AI-Driven Media Representation Analysis for Social Equity (AIRAS)

Research Question  •   How can natural language processing and computer vision techniques, together with explainable artificial intelligence techniques, assist in analysing gender portrayal in news content?

Traditional manual methods used to analyse how gender is represented in media can be inconsistent, lack scalability, and fail to capture subtle biases. This project addresses these limitations by proposing an AI-powered system to monitor gender portrayals within the Maltese online news platforms. The system will integrate natural language processing (NLP), computer vision (CV), and explainable AI (XAI) techniques to provide a nuanced and reliable analysis.Specifically, the project will investigate how sentiment analysis, named-entity recognition, and other NLP methods can reveal implicit gender biases in news text. Furthermore, CV models will be trained to identify visual disparities in the representation of men and women. XAI will be used to ensure transparency and explain the reasoning behind the system's outputs, hence ensuring the ethical alignment of this system.This project utilises a two-fold evaluation process. Media studies experts, journalists, and the general public in Malta will provide qualitative annotations, while the AI system's results will be compared to this ground truth for quantitative evaluation. The goal is to explore the feasibility and realisation of AI-powered tools that align with human perceptions and meaningfully support critical media analysis within the Maltese context, promoting greater awareness and a more equitable media landscape.

• Arrow Left

Exploring Visual Bias in News Content using Explainable AI (NBxAI)

Exploring Visual Bias in News Content using Explainable AI (NBxAI)  •  1 min read

The NBxAI project analyses news articles from six Maltese newspapers using advanced AI techniques. It employs named entity recognition, keyword extraction, and sentiment analysis on article text. The system also generates captions for images based on visual content and compares them with actual captions. Image-text matching is performed to assess the relevance of images to the article content.The project aims to detect potential biases by examining the relationship between article sentiment and image relevancy. Results showed varying performance across newspapers, with some displaying higher similarity scores between synthetic captions and actual content. This tool enables journalists and readers to critically evaluate the coherence and potential biases in news articles.This project is financed by Xjenza Malta through the FUSION: R&I Research Excellence Programme of 2023.

• Arrow Left

Data-Informed Media Analysis Suite (DIMAS)

Research Question  •   In the current context of information overload and diverse news sources, does an AI-powered portal with tailored analysis tools significantly increase the ability of journalists and the public to assess media content critically?

The DIMAS project aims to democratise access to Artificial Intelligence (AI) tools for fair journalism, empowering both journalists and the public to evaluate news content critically. It addresses the challenge of misinformation and bias in the digital age by integrating cutting-edge AI technologies into a user-friendly web platform. The project builds upon existing research from the University of Malta, leveraging AI models for bias detection, sentiment analysis, computer vision and natural language understanding. By incorporating large language models, DIMAS will enable conversational news exploration, personalised recommendations, and fact-checking capabilities. The project prioritises user-centred design, ensuring accessibility for both technical and non-technical users. Through rigorous evaluation and iterative development, DIMAS aims to foster a more informed, engaged, and critical media landscape, ultimately contributing to a more transparent and equitable society.

• Arrow Left

About this initiative

In today's digital age, the abundance of information and the prevalence of misinformation pose significant challenges to informed public discourse and democracy. The ability to critically evaluate news content and differentiate factual reporting from biased or misleading information is essential for citizens in a democratic society.To address this challenge, the Department of Artificial Intelligence at the University of Malta is undertaking a research initiative focused on developing innovative AI-powered tools for news analysis. Dr Dylan Seychell, a lecturer in the same department, is leading this initiative together with a team of other academics, research support officers and industry collaborators. Our focus is on developing cutting-edge AI tools to analyse and understand how news and media content is created and consumed.Our ApproachWe believe in a collaborative approach to research and development. Therefore, we actively engage with journalists and media entities throughout our research process. This collaboration ensures that our AI tools are relevant to the needs of the journalism profession and contribute to strengthening journalistic practices.Furthermore, we conduct public focus groups to gather diverse perspectives and ensure our research remains relevant and accessible to a wide audience. This participatory approach allows us to develop AI tools that truly empower both journalists and the public to navigate the complexities of news in the digital age.Our ProjectsOur current research projects include:- NBxAI: This project analyses news articles from six Maltese newspapers using advanced AI techniques, including named entity recognition, keyword extraction, and sentiment analysis. By examining the relationship between article sentiment and image relevancy, NBxAI aims to detect potential biases and empower journalists and readers to critically evaluate news content.- AIRAS: This project addresses the limitations of traditional media analysis methods by developing an AI-powered system to monitor gender portrayals in Maltese online news platforms. Integrating natural language processing, computer vision, and explainable AI, this system aims to provide a nuanced and reliable analysis of gender representation, promoting greater awareness and a more equitable media landscape.- DIMAS: This project aims to democratise access to AI tools for fair journalism, empowering both journalists and the public to critically evaluate news content. By integrating cutting-edge AI technologies into a user-friendly web platform, DIMAS will provide bias detection, sentiment analysis, and fact-checking capabilities, fostering a more informed and critical media landscape.These projects are currently in their research phase. We invite individuals with an interest in media literacy, artificial intelligence, and the intersection of these fields to join our growing community. By participating in this initiative, you can contribute to the development of tools that have the potential to transform news consumption and promote informed citizenship.

Related Publications:

- D. Seychell, G. Hili, J. Attard and K. Makantatis, "AI as a Tool for Fair Journalism: Case Studies from Malta," 2024 IEEE Conference on Artificial Intelligence (CAI), Singapore, Singapore, 2024, pp. 127-132- G. Hili and D. Seychell, "Using Machine Learning to Investigate Potential Image Bias in News Articles," 2024 IEEE 22nd Mediterranean Electrotechnical Conference (MELECON), Porto, Portugal, 2024, pp. 174-179

• Arrow Left