A Novel Stacking Approach for Accurate Detection of Fake News

Abstract

With the increasing popularity of social media, people has changed the way they access news. News online has become the major source of information for people. However, much information appearing on the Internet is dubious and even intended to mislead. Some fake news are so similar to the real ones that it is difficult for human to identify them. Therefore, automated fake news detection tools like machine learning and deep learning models have become an essential requirement. In this paper, we evaluated the performance of five machine learning models and three deep learning models on two fake and real news datasets of different size with hold out cross validation. We also used term frequency, term frequencyinverse document frequency and embedding techniques to obtain text representation for machine learning and deep learning models respectively. To evaluate models' performance, we used accuracy, precision, recall and F1-score as the evaluation metrics and a corrected version of McNemar's test to determine if models' performance is significantly different. Then, we proposed our novel stacking model which achieved testing accuracy of 99.94% and 96.05 % respectively on the ISOT dataset and KDnugget dataset. Furthermore, the performance of our proposed method is high as compared to baseline methods. Thus, we highly recommend it for fake news detection.

Thai Fake News Detection Based on Information Retrieval, Natural Language Processing and Machine Learning

“…Equations ( 1) and ( 2) represent news document vectors and a news query document vector, respectively. Equation ( 3) is a proximity formula based on cosine similarity to measure the similarity among vectors, while ( 4) is used in (3). Equation ( 3) determines the relevant news document vectors returned by the web crawlers.…”

“…For feature extraction, Algorithm 2, we use words that usually appear on fake or real news. The feature extraction takes acts when the web crawlers retrieve the news contents relevant to the query based on the truncated cosine similarity defined in (3). We used ( 7)-( 13) for feature extraction from each news document.…”

“…The readers can easily access from any place at any time via the Internet. People are now comfortable accessing online news and can quickly share the news contents across the social network media such as WWW, Google, YouTube, Google+, Facebook, Twitter, Instagram, and Line [ 3 , 4 ]. Fake news is a threat to democracy around the world, which has weakened the confidence of governments, newspapers, and civil society.…”

A Systematic Review of Machine Learning Algorithms in Cyberbullying Detection: Future Directions and Challenges

“…Early detection of different social anomalies, including cyberbullying, hate speech [13,14], trolling [15], fake news [16], rumors [17], counterfeit profile detection [18], misogyny [19], etc., is becoming a trend in recent social media-based research. Abusive text can be detected in the messaging/comments on social media by maintaining a list of offensive words.…”

Multi-Level Multi-Modal Cross-Attention Network for Fake News Detection

“…This method exploits latent relationships among users to model the influence of the users with high prestige on the other users for detecting fake news efficiently. To obtain text representation for fake news detection, Jiang et al [17] propose a stacking model based on five machine learning models and three deep learning models, such as CNN, LSTM, SVM, etc. Umer et al [18] design a hybrid Neural Network architecture combining the capabilities of CNN and LSTM, which is used with two different dimensionality reduction approaches containing Principle Component Analysis (PCA) and Chi-Square to reduce the dimensionality of the feature vectors.…”