|
A Step towards Automated Support in Assessing Emotional Mirroring Adriana-Mihaela Guran, Grigoreta-Sofia Cojocar, Dan Cojocar |
97-112 |
|
FootballInsight: A Web Portal for Football Match Analysis and Machine Learning-Driven Prediction Cristian-Răzvan Taudor, Paul Stefan Popescu |
113-130 |
Babeș-Bolyai University
1 M. Kogălniceanu, Cluj-Napoca, Romania
Abstract. Empathy is a central construct in social cognition and is defined as the ability to recognize and adequately react emotionally to an affective message transferred by a human counterpart by sharing their emotion mirroring). In the therapy process, empathy plays an important role in building a strong connection between the participants. In this paper, we describe a support system that provides visual representations of the emotional mirroring between two people based on the analysis and correlation of video and audio data and EMG sensor outputs. A set of metrics related to the emotion evolution of the two participants is proposed, and the identified correlations on a set of 10 dyads are presented.
Keywords: AI, emotions, mirroring, empathy, automation
Cite this paper as:
Guran, A.-M., Cojocar, G.-S., Cojocar, D. A Step towards Automated Support in Assessing Emotional Mirroring.
International Journal of User-System Interaction 17(4),
97-112, 2024.
DOI: 10.37789/ijusi.2023.17.4.1
University of Craiova
Al. I. Cuza Street, 13, Craiova, Romania
Abstract. The increasing availability of sports data has created new opportunities for building interactive analytical tools that bridge the gap between raw statistics and actionable insights. This paper presents FootballInsight, a full-stack web application that integrates machine learning (ML)-based outcome prediction within a user-centered, transparent interface for European football match analysis. Unlike existing closed platforms such as Opta Sports or SofaScore, FootballInsight exposes prediction rationale through contextual statistics, head-to-head history, and expected goals visualization, enabling users to understand the factors driving each prediction. The architecture combines a React/Vite frontend, a Node.js orchestration backend, and a Python ML microservice providing three complementary prediction types: match outcome (1/X/2 classification), total goals category (Under 2.5 / 2–3 / Over 3.5), and expected goals per team. Four ML algorithms were trained and evaluated on a dataset of European league matches from 2011 to 2025. XGBoost (Chen & Guestrin, 2016) achieved the best performance (accuracy: 0.68, macro F1-score: 0.66). A retrospective heuristic evaluation using Nielsen's ten usability heuristics (Nielsen, 1994) confirms the interface's strong alignment with established usability principles, while identifying targeted areas for improvement.
Keywords: football match prediction, machine learning, human-computer interaction, XGBoost, sports analytics, transparency in AI, web application, usability heuristics
Cite this paper as:
Taudor, C.-R., Popescu, P. S. FootballInsight: A Web Portal for Football Match Analysis and Machine Learning-Driven Prediction.
International Journal of User-System Interaction 17(4),
113-130, 2024.
DOI: 10.37789/ijusi.2023.17.4.2