PROJECT TITLE :

Deep Feature-Based Text Clustering and Its Explanation

ABSTRACT:

The text mining community has devoted a significant amount of time and energy to the research of text clustering as it is an essential step in the text data analysis process. The majority of the text clustering algorithms that are currently in use are based on a model called the bag-of-words, which suffers from problems such as high-dimensionality and sparsity and ignores text structural and sequence information. The models that are based on deep learning, such as convolutional neural networks and recurrent neural networks, regard texts as sequences; however, these models do not have supervised signals and do not produce results that can be explained. In this paper, we propose a deep feature-based text clustering (DFTC) framework that integrates pretrained text encoders into text clustering tasks. DFTC stands for deep features, deep features-based text clustering, and deep feature-based text clustering. The dependence on supervision is broken with the help of this model, which is predicated on sequence representations. The results of the experiments show that our model performs better than traditional text clustering algorithms and the most advanced pretrained language model available today, known as BERT, on almost all of the datasets that were taken into consideration. In addition, understanding the principles underlying the deep learning approach is significantly aided by the explanation of the clustering results. The explanation module that is included in our proposed framework for clustering is designed to provide users with assistance in comprehending the significance of the clustering results as well as their overall quality.