ATE_ABSITA dataset

The ATE_ABSITA dataset comprehends 4364 real user reviews, written in the Italian language, involving 23 products and collected from a famous eCommerce platform. Each sentence has been annotated by three subjects for its polarity, for the score assigned by the author, and for the presence of aspect terms. The dataset has been randomly divided into training, development and test sets. The data format used is NDJSON 1 with UTF-8 encoding and newline as delimiter. <p>The dataset has been produced and used in the context of the ATE_ABSITA task (http://www.di.uniba.it/~swap/ate_absita/index.html) during the Evalita 2020 evaluation campaign (http://www.evalita.it/2020).