Automatic generation of summaries from multiple news articles is a valuable tool as the number of online publications grows rapidly. Single document summarization (SDS) systems have benefited from advances in neural encoder-decoder model thanks to the availability of large datasets. However, multidocument summarization (MDS) of news articles has been limited to datasets of a couple of hundred examples. In this paper, we introduce Multi-News, the first large-scale MDS news dataset. Additionally, we propose an end-to-end model which incorporates a traditional extractive summarization model with a standard SDS model and achieves competitive results on MDS datasets. We benchmark several methods on Multi-News and release our data and code in hope that this work will promote advances in summarization in the multidocument setting 1 .

devices, [26,27] to small-scale nano- [28][29][30] and DNA origamis. [31] A common theme in these studies is to exploit the sophisticated shape transformations from folding. For example, an origami robot is typically fabricated in a 2D flat configuration and then folded into the prescribed 3D shape to perform its tasks. The origamis have been treated essentially as linkage mechanisms in which rigid facets rotate around hingelike creases (aka "rigid-folding origami"). Elastic deformation of the constituent sheet materials or the dynamics of folding are often neglected. Such a limitation in scope indeed resonates the origin of this field, that is, folding was initially considered as a topic in geometry and kinematics.However, the increasingly diverse applications of origami require us to understand the force-deformation relationship and other mechanical properties of folded structures. Over the last decade, studies in this field started to expand beyond design and kinematics and into the domain of mechanics and dynamics. Catalyzed by this development, a family of architected origami materials quickly emerged (Figure 1). These materials are essentially assemblies of origami sheets or modules with carefully designed crease patterns. The kinematics of folding still plays an important role in creating certain properties of these origami materials. For example, rigid folding of the classical Miura-ori sheet induces an in-plane deformation pattern with auxetic properties (aka negative Poisson's ratios). [32,33] However, elastic energy in the deformed facets and creases, combined with their intricate spatial distributions, impart the origami materials with a rich list of desirable and even unorthodox properties that were never examined in origami before. For example, the Ron-Resch fold creates a unique tri-fold structure where pairs of triangular facets are oriented vertically to the overall origami sheet and pressed against each other. Such an arrangement can effectively resist buckling and create very high compressive load bearing capacity. [34] Other achieved properties include shape-reconfiguration, tunable nonlinear stiffness and dynamic characteristics, multistability, and impact absorption.Since the architected origami materials obtain their unique properties from the 3D geometries of the constituent sheets or modules, they can be considered a subset of architected cellular solids or mechanical metamaterials. [35][36][37][38][39] However, the origami materials have many unique characteristics. The rich geometries of origami offer us great freedom to tailor targeted Origami, the ancient Japanese art of paper folding, is not only an inspiring technique to create sophisticated shapes, but also a surprisingly powerful method to induce nonlinear mechanical properties. Over the last decade, advances in crease design, mechanics modeling, and scalable fabrication have fostered the rapid emergence of architected origami materials. These materials typically consist of folded origami sheets or modules with intricate 3D geomet...

No abstract

The Singapore Strait is considered as the bottleneck and chokepoint of the shipping routes connecting the Indian and the Pacific Ocean. Therefore, the ship collision risk assessment is of significant importance for ships passing through the narrow, shallow, and busy waterway. In this paper, three ship collision risk indices are initially proposed to quantitatively assess the ship collision risks in the Strait: index of speed dispersion, degree of acceleration and deceleration, and number of fuzzy ship domain overlaps. These three risk indices for the Singapore Strait are estimated by using the real-time ship locations and sailing speeds provide by Lloyd's MIU automatic identification system (AIS). Based on estimation of these three risk indices, it can be concluded that Legs 4W, 5W, 11E, and 12E are the most risky legs in the Strait. Therefore, the ship collision risk reduction solutions should be prioritized being implemented in these four legs. This study also finds that around 25% of the vessels sail with a speed in excess of the speed limit, which results in higher potentials of ship collision. Analysis indicates that the safety level would be significantly improved if all the vessels follow the passage guidelines.

We present CoSQL, a corpus for building cross-domain, general-purpose database (DB) querying dialogue systems. It consists of 30k+ turns plus 10k+ annotated SQL queries, obtained from a Wizard-of-Oz (WOZ) collection of 3k dialogues querying 200 complex DBs spanning 138 domains. Each dialogue simulates a real-world DB query scenario with a crowd worker as a user exploring the DB and a SQL expert retrieving answers with SQL, clarifying ambiguous questions, or otherwise informing of unanswerable questions. When user questions are answerable by SQL, the expert describes the SQL and execution results to the user, hence maintaining a natural interaction flow. CoSQL introduces new challenges compared to existing task-oriented dialogue datasets: (1) the dialogue states are grounded in SQL, a domain-independent executable representation, instead of domain-specific slotvalue pairs, and (2) because testing is done on unseen databases, success requires generalizing to new domains. CoSQL includes three tasks: SQL-grounded dialogue state tracking, response generation from query results, and user dialogue act prediction. We evaluate a set of strong baselines for each task and show that CoSQL presents significant challenges for future research. The dataset, baselines, and leaderboard will be released at https:// yale-lily.github.io/cosql.