What is Fake News?
The purposeful presenting of (usually) incorrect or misleading statements as news, where the assertions are deceptive by design, is known as fake news. How did digital media and news evolve? Newspapers, tabloids, and magazines gave way to digital news platforms, blogs, social media feeds, and a plethora of news mobile apps. News organizations profited from the increasing usage of social media and mobile platforms by offering their subscribers up-to-date news almost instantly. Customers can now more easily obtain the most recent news at their fingertips. Therefore, the ease with which these digital media platforms can be accessed by everyone and their capacity to facilitate user discussion and idea sharing as well as debate about topics like such as democracy, education, health, research and history. Apart from this benefit, however, false or fake news pieces are becoming more and more prevalent on digital platforms. They are primarily used negatively for their own gain, including financial and political gain, the creation of prejudiced attitudes, the manipulation of mindsets, and the dissemination of ridiculousness.
Anyone can disseminate false and biased information due to the quick uptake of the Internet, social media, and digital platforms (like Facebook, Twitter, news portals, or other social media). It is difficult to stop the creation of fake news. The past ten years have seen a sharp rise in the dissemination of fake news, which now affects a wide range of topics including politics, sports, health, history, entertainment, and science and research. Consider the 2016 US presidential election, which was heavily influenced by the publication of false and slanted news. Another example would be COVID-19. Every day, we typically encounter a lot of false or misleading information that could have major repercussions, incite panic among the populace, and hasten the spread of the epidemic.
As a result, it's critical to recognize and distinguish between fake and legitimate news. One approach is to assess each piece of news by an expert and fact-checker, but this takes time and involves expertise that is not transferable. Second, by utilizing machine learning and artificial intelligence tools, we can automate the detection of fake news. While there are many different unstructured format data types (such documents, films, and audios) in online news material, we will focus on text format news here. Natural language processing has advanced to the point where we can now recognize when an article or statement is false or misleading. The goal of this field's extensive research and experimentation is to identify fake news across all media (text, audio, and video).
In my study, I employed the Kaggle Fake News dataset to apply the Deep Learning Technique Sequence to Sequence programming to identify news stories that lacked credibility as fake news.
A full training dataset with the following attributes
id : unique id for a news article
title: the title of a news article
author: author of the news article
text : the text of the article; could be incomplete
label : a label that marks the article as potentially unreliable
1 : unreliable
0 : reliable
Tech Stack
Language : Python
Libraries : Scikit-learn , Tensorflow , Keras, Glove, nltk, pandas, numpy
Understanding the problem statement
Understanding the Sequence problem and their types
Understanding Sequence neural network approach like: RNN, GRU and LSTM
Importing the dataset and importing libraries
Performing basic text cleaning
Perform text preprocessing: Stop word removal, Stemming etc
Text tokenization using keras tokenizer
Sequence data preparation with tokenizer and padding
Train and test split for model validation
Explaining Text vectorization and Word embedding
Build Word embedding layer with Glove
Important parameter before train model
Explaining Sequence model steps
Implementing Simple RNN
Implementing LSTM and GRU
Making Test predictions using the trained model
Comparing Model training and validation loss and performance
Comparing LSTM and GRU performance and computation Time