A pipeline combining DeBERTa text embeddings, Node2Vec user embeddings, and CNN/BiRNN classifiers detects rumors with 0.758/0.823 accuracy on Twitter15/16, while SparseShield reduces simulated spread by roughly half.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SI 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
CleanNews: a Network-aware Fake News Mitigation Architecture for Social Media
A pipeline combining DeBERTa text embeddings, Node2Vec user embeddings, and CNN/BiRNN classifiers detects rumors with 0.758/0.823 accuracy on Twitter15/16, while SparseShield reduces simulated spread by roughly half.