This paper describes the SemEval 2016 shared task on Aspect Based Sentiment Analysis (ABSA), a continuation of the respective tasks of 2014 and 2015. In its third year, the task provided 19 training and 20 testing datasets for 8 languages and 7 domains, as well as a common evaluation procedure. From these datasets, 25 were for sentence-level and 14 for text-level ABSA; the latter was introduced for the first time as a subtask in SemEval. The task attracted 245 submissions from 29 teams.

Several medications commonly used for a number of medical conditions share a property of functional inhibition of acid sphingomyelinase (ASM), or FIASMA. Preclinical and clinical evidence suggest that the (ASM)/ceramide system may be central to SARS‐CoV‐2 infection. We examined the potential usefulness of FIASMA use among patients hospitalized for severe COVID‐19 in an observational multicenter study conducted at Greater Paris University hospitals. Of 2,846 adult patients hospitalized for severe COVID‐19, 277 (9.7%) were taking a FIASMA medication at the time of their hospital admission. The primary endpoint was a composite of intubation and/or death. We compared this endpoint between patients taking vs. not taking a FIASMA medication in time‐to‐event analyses adjusted for sociodemographic characteristics and medical comorbidities. The primary analysis was a Cox regression model with inverse probability weighting (IPW). Over a mean follow‐up of 9.2 days (SD=12.5), the primary endpoint occurred in 104 patients (37.5%) receiving a FIASMA medication, and 1,060 patients (41.4%) who did not. Despite being significantly and substantially associated with older age and greater medical severity, FIASMA medication use was significantly associated with reduced likelihood of intubation or death in both crude (HR=0.71; 95%CI=0.58‐0.87; p<0.001) and primary IPW (HR=0.58; 95%CI=0.46‐0.72; p<0.001) analyses. This association remained significant in multiple sensitivity analyses and was not specific to one particular FIASMA class or medication. These results show the potential importance of the ASM/ceramide system in COVID‐19 and support the continuation of FIASMA medications in these patients. Double‐blind controlled randomized clinical trials of these medications for COVID‐19 are needed.

Objective: Preliminary data from different cohorts of small sample size or with short follow-up indicate poorer prognosis in people with obesity compared with other patients. This study aims to precisely describe the strength of association between obesity in patients hospitalized with coronavirus disease 2019 (COVID-19) and mortality and to clarify the risk according to usual cardiometabolic risk factors in a large cohort. Methods: This is a prospective cohort study including 5,795 patients aged 18 to 79 years hospitalized from February 1 to April 30, 2020, in the Paris area, with confirmed infection by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Adjusted regression models were used to estimate the odds ratios (ORs) and 95% CIs for the mortality rate at 30 days across BMI classes, without and with imputation for missing BMI values. Results: Eight hundred ninety-one deaths had occurred at 30 days. Mortality was significantly raised in people with obesity, with the following ORs for BMI of 30 to 35 kg/m 2 , 35 to 40 kg/m 2 , and >40 kg/m 2 : 1.89 (95% CI: 1.45-2.47), 2.79 (95% CI: 1.95-3.97), and 2.55 (95% CI: 1.62-3.95), respectively (18.5-25 kg/m 2 was used as the reference class). This increase holds for all age classes. Conclusions: Obesity doubles mortality in patients hospitalized with COVID-19.

We present a neural architecture for containment relation identification between medical events and/or temporal expressions. We experiment on a corpus of deidentified clinical notes in English from the Mayo Clinic, namely the THYME corpus. Our model achieves an F-measure of 0.613 and outperforms the best result reported on this corpus to date.

This paper presents a generative model to event schema induction. Previous methods in the literature only use head words to represent entities. However, elements other than head words contain useful information. For instance, an armed man is more discriminative than man. Our model takes into account this information and precisely represents it using probabilistic topic distributions. We illustrate that such information plays an important role in parameter estimation. Mostly, it makes topic distributions more coherent and more discriminative. Experimental results on benchmark dataset empirically confirm this enhancement.