Truth of Varying Shades: Analyzing Language in Fake News and
            Political Fact-Checking

Rashkin, Hannah; Choi, Eunsol; Jang, Jin Yea; Volkova, Svitlana; Choi, Yejin

doi:10.18653/v1/d17-1317

Cited by 561 publications

(327 citation statements)

References 21 publications

Supporting

Mentioning

299

Contrasting

Unclassified

Order By: Relevance

“…We call this value the propaganda index, since it reflects the probability for an article to have a propagandistic intent. We use four families of features: Word n-gram features We use tf.idf -weighted word [1, 3]-grams (Rashkin et al 2017).…”

Section: Propaganda Index Computationmentioning

confidence: 99%

Proppy: A System to Unmask Propaganda in Online News

Barrón-Cedeño

Martino

Jaradat

et al. 2019

AAAI

View full text Add to dashboard Cite

We present proppy, the first publicly available real-world, real-time propaganda detection system for online news, which aims at raising awareness, thus potentially limiting the impact of propaganda and helping fight disinformation. The system constantly monitors a number of news sources, deduplicates and clusters the news into events, and organizes the articles about an event on the basis of the likelihood that they contain propagandistic content. The system is trained on known propaganda sources using a variety of stylistic features. The evaluation results on a standard dataset show stateof-the-art results for propaganda detection.

show abstract

Section: Propaganda Index Computationmentioning

confidence: 99%

Proppy: A System to Unmask Propaganda in Online News

Barrón-Cedeño

Martino

Jaradat

et al. 2019

AAAI

View full text Add to dashboard Cite

show abstract

“…The syntax, semantics, and style of the written text can provide significant information about the intention of the authors. It has been widely observed that the language and tone of fake news presentation are more aggressive in general, and it involves a choice of words depicting strong emotions and biases (Rashkin et al, 2017). Our model uses a deep, bidirectional LSTM architecture.…”

Section: Methodsmentioning

confidence: 99%

“…Although the dataset was developed for a similar problem, we made slight modifications to make it more generalizable. For example, we removed all sentences which were labeled as satire as we theorize that satire is more of a linguistic phenomenon (intended for humor) than fake news (Rashkin et al, 2017).…”

Section: Datasetmentioning

confidence: 99%

Towards automatic fake news classification

Ghosh

Shah

2018

Proc. Assoc. Info. Sci. Tech.

View full text Add to dashboard Cite

The interaction of technology with humans has many adverse effects. The rapid growth and outreach of the social media and the Web have led to the dissemination of questionable and untrusted content among a wider audience, which has negatively influenced their lives and judgment. Many research studies have been conducted to tackle the detection and spreading of fake news, which is misinformation that looks genuine. While the first step of such tasks would be to classify claims associated based on their credibility, the next steps would involve identifying hidden patterns in style, syntax, and content of such news claims. We propose a generalized method based on Deep Neural Networks to detect if a given claim is fake or genuine. We have used a modular approach by combining techniques from information retrieval, natural language processing, and deep learning. Our classifier comprises two main submodules. The first submodule uses the claim to retrieve relevant articles from the knowledge base which can then be used to verify the truth of the claim. It also uses word‐level features for prediction. The second submodule uses a deep neural network to learn the underlying style of fake content. Our experiments conducted on benchmark datasets show that for the given classification task we can obtain up to 82.4% accuracy by using a combination of two models; the first model was up to 72% accurate while the second model was around 81% accurate. Our detection model has the potential to automatically detect and prevent the spread of fake news, thus, limiting the caustic influence of technology in the human lives.

show abstract

“…Deep semantics Ma et al 2016Ma et al 2018Wu et al 2018Reis et al 2019Rashkin et al 2017Martin et al 2018Tseng et al 1999Karimi et al 2018Long et al 2017Shu et al 2019 Profiles Influence Interests Yang et al 2018Ghenai et al 2018 Dynamic networks Static networks Jin et al 2016Tacchini et al 2017Ma et al 2017Ruchansky et al 2017Wu et al 2018 Stance-based Ma et al 2018Lukasik et al 2019 Meta-data based Figure 2. The review of information credibility evaluation methods Wu et al, 2019a).…”

Section: Shallow Semanticsmentioning

confidence: 99%

Discovering differential features: Adversarial learning for information credibility evaluation

Rao

Nazir

et al. 2020

Information Sciences

View full text Add to dashboard Cite

A series of deep learning approaches extract a large number of credibility features to detect fake news on the Internet. However, these extracted features still suffer from many irrelevant and noisy features that restrict severely the performance of the approaches. In this paper, we propose a novel model based on Adversarial Networks and inspirited by the Shared-Private model (ANSP), which aims at reducing common, irrelevant features from the extracted features for information credibility evaluation. Specifically, ANSP involves two tasks: one is to prevent the binary classification of true and false information for capturing common features relying on adversarial networks guided by reinforcement learning. Another extracts credibility features (henceforth, private features) from multiple types of credibility information and compares with the common features through two strategies, i.e., orthogonality constraints and KL-divergence for making the private features more differential. Experiments first on two six-label LIAR and Weibo datasets demonstrate that ANSP achieves the state-of-theart performance, boosting the accuracy by 2.1%, 3.1%, respectively and then on four-label Twit-ter16 validate the robustness of the model with 1.8% performance improvements.

show abstract

Truth of Varying Shades: Analyzing Language in Fake News and Political Fact-Checking

Cited by 561 publications

References 21 publications

Proppy: A System to Unmask Propaganda in Online News

Proppy: A System to Unmask Propaganda in Online News

Towards automatic fake news classification

Discovering differential features: Adversarial learning for information credibility evaluation

Contact Info

Product

Resources

About