CLASSIfying Microaggressions Using MACHINE LEARNING Algorithms.

ABL-Micro: Opportunities for Affective AI Built Using a Multimodal Microaggression Dataset

Interdisciplinary research has begun to study how technology can assist humans with improving their communications and reducing racist, sexist, and/or hate speech. Many of these technologies are built using textual examples taken from social media statuses and updates. Models are rarely built containing multimodal examples that may provide more context into abusive speech. This paper explores the creation of a multimodal dataset of microaggressions built from listening and annotating speech from popular American television shows, and also from mining text from websites containing microaggressions. American television shows were chosen because they are readily available online and provide context that often mimics natural human conversations. The dataset, called ABL-MICRO, contains over three thousand text and sound instances of racist, homophobic, and sexist remarks, mostly geared towards people of color and women. Finally, a discussion over opportunities for researchers to begin to analyze affective content from this dataset is provided. Read More...

Citation: Washington, G., Mance, G., Aryal, S., Ngueajio, M., Salaam, C., & Alim, C. (2021). ABL-MICRO: Opportunities for Affective AI Built Using a Multimodal Microaggression Dataset. In Proceedings of the AAAI-21 Workshop on Affective Content Analysis, New York, USA, AAAI.

Classifying Microaggressions using SVM

A microaggression is an action or statement often described as unintentional discrimination prevailing against marginalized groups of people, with the offender generally unaware of their impact on the recipient. Most people are exposed to these derogatory speeches but lack proper understanding and means to respond to them. Hence, victims of these Microaggressions usually “manage and deal with it” altogether. Microaggressions are shown in verbal(speech) and non-verbal forms (text), and researchers have found that long-term exposure can result in hypertension, fatigue, low self-esteem, and even suicide. In this research, we use Artificial Intelligence (AI), Machine Learning (ML) algorithms, and Natural Language Processing (NLP) techniques to identify and predict two different types of microaggressions- Racial microaggression and Gender microaggression. The main goal of this research is to help humans become more aware of what microaggressions are, to improve our emotional intelligence and empathy vis-a-vis one another. Our proposed methodology involves the creation of a racial and gender datasets, which we later combine to form a binary classification dataset. The racial dataset is created by extracting microaggressive dialogues from famous American TV sitcoms, while the gender dataset is web scraped from the website - ‘microagression.com’. Then, we build an automated racial and gender microaggression identification tool, by applying a ‘Support Vector Machine (SVM)’ algorithm to our dataset. The SVM was trained and tested on the corpus using five and ten-fold cross-validation. An F1 score of 65.9% and an accuracy score of 72.8% was achieved on the test set. We believe the results from these projects could help create smarter diversity and inclusion training tools and applications that will lead to more productive, empathetic, and self-aware employees across the nation These results could also play an important role in filtering out harmful content as part of speech recognition technologies such as Grammerly.