Timeline Summarization identifies major events from a news collection and describes them following temporal order, with key dates tagged. Previous methods generally generate summaries separately for each date after they determine the key dates of events. These methods overlook the events’ intra-structures (arguments) and inter-structures (event-event connections). Following a different route, we propose to represent the news articles as an event-graph, thus the summarization becomes compressing the whole graph to its salient sub-graph. The key hypothesis is that the events connected through shared arguments and temporal order depict the skeleton of a timeline, containing events that are semantically related, temporally coherent and structurally salient in the global event graph. A time-aware optimal transport distance is then introduced for learning the compression model in an unsupervised manner. We show that our approach significantly improves on the state of the art on three real-world datasets, including two public standard benchmarks and our newly collected Timeline100 dataset.

Multi-hop reasoning approaches over knowledge graphs infer a missing relationship between entities with a multi-hop rule, which corresponds to a chain of relationships. We extend existing works to consider a generalized form of multi-hop rules, where each rule is a set of relation chains. To learn such generalized rules efficiently, we propose a two-step approach that first selects a small set of relation chains as a rule and then evaluates the confidence of the target relationship by jointly scoring the selected chains. A game-theoretical framework is proposed to this end to simultaneously optimize the rule selection and prediction steps. Empirical results show that our multi-chain multi-hop (MCMH) rules result in superior results compared to the standard single-chain approaches, justifying both our formulation of generalized rules and the effectiveness of the proposed learning framework.

Introducing common sense to natural language understanding systems has received increasing research attention. It remains a fundamental question on how to evaluate whether a system has the sense-making capability. Existing benchmarks measure common sense knowledge indirectly or without reasoning. In this paper, we release a benchmark to directly test whether a system can differentiate natural language statements that make sense from those that do not make sense. In addition, a system is asked to identify the most crucial reason why a statement does not make sense. We evaluate models trained over large-scale language modeling tasks as well as human performance, showing that there are different challenges for system sense-making.

General Question Answering (QA) systems over texts require the multi-hop reasoning capability, i.e. the ability to reason with information collected from multiple passages to derive the answer. In this paper we conduct a systematic analysis to assess such an ability of various existing models proposed for multi-hop QA tasks. Specifically, our analysis investigates that whether providing the full reasoning chain of multiple passages, instead of just one final passage where the answer appears, could improve the performance of the existing QA models. Surprisingly, when using the additional evidence passages, the improvements of all the existing multi-hop reading approaches are rather limited, with the highest error reduction of 5.8% on F1 (corresponding to 1.3% improvement) from the BERT model. To better understand whether the reasoning chains indeed could help find the correct answers, we further develop a co-matching-based method that leads to 13.1% error reduction with passage chains when applied to two of our base readers (including BERT). Our results demonstrate the existence of the potential improvement using explicit multi-hop reasoning and the necessity to develop models with better reasoning abilities.

Multi-hop question answering (QA) requires an information retrieval (IR) system that can find multiple supporting evidence needed to answer the question, making the retrieval process very challenging. This paper introduces an IR technique that uses information of entities present in the initially retrieved evidence to learn to ‘hop’ to other relevant evidence. In a setting, with more than 5 million Wikipedia paragraphs, our approach leads to significant boost in retrieval performance. The retrieved evidence also increased the performance of an existing QA model (without any training) on the benchmark by 10.59 F1.

Discourse structure is a key aspect of all forms of text, providing valuable information both to humans and machines. We applied the hierarchical theory of news discourse developed by van Dijk to examine how paragraphs operate as units of discourse structure within news articles—what we refer to here as document-level discourse. This document-level discourse provides a characterization of the content of each paragraph that describes its relation to the events presented in the article (such as main events, backgrounds, and consequences) as well as to other components of the story (such as commentary and evaluation). The purpose of a news discourse section is of great utility to story understanding as it affects both the importance and temporal order of items introduced in the text—therefore, if we know the news discourse purpose for different sections, we should be able to better rank events for their importance and better construct timelines. We test two hypotheses: first, that people can reliably annotate news articles with van Dijk’s theory; second, that we can reliably predict these labels using machine learning. We show that people have a high degree of agreement with each other when annotating the theory (F1 > 0.8, Cohen’s kappa > 0.6), demonstrating that it can be both learned and reliably applied by human annotators. Additionally, we demonstrate first steps toward machine learning of the theory, achieving a performance of F1 = 0.54, which is 65% of human performance. Moreover, we have generated a gold-standard, adjudicated corpus of 50 documents for document-level discourse annotation based on the ACE Phase 2 corpus.