Corey Harper et. al. won the best task paper award at SemEval 2021! See https://semeval.github.io/SemEval2021/awards for the conference organizer’s writeup on it, and https://aclanthology.org/2021.semeval-1.38.pdf for the paper itself.

SemEval (SEMantic EVALuations) is a highly-regarded conference where each year a number of tasks are proposed for some kind of semantic evaluation of text, images, or data. The tasks are challenges where different research groups develop competing systems. The task organizers provide training and testing data, as well as an evaluation procedure, for the different systems to use. After the end of the competition period, they write a paper on the task and on the results of the various competitors. Corey proposed a task on extracting measurements and surrounding context from STM text. For example, in addition to spotting measurements like 273.6 MPa, additional information on the property being measured (e.g. fracture pressure), the entity being measured (concrete with a certain additive), and other special information on conditions (.e.g that the samples were cured at a temperature of -10 degrees C). We expect this kind of additional information will allow us to merge measurements described in the running text with measurements provided databases or in tables in other articles. We think that will be very important as Elsevier continues to develop it data and analytics strategy.