Neural machine translation for fact-checking temporal claims

Computational fact-checking aims at supporting the verification process of textual claims

by exploiting trustworthy sources. However, there are large classes of complex claims that

cannot be automatically verified, for instance those related to temporal reasoning. To this

aim, in this work, we focus on the verification of economic claims against time series sources. Starting from given textual claims in natural language, we propose a neural machine translation approach to produce respective queries expressed in a recently proposed temporal fragment of the Datalog language. The adopted deep neural approach shows promising preliminary results for the translation of 10 categories of claims extracted from real use cases.