Improving Slot Filling Performance with Attentive Neural Networks on Dependency Structures

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 2588–2597 Copenhagen, Denmark, September 7–11, 2017. c⃝2017 Association for Computational Linguistics Improving Slot Filling Performance with Attentive Neural Networks on Dependency Structures Lifu Huang1∗, Avirup Sil2, Heng Ji1, Radu Florian2 1 Rensselaer Polytechnic Institute, 2 IBM T.J. Watson Research Center 1{huangl7,jih}@rpi.edu, 2{avi,raduf}@us.ibm.com Abstract Slot Filling (SF) aims to extract the values of certain types of attributes (or slots, such as person:cities of residence) for a given entity from a large collection of source documents. In this paper we propose an effective DNN architecture for SF with the following new strategies: (1). Take a regu- larized dependency graph instead of a raw sentence as input to DNN, to compress the wide contexts between query and candi- date filler; (2). Incorporate two attention mechanisms: local attention learned from query and candidate filler, and global at- tention learned from external knowledge bases, to guide the model to better se- lect indicative contexts to determine slot type. Experiments show that this frame- work outperforms state-of-the-art on both relation extraction (16% absolute F-score gain) and slot filling validation for each individual system (up to 8.5% absolute F- score gain). 1 Introduction The goal of Slot Filling (SF) is to extract pre-defined types of attributes or slots (e.g., per:cities of residence) for a given query entity from a large collection of documents. The slot filler (attribute value) can be an entity, time expres- sion or value (e.g., per:charges). The TAC-KBP slot filling task (Ji et al., 2011a; Surdeanu and Ji, 2014) defined 41 slot types, including 25 types for person and 16 types for organization. One critical component of slot filling is rela- tion extraction, namely to classify the relation be- tween a pair of query entity and candidate slot ∗ This work was carried out during an internship at IBM Research. filler into one of the 41 types or none. Most pre- vious studies have treated SF in the same way as within-sentence relation extraction tasks in ACE 1 or SemEval (Hendrickx et al., 2009). They created training data based on crowd-sourcing or distant supervision, and then trained a multi-class classi- fier or multiple binary classifiers for each slot type based on a set of hand-crafted features. Although Deep Neural Networks (DNN) such as Convolutional Neural Networks (CNN) and Re- current Neural Networks (RNN) have achieved state-of-the-art results on within-sentence relation extraction (Zeng et al., 2014; Liu et al., 2015; Santos et al., 2015; Nguyen and Grishman, 2015; Yang et al., 2016; Wang et al., 2016), there are lim- ited studies on SF using DNN. Adel and Sch¨utze (2015) and Adel et al. (2016) exploited DNN for SF but did not achieve comparable results as tradi- tional methods. In this paper we aim to answer the following questions: What is the difference between SF and ACE/SemEval relation extraction task? How can we make DNN work for SF? We argue that SF is different and more chal- lenging than traditional relation extraction. First, …

Content truncated to 500 words