Improving Temporal Relation Extraction with a Globally Acquired Statistical Resource

Proceedings of NAACL-HLT 2018, pages 841–851 New Orleans, Louisiana, June 1 - 6, 2018. c⃝2018 Association for Computational Linguistics Improving Temporal Relation Extraction with a Globally Acquired Statistical Resource Qiang Ning,1 Hao Wu,2 Haoruo Peng,1 Dan Roth1,2 Department of Computer Science 1University of Illinois at Urbana-Champaign, Urbana, IL 61801, USA 2University of Pennsylvania, Philadelphia, PA 19104, USA {qning2,hpeng7}@illinois.edu, {haowu4,danroth}@seas.upenn.edu Abstract Extracting temporal relations (before, after, overlapping, etc.) is a key aspect of under- standing events described in natural language. We argue that this task would gain from the availability of a resource that provides prior knowledge in the form of the temporal order that events usually follow. This paper devel- ops such a resource – a probabilistic knowl- edge base acquired in the news domain – by extracting temporal relations between events from the New York Times (NYT) articles over a 20-year span (1987–2007). We show that ex- isting temporal extraction systems can be im- proved via this resource. As a byproduct, we also show that interesting statistics can be re- trieved from this resource, which can poten- tially benefit other time-aware tasks. The pro- posed system and resource are both publicly available1. 1 Introduction Time is an important dimension of knowledge rep- resentation. In natural language, temporal infor- mation is often expressed as relations between events. Reasoning over these relations can help figuring out when things happened, estimating how long things take, and summarizing the time- line of a series of events. Several recent SemEval workshops are a good showcase of the importance of this topic (Verhagen et al., 2007, 2010; Uz- Zaman et al., 2013; Llorens et al., 2015; Minard et al., 2015; Bethard et al., 2015, 2016, 2017). One of the challenges in temporal relation ex- traction is that it requires high-level prior knowl- edge of the temporal order that events usually fol- low. In Example 1, we have deleted events from several snippets from CNN, so that we cannot use our prior knowledge of those events. We are also 1http://cogcomp.org/page/publication_ view/830 told that e1 and e2 have the same tense, and e3 and e4 have the same tense, so we cannot resort to their tenses to tell which one happens earlier. As a result, it is very difficult even for humans to figure out the temporal relations (referred to as “TempRels” hereafter) between those events. This is because rich temporal information is encoded in the events’ names, and this often plays an in- dispensable role in making our decisions. In the first paragraph of Example 1, it is difficult to un- derstand what really happened without the actual event verbs; let alone the TempRels between them. In the second paragraph, things are even more in- teresting: if we had e3:dislike and e4:stop, then we would know easily that “I dislike” occurs af- ter “they stop the column”. However, if we had e3:ask and e4:help, then the relation between e3 and e4 is now reversed and e3 is before e4. We are in need of the event …

Content truncated to 500 words