Deep learning applications to classify cross-topic natural language texts based on their argumentation

Abstract

This assignment poses a challenging but essential problem for the rapidly developing discipline of natural language processing (NLP), which tries to classify information that is contested within texts that cover a variety of subject domains. The assignment in question seeks to classify information that is contested within texts. Enhanced Deep Inference for Text Segmentation (EDITS) is the name of the model that was used in this research project to demonstrate how deep learning can be integrated with more conventional approaches to machine learning such as Random Forest, Support Vector Machine, and Naive Bayes. The purpose of this research was to demonstrate the usefulness of deep learning in practice. The naive Bayes algorithm, support vector machines, and random forests are some of the approaches that fall under this category. In order to extract features and semantic information from the writings that are a part of the Quantitative Argumentation Dataset (QuAD), which is made of writings on a wide variety of topics, we use methods such as term frequency-inverse document frequency (TF-IDF) analysis. This allows us to extract information from the writings that is relevant to our research. This allows us to extract information from the texts with a greater degree of precision. Because of this, we are able to glean information from the sentences in a more accurate and exact manner. In this study, we compare a deep learning model called EDITS, which was developed specifically for lectures, to models that are more traditional. EDITS was created with the intention of improving speech recognition. This research expands our understanding of argumentation in a wide variety of contexts and demonstrates how cross-topic dispute may be classified via EDITS in real-world scenarios.

Topics & Concepts