Natural Language Processing using Graph Neural Network for Text Classification
Vimal Kumar, Ahmed Al‐Emran, D.A. Karras, Shashi Kant Gupta, Chandra Kumar Dixit, Bhadrappa Haralayya
Abstract
The boom of the technological area has given rise to numerous new applications that actually rule the entire world. Some of them mainly are the social media networks like the Twitter, Gmail and Facebook. Lot of text and content are received to millions of people across the work. It has become very essential to be surer whether the content is fake, or isit malicious or is it a really important content. To classify these contents, text classification techniques are being widely used across the globe. This has invented a new technical world of Natural Language Processing where it becomes very important to process the texts and images before we actually use them. In this paper, we have introduced the use of Graph Neural Networks (GNN) to classify the text according to their content. The use of GNN is done as they work well for 2D vectors and texts are in two dimensions. Computation of Self-Organizing Maps (SOM) are done to compute the nearest neighbors in the graphs and it calculates the actual distances between the neighbors. This process is used in classifying the text and performs better when compared to the existing techniques.