Chaining Event Spans for Temporal Relation Grounding

TL;DR

This paper introduces the Timeline Reasoning Network (TRN), a novel model that improves temporal relation understanding by chaining event spans to generate timelines, outperforming previous methods on multiple tasks.

Contribution

The paper proposes TRN, a new two-step inductive reasoning model that chains questions to predict timelines, addressing answer overlap issues in temporal relation tasks.

Findings

TRN outperforms previous methods on TORQUE and TB-dense datasets.

Effective resolution of spurious answer overlaps.

Improved temporal relation grounding accuracy.

Abstract

Accurately understanding temporal relations between events is a critical building block of diverse tasks, such as temporal reading comprehension (TRC) and relation extraction (TRE). For example in TRC, we need to understand the temporal semantic differences between the following two questions that are lexically near-identical: "What finished right before the decision?" or "What finished right after the decision?". To discern the two questions, existing solutions have relied on answer overlaps as a proxy label to contrast similar and dissimilar questions. However, we claim that answer overlap can lead to unreliable results, due to spurious overlaps of two dissimilar questions with coincidentally identical answers. To address the issue, we propose a novel approach that elicits proper reasoning behaviors through a module for predicting time spans of events. We introduce the Timeline Reasoning Network (TRN) operating in a two-step inductive reasoning process: In the first step model initially answers each question with semantic and syntactic information. The next step chains multiple questions on the same event to predict a timeline, which is then used to ground the answers. Results on the TORQUE and TB-dense, TRC and TRE tasks respectively, demonstrate that TRN outperforms previous methods by effectively resolving the spurious overlaps using the predicted timeline.