Detecting Microaggressions

• Author(s): Mbachu Chizaram
• Paper ID: 1709812
• Page: 531-559
• Published Date: 31-01-2022
• Published In: Iconic Research And Engineering Journals
• Publisher: IRE Journals
• e-ISSN: 2456-8880
• Volume/Issue: Volume 5 Issue 7 January-2022

Abstract

Problem: While we appreciate the numerous advantages and opportunities that using social media platforms provides, it is undeniable that many people are embarrassed, insulted, tormented, and harassed by anonymous users, strangers, or peers. These insults may be communicated openly or subtly, and while it may be easy to spot an insult outright in some cases, it may not be in others, this brings about the term microaggression. Microaggressions are subtle, small insults that are difficult to identify. To address this issue, the detection of microaggressions is coming to play, this detection has been a top concern in recent years, with the goal of reducing online toxicity and bullying, which are prevalent in today's environment. Objectives: While it is the intended goal of this dissertation is to develop a model using different machine learning algorithms to accurately detect microaggressions, I will be concentrating on comments that contain hate speech rather than microaggressions for a variety of reasons that will be explained in greater detail in other chapters as we progress through the dissertation. Hate speech, like microaggression, is difficult to define but is considered as any sort of verbal expression calling for cultural, racial, or religious violence. The difference between hate speech and microaggression is basically in its usage, while hate speech is overtly expressed, microaggression is covertly expressed Methodology: A pre-labelled dataset from https://www.kaggle.com/arkhoshghalb/twitter-sentiment-analysis-hatred-speech). was used and the terms used in the labelling were gotten from https://hatebase.org/ an organization that maintains a glossary of hate speech vocabulary. Each tweet is annotated for whether it contains hate speech or not, and the total number of tweets in this collection is a little over 31,000. A supervised machine learning method for detecting hate speech using machine learning algorithms such as Multinomial Nave Bayes, Random Forest Classifier, Extra Trees Classifier, Light Gradient Boosting (LGBM), Extreme Gradient Boosting (Xgboost), Adaptive Boosting (Adaboost), Logistic Regression, Logistic RegressionCV, Stochastic Gradient Descent (SGD) Classifier, and Support Vector Machine algorithms were all used. Achievements: The findings show that our suggested methodology is a viable solution for detecting hate speech in online social media platforms. Finally, I compared the outcome of the chosen model to model used by other academics.

Citations

IRE Journals:
Mbachu Chizaram "Detecting Microaggressions" Iconic Research And Engineering Journals Volume 5 Issue 7 2022 Page 531-559

IEEE:
Mbachu Chizaram "Detecting Microaggressions" Iconic Research And Engineering Journals, 5(7)