Scientific paper classification using Convolutional Neural Networks
Monir Ech-Chouyyekh, Hicham Omara, Mohamed Lazaar · 2019
The Convolutional Neural Network (CNN), a class of artificial neural networks, has become dominant in various fields, including analysis and text processing. It designed to learn, automatically and adaptively, the spatial hierarchies of backscattered entities using multiple building blocks, such as convolutional layers, grouping layers, and fully connected layers. This article aims to present an approach based on CNN to classify scientific articles by their domains (7 different domains) from their abstracts, the process will base on several features extracted automatically from their summaries. The proposed approach has shown remarkable results for the automatic classification of text compared to other usual automatic learning algorithms.