Textual Entailment For Cybersecurity: An Applicative Case

Recognizing Textual Entailment (RTE) is the task of recognizing the relation between two sentences, in order to measure whether and to what extent one of the two is inferred from the other. It is used in many Natural Language Processing (NLP) tasks. In the last decades, with the digitization of many legal documents, NLP applied to the legal domain has became prominent, due to the need of knowing which norms are complied with in case other norms are. In this context, from a set of obligations that are known to be complied with, RTE may be used to infer which other norms are complied with as well. We propose a dataset, regarding cybersecurity controls, for RTE on the legal domain. The dataset has been constructed using information available online, provided by domain experts from NIST (https://www.nist.gov).