Deceptive Report Analysis for English and German Language Using Machine Learning Algorithms

In today's contemporary age, where the internet is omnipresent, individuals depend on a multitude of online sources to stay informed. With the proliferation of social media platforms such as Facebook, Twitter, and others, news dissemination has become remarkably rapid, reaching millions of users in a remarkably short time. The widespread circulation of false information carries significant ramifications, including the formation of biased opinions and the potential to influence election results in favor of specific candidates. Additionally, spammers exploit enticing news headlines to generate revenue through click-bait advertisements. This study focuses on applying Artificial Intelligence, Machine Learning and NLP concepts to accomplish binary classification of diverse online news articles. Our objective is to empower users with the ability to distinguish between fake and genuine news while also verifying the credibility of the publishing website.In today's era, users of these social media platforms are generating and exchanging a vast amount of information in different languages, a portion of which is deceptive and lacks any connection to the truth. Determining whether a text article is misinformation or disinformation presents a formidable challenge when relying solely on human judgment. Even domain experts need to carefully consider multiple factors before reaching a conclusion regarding the article's veracity. In this research, we propose the utilization of a machine learning ensemble approach to automatically classify news reports. Our research explores different textual attributes that can successfully distinguish between counterfeit and authentic content in English and German language. Leveraging these properties, we train an ensemble of diverse machine learning algorithms using different ensemble methods and assess their performance on various real-world datasets. It pertains to intentionally and provably false accounts that are largely propagated through social media networks. Given its immense power to influence, it is imperative to devise tactics for discerning and critically assessing news encountered on social media platforms.While each study concentrated on English language evaluations, we paid particular attention to German language-based datasets in order to evaluate the algorithms' performance objectively and classify the data based on the authenticity of the reviews. In order to classify the German reviews as fraudulent, we looked for two or more special characters. Given that NLP models learn best from a range of datasets, we have added a German dataset to our training set. This makes it easier to build models that can effectively understand and produce German language content.