Veracity vs. Reliability: Changing the Approach of Our Annotation Guideline
| Autor: | Bonet-Jover, Alba |
|---|---|
| Přispěvatelé: | Universidad de Alicante. Departamento de Lenguajes y Sistemas Informáticos |
| Rok vydání: | 2022 |
| Předmět: | |
| Zdroj: | RUA. Repositorio Institucional de la Universidad de Alicante Universidad de Alicante (UA) |
| Popis: | This paper presents the evolution of an annotation guideline designed for the disinformation detection task, an essential step of my doctoral thesis. The annotation proposal aims to label all the structural and content elements of a news item, as well as to classify them as Reliable or Unreliable. The initial objective was to annotate those elements into Fake or True, but for that classification, world knowledge is needed. Our current goal is to annotate news on the basis of a purely textual, semantic and linguistic analysis, without using external knowledge and, for that reason, the annotation was redirected towards a reliability rating, rather than a veracity classification. This article justifies the change of perspective at this stage of the thesis, explains the difference between veracity and reliability and shows the concrete changes that have been adopted in our annotation proposal with this new approach. This research work has been partially funded by the Spanish Government and Fondo Europeo de Desarrollo Regional (FEDER) through the project Modelang: Modeling the behavior of digital entities by Human Language Technologies (RTI2018-094653-B-C22) as well as supported by a grant from the Consellería de Innovación, Universidades, Ciencia y Sociedad Digital (ACIF/2020/177) from the Spanish Government. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|