Scientific Researcher
Research work across artificial intelligence, educational media, and graph-based learning systems.
HET MEHTA
Het Darshan Mehta
Het Darshan MehtaPhD ResearcherAI Research | Algorithmic Fairness & Bias Detection | Recommender SystemsMagdeburg / Braunschweig, Germany
Published AI Research
Publications and projects across recommender systems, LLMs, NLP, and graph-based personalization.
Human-Centered AI Tutor
Tutored HCAI, HCNLP, and human-centered technology courses at Otto-von-Guericke University.
About
Research Profile
I am a Scientific Researcher at the Leibniz Institute for Educational Media | Georg Eckert Institute and a PhD student at Otto von Guericke University Magdeburg. My current work focuses on measuring and evaluating algorithmic fairness in graph neural networks, as well as developing fairness-aware models for bias detection in educational textbooks. My doctoral research examines multi-group and multi-class fairness for knowledge-aware recommender systems. I have also supported teaching in Human-Centered Artificial Intelligence, Human-Centered Natural Language Processing, and Human-Centered Approaches and Technology.
Selected Work
Research & Scientific Publications
Scientific Publication
Large language model enhanced embeddings for knowledge aware recommender systems
Journal Article
Discover Artificial Intelligence, Springer Nature, 2026
Het Darshan Mehta, Marco Polignano, Giovanni Semeraro, Ernesto William De Luca
Open LinkScientific Publication
When Multimodal Features Do Not Improve Fairness in Learning Systems
Conference Workshop Paper
MAAI4AIED - Multimodal Affect in AI for Education Design, Application, and Ethical Implications, Festival of Learning 2026
Het Darshan Mehta, Ernesto William De Luca
Link can be added in portfolio-data.ts.
Scientific Publication
A Comprehensive Strategy to Bias and Mitigation in Human Resource Decision Systems
Workshop Paper
5th Italian Workshop on Explainable Artificial Intelligence, co-located with AIxIA 2024, Bolzano, Italy
Silvia D'Amicantonio, Mishal Kizhakkam Kulangara, Het Darshan Mehta, Shalini Pal, Marco Levantesi, Marco Polignano, Erasmo Purificato, Ernesto William De Luca
Open LinkScientific Publication
Temporal Linguistic Escalation Patterns in Clinical Narratives Prior to Clinical Deterioration
Accepted Conference Paper
CLiC-it 2026 - Twelfth Italian Conference on Computational Linguistics, Palermo, Italy
Faizan Nadeem Mufti, Ajithkumar Jayasree Jayakrishna, Het Darshan Mehta, Marco Polignano, Giovanni Semeraro, Ernesto William De Luca
Link can be added in portfolio-data.ts.
Scientific Publication
A Comprehensive Evaluation Framework for Multi-Level Bias Analysis in Graph-Based Personalization Systems
Workshop Paper
8th International Workshop on Explainable User Modeling and Personalised Systems, co-located with UMAP 2026, Gothenburg, Sweden
Hrushikesh Ahire, Gavin Rony Correia, Pinky Sherwani, Het Darshan Mehta, Marco Polignano, Giovanni Semeraro, Ernesto William De Luca
Open LinkCareer
Experience
Scientific Researcher
Leibniz Institute for Educational Media | Georg Eckert Institute
Feb 2026 - Present
Braunschweig, Germany
- Measuring and evaluating algorithmic fairness in graph neural networks.
- Developing fairness-aware models for bias detection in educational textbooks.
Tutor: HCAI, HCNLP, HCAT
Otto-von-Guericke University
Oct 2024 - Jan 2026
Magdeburg, Germany
- Supported teaching across Human-Centered AI, Natural Language Processing, and Human-Centered Approaches and Technology.
Research Student
Leibniz Institute for Neurobiology
May 2024 - Jan 2025
Magdeburg, Germany
- Measured and processed real-time brain signals at the Brain-Machine Interface AG.
- Supported neurofeedback software development for a mobile EEG device.
Graduate Intern
Siemens Energy and Performance
Dec 2022 - Aug 2023
Mumbai, India
- Designed digital twins for optimization of a pasteurization system at AMUL.
- Contributed to a factory digitization project at Hindustan Unilever, Pune.
Applied Systems
Projects
MEG-RW
Multi-Group Exposure calibrated Graph Reweighting for Fairness in Recommendation.
MMRec-FAIR
Fairness in multimodal recommender systems.
EUI_LLM_KARS
Knowledge-aware recommender system combining knowledge graph embeddings and large language models.
LLM-KG-Extraction
Pipeline for extracting and evaluating knowledge graphs from the MovieLens 1M dataset with large language models.
Deep-Reinforcement-Learning-A2C-
An implementation of Advantage Actor Critic (A2C) reinforcement learning.
Evaluating-Algorithmic-Fairness-in-GNNs
Evaluation of algorithmic fairness in graph neural networks using the Adult dataset and mitigation strategies.
Technical Depth
Skills
Languages
PythonC/C++MySQLHTMLCSS
AI & Deep Learning
PyTorchTensorFlowGNNsA2CA3CGraph Contrastive Learning
NLP & Recommenders
RAGSentence-BERTTransformersFairness MetricsBias Mitigation
Knowledge & Reasoning
Knowledge GraphsPrompt EngineeringCoTKGoTTree-of-Thoughts
Frameworks & Developer Tools
ClayRSElliotDockerLangChainNeo4jOllamaCoppeliaSimOpenAI API
Libraries & APIs
MMRecRecBolefastFMSentence-TransformersTransformersscikit-learnpandasNumPyMatplotlibSeabornTogether AI APIs
Public Datasets
MovieLensDBbooksLastFM
Academic Path
Education
Feb 2026 - Present
PhD Student, Multi-group/Multi-class Fairness for Knowledge-Aware Recommender Systems
Otto-von-Guericke University
Magdeburg, Germany
Oct 2023 - Jan 2026
M.Sc. Digital Engineering
Otto-von-Guericke University
Magdeburg, Germany
May 2025 - Oct 2025
Erasmus+, Computer Science
University of Bari Aldo Moro
Bari, Italy
2019 - 2023
B.E. Electronics and Communication
Dharmsinh Desai University
Nadiad, India
Contact
Let's Connect
Focused on AI fairness, human-centered AI, educational technology, and trustworthy recommender systems.
Languages Spoken
English C1German A2Italian A1Hindi C2Gujarati C2
Emailhet.mehta@ovgu.dePhone+49 176 77595077ORCID0009-0000-8372-3877
LocationMagdeburg / Braunschweig, Germany