A Hybrid Approach for Identifying Authentic News Using Deep Learning Methods on Popular Twitter Threads

摘要

Now a day’s social networks have become a crucial part and parcel in our day to day life. The gradual rise of the amount of information available in different kind of social media is the sole reason behind this impact. However, having the scope of accessing this huge amount of information so easily has made it complicated to find the contrast between counterfeit and an actual news. Some group of people often grabs this chance to share manipulated information to a huge number of people with an intent of destroying the image of any people, person or any specific groups. This results in creating such agendas which creates havoc to social welfare and peacefulness. Due to this fast circulation in such a small amount of time, it is almost impossible to manually detect the sources and authenticity of fake news effectively and efficiently. This is where computational and automated methods come handy to detect such kind of act in social media. We can decide whether any piece of text is either fabricated or not based on previously experienced real or fabricated news. In this study our contribution has three principal parts. First, we introduce our manually collected dataset and its details, then we go on to use some significant concepts of Natural Language Processing and Machine Learning models to train our model and lastly, we provided some comparison with the existing algorithms from different machine learning techniques which proves the superiority of our proposed model. Our model achieved a highest accuracy of 80% using Bayesian Classifier and 77% using a Hybrid Long Short-Term Memory Neural Network Architecture.