Fake News Detection Using Machine Learning Methods

Nagaraja, Arun; Soumya, K.; Sinha, Anubhav; Kumar, Jain Vinay; Nayak, Prajwal

doi:10.1145/3460620.3460753

Cited by 10 publications

(1 citation statement)

References 81 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…After considerable testing, it was found that the Support Vector Machine (SVM) is superior than any other classi ers. SVM is a popular model for situations with binary and multiple categories [27] The produced dataset can be enhanced over time to achieve balance between the two classes. The news label that is either Fake or True in most dataset, can be divided into several classes.…”

Section: Machine Learning Classi Er Evaluationmentioning

confidence: 99%

Uncovering the Truth: A Deep Learning Ensemble Model for Identifying Fake News

Namatullah,

Khurshid,

Jaleel

et al. 2024

Preprint

View full text Add to dashboard Cite

The rapid outspread of misinformation and its continuous spreading on digital platforms have raised a serious concern due to its ability to create harmful effects. Over the past ten years, fake news has become increasingly popular in Pakistan. Now it’s a challenging task to identify or differentiate among fake news and real news. Several researchers have made tremendous advancements to detect misleading information in previous years, but due to the nature of the problem, there are still several unresolved problems. The main goal of this research is to create the detection dataset for Pakistani news by semantically extracting news data from various sources and through social media platforms. We have categorized the textual properties of news article. To evaluate our proposed dataset, we used various learning algorithms namely Naive Bayes, Support Vector Machine (SVM), Random Forrest, Logistic Regression, Recurrent neural network (RNN), Long short-term memory (LSTM) and Bidirectional Long short-term memory (Bi-LSTM). By merging them, we built an Ensemble Learning classifier (Meta Model) to achieve higher accuracy. Our findings proved that our proposed Deep Ensemble Learning model outperformed others with an accuracy of 89 percent. The outcomes also demonstrated that an ensemble model outperformed the individual Base model. The findings suggested that an ensemble model can maximize not only accuracy but it can also be used on small datasets. The metrics like accuracy, Precision, F1-score and recall are used to measure an effectiveness of each applied model.

show abstract

Section: Machine Learning Classi Er Evaluationmentioning

confidence: 99%