FarFetched: Entity-centric Reasoning and Claim Validation for the Greek Language based on Textually Represented Environments.

The abundance of online information narrows our collective attention span. We address the need for automated claim validation, based on the incorporated evidence from dynamic, textually represented environments created by online news sources. We present FarFetched, an entity-centric reasoning framework based on news, where latent connections between events, actions or statements are discovered via their identified entity mentions and are represented in a structured form with the help of a graph database. We propose an evidence construction approach that combines relevant extracts of the stored information from various online sources in order to support or refute a given claim in free text, relying on entity linking and semantic similarity. We then leverage textual entailment recognition to provide a measurable way for assessing whether this claim is plausible based on the constructed evidence. Our approach tries to fill the gap in automated claim validation for less-resourced languages and is showcased on the Greek language, complemented by the training of relevant semantic textual similarity (STS) and natural language inference (NLI) models that are evaluated on translated versions of common benchmarks.