Improving Distant Supervision for Information Extraction Using Label Propagation Through Lists

Improving Distant Supervision for Information Extraction Using Label Propagation Through Lists Lidong Bing§ Sneha Chaudhari§ Richard C. Wang♮ William W. Cohen§ §Carnegie Mellon University, Pittsburgh, PA 15213 ♮US Development Center, Baidu USA, Sunnyvale, CA 94089 §{lbing@cs, sschaudh@andrew, wcohen@cs}.cmu.edu ♮richardwang@baidu.com Abstract Because of polysemy, distant labeling for information extraction leads to noisy train- ing data. We describe a procedure for re- ducing this noise by using label propaga- tion on a graph in which the nodes are entity mentions, and mentions are cou- pled when they occur in coordinate list structures. We show that this labeling ap- proach leads to good performance even when off-the-shelf classifiers are used on the distantly-labeled data. 1 Introduction In distantly-supervised information extraction (IE), a knowledge base (KB) of relation or concept instances is used to train an IE sys- tem. For instance, a set of facts like adverse- EffectOf(meloxicam, stomachBleeding), interacts- With(meloxicam, ibuprofen), might be used to train an IE system that extracts these relations from documents. In distant supervision, instances are first matched against a corpus, and the match- ing sentences are then used to generate training data consisting of labeled entity mentions. For in- stance, matching the KB above might lead to la- beling passage 1 from Table 1 as support for the fact adverseEffectOf(, stomachBleeding). A weakness of distant supervision is that it pro- duces noisy training data: for instance, match- ing the adverse effect weakness might lead to incorrectly-labeled mention examples. Distant su- pervision is often coupled with learning methods that allow for this sort of noise by introducing la- tent variables for each entity mention (e.g., (Hoff- mann et al., 2011; Riedel et al., 2010; Surdeanu et al., 2012)); by carefully selecting the entity mentions from contexts likely to include specific KB facts (Wu and Weld, 2010); by careful filter- 1. “Avoid drinking alcohol. It may increase your risk of stomach bleeding.” 2. “Get emergency medical help if you have chest pain, weakness, shortness of breath, slurred speech, or problems with vision or balance.” Table 1: Passages from a page discussing the drug meloxicam. ing of the KB strings used as seeds (Movshovitz- Attias and Cohen, 2012); or by making use of named-entity linking methods and co-reference to improve the matching phase of distant learning (Koch et al., 2014). Here we explore an alternative approach of Distant IE using coordinate-term Lists (DIEL) based on detection of lists in text, such as the one illustrated in passage 2 in Table 1. Since list items are usually of the same type, the unambiguous mention chest pain here disambiguates the men- tion weakness. Label propagation methods (Zhu et al., 2003; Lin and Cohen, 2010) can be used to exploit this intuition, by propagating the low- confidence labels associated with distance super- vision through an appropriate graph. Here we describe a pipelined system which (1) identifies lists of semantically-related items us- ing lexico-syntactic patterns (2) uses distant su- pervision, in combination with a label-propagation method, to find entity mentions that can be confi- dently …

Content truncated to 500 words