UPV-28-UNITO at SemEval-2019 Task 7: Exploiting Post’s Nesting and Syntax Information for Rumor Stance Classification

Abstract

In the present paper we describe the UPV-28-UNITO system's submission to the Ru-morEval 2019 shared task. The approach we applied for addressing both the subtasks of the contest exploits both classical machine learning algorithms and word embeddings, and it is based on diverse groups of features: stylistic, lexical, emotional, sentiment, meta-structural and Twitter-based. A novel set of features that take advantage of the syntactic information in texts is moreover introduced in the paper.

FakeFlow: Fake News Detection by Modeling the Flow of Affective Information

“…Previous work on fake news detection is mainly divided into two main lines, namely with a focus on social media (Zubiaga et al, 2015;Aker et al, 2017;Ghanem et al, 2019) or online news articles (Tausczik and Pennebaker, 2010;Horne and Adali, 2017;Rashkin et al, 2017;Barrón-Cedeno et al, 2019). In this work we focus on the latter one.…”

Multilingual Irony Detection with Dependency Syntax and Neural Models

“…For example, Sidorov et al (2012) exploited syntactic dependency-based n-grams for general-purpose classification tasks, Socher et al (2013) investigated sentiment and syntax with the development of a sentiment treebank, and Kanayama and Iwamoto (2020) showed a pipeline method that makes the most of syntactic structures based on Universal Dependencies, achieving high precision in sentiment detection for 17 languages. Morphology and syntax have also been proved useful in a number of other tasks, such as rumor detection (Ghanem et al, 2019), authorship attribution (Posadas-Duran et al, 2014; and humor detection (Liu et al, 2018a). To the best of our knowledge, very few studies use syntactic information specifically for irony detection.…”

On the Detection of False Information: From Rumors to Fake News

“…• Top-k replies, likes, or re-tweets: Some approaches in rumors detection use the number of replies, likes, and re-tweets to detect rumors [78]. Thus, we extract top k replied, liked or re-tweeted tweets from each account to assess the accounts factuality.…”

“…Using this feature, we aim to detect the stance of the users regarding the different topics we extracted. To model the stance we use a set of stance lexicons employed in previous works [9,78] . Concretely, we focus on the following categories: belief, denial, doubt, fake, knowledge, negation, question, and report (8 Features).…”