Deep learning Applications to classify Cross-Topic Natural Language Texts Based on Their Argumentative Form

Argument mining is a public blockchain for optimizing the analysis for reasons in broad data sets. Despite their utility for this role, most current techniques for contention processing are optimized for particular text types and fail when implemented in interdependent texts. This paper suggests a new sentences annotation structure that can be reliably applied to random Web texts by crowd staff. Commentaries for over 25,000 cases layer protocol contentious topics were gathered. Cross-topic tests show that our attn computer program generalizes the best to unknown subjects, outperforming standard BiLSTM prototypes by 6% in specificity and 11% in F-score.