Tools
Tools
Thakur, Sayali Machhindra (2024) Hybrid Approaches to Sarcasm Detection in Social Media: Comparing Rule-Based, Statistical, and Deep Learning Models. Masters thesis, Dublin, National College of Ireland.
Preview |
PDF (Master of Science)
Download (1MB) | Preview |
Preview |
PDF (Configuration Manual)
Download (1MB) | Preview |
Abstract
Instagram, YouTube, or Twitter—people regularly use sarcasm in their comments, processing sarcasm as a text-emoji interaction, which creates difficulties for sarcasm detection as a computational challenge. This research endeavour therefore seeks to overcome this challenge in the following manner: We assess the performance of several methods for identifying sarcasm that hail from various categories: rule-based systems, traditional machine learning, and the recent complex deep learning models. The first goal is to investigate how successful these models are in recognizing if the sarcastic comment is polite or rude, understanding a set of data from Twitter encompassing textual content and emojis. the study shows the relationship between the forms of words and contextual signals and the fact that state-of-the-art models are more effective than previous purely language-based ones in addressing this relationship. The findings show that incorporating text and emoji features improves the model’s recognition and assessment of sarcasm. This work offers theoretical suggestion concerning the linguistic and contextual characteristics of sarcasm on one hand and practical implications for sentiment analysis and social media moderation on the other hand. These features have become the object of future research with regard to the enhancement of cross-domain adaptability and refining model efficiency.
Actions (login required)
 |
View Item |