Knowledge Base Relation Extraction

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 1784–1789 Copenhagen, Denmark, September 7–11, 2017. c⃝2017 Association for Computational Linguistics Context-Aware Representations for Knowledge Base Relation Extraction Daniil Sorokin and Iryna Gurevych Ubiquitous Knowledge Processing Lab (UKP) Research Training Group AIPHES Department of Computer Science Technische Universit¨at Darmstadt www.ukp.tu-darmstadt.de Abstract We demonstrate that for sentence-level re- lation extraction it is beneficial to consider other relations in the sentential context while predicting the target relation. Our architecture uses an LSTM-based encoder to jointly learn representations for all rela- tions in a single sentence. We combine the context representations with an attention mechanism to make the final prediction. We use the Wikidata knowledge base to construct a dataset of multiple relations per sentence and to evaluate our approach. Compared to a baseline system, our method results in an average error reduction of 24% on a held-out set of relations. The code and the dataset to replicate the experiments are made available at https://github.com/ukplab. 1 Introduction The main goal of relation extraction is to determine a type of relation between two target entities that appear together in a text. In this paper, we consider the sentential relation extraction task: to each oc- currence of the target entity pair ⟨e1,e2⟩ in some sentence s one has to assign a relation type r from a given set R (Hoffmann et al., 2011). A triple ⟨e1,r,e2⟩ is called a relation instance and we refer to the relation of the target entity pair as target re- lation. Relation extraction is a fundamental task that enables a wide range of semantic applications from question answering (Xu et al., 2016) to fact checking (Vlachos and Riedel, 2014). For relation extraction, it is crucial to be able to extract relevant features from the sentential context (Riedel et al., 2010; Zeng et al., 2015). Modern ap- proaches focus just on the relation between the tar- get entities and disregard other relations that might be present in the same sentence (Zeng et al., 2015; Lin et al., 2016). For example, in order to correctly identify the relation type between the movie e1 and the director e2 in (1), it is important to separate out the INSTANCE OF relation between the movie and its type e3: (1) [e1 Star Wars VII] is an American [e3 space opera epic film] directed by [e2 J. J. Abrams]. We present a novel architecture that considers other relations in the sentence as a context for predicting the label of the target relation. We use the term context relations to refer to them throughout the pa- per. Our architecture uses an LSTM-based encoder to jointly learn representations for all relations in a single sentence. The representation of the target relation and representations of the context relations are combined to make the final prediction. To facilitate the experiments we construct a dataset that contains multiple positive and nega- tive relation instances per sentence. We employ a fast growing community managed knowledge base (KB) Wikidata (Vrandeˇci´c …

Content truncated to 500 words