NetGuardAI is an intelligent, distributed system that leverages advanced deep learning and Natural Language Processing to detect harmful online content such as disinformation, hate speech, and propaganda. By combining high-precision AI detection with innovative social network immunization strategies, the project aims to actively halt the spread of toxic behavior and shield society from its negative political, economic, and psychological impacts.
Enhance AI Detection Accuracy: Design and train novel deep neural network architectures and machine learning algorithms to detect various forms of harmful content with exceptionally high precision.
Develop Network Immunization Strategies: Formulate innovative graph-immunization algorithms and strategies to proactively block and halt the spread of toxic messages across social networks.
NetGuardAI project goes beyond simple filtering to comprehensively understand complex context, syntax, and intent across diverse harmful categories (such as trolling, hate speech, and coordinated propaganda). The project aims to move the paradigm from merely tagging toxic behavior to actively immunizing the network, thereby isolating malicious nodes before their content goes viral.
Elena-Simona Apostol, Alin-Georgian Pisică, Ciprian-Octavian Truică. ATESA-BÆRT: A heterogeneous ensemble learning model for Aspect-Based Sentiment Analysis. Knowledge-Based Systems, 326:1-13(113987), ISSN 0950-7051, September 2025. DOI: 10.1016/j.knosys.2025.113987 (Q1 Journal)
Elena-Simona Apostol, Ciprian-Octavian Truică, Mariana Damova, Purificação Silvano, Giedre Valunaite Oleškeviciene, Chaya Liebeskind, Dimitar Trajanov, Anna Baczkowska, Emma Angela Montecchiari, Christian Chiarcos. Multiword Discourse Markers Across Languages: A Linguistic and Computational Perspective, International Journal of Applied Linguistics, Wiley, 35(4):2078-2090, ISSN 0802-6106, November 2025 (Accepted for publication April 2025). DOI: 10.1111/ijal.12755 (Q1 Journal) [pdf]
Ana-Maria Luisa Mocanu, Sebastian Mocanu, Ciprian-Octavian Truică, Elena-Simona Apostol. IROH: Insightful Ranking of Humor using Multi-Stage Hybrid Retrieval with Rationale-Distilled LLM Judges for JOKER 2026 Track Task 1 English, Working Notes of the Conference and Labs of the Evaluation Forum (CLEF 2026), September 2026
Maria-Diana Cotelin, Ciprian-Octavian Truică, Elena-Simona Apostol. Awakened at CheckThat! 2026: X-IBIS: an Iterative Bi-Encoder Fine-Tuning with Hard-Negative Mining for Cross-Lingual Scientific Source Retrieval, Working Notes of the Conference and Labs of the Evaluation Forum (CLEF 2026), September 2026
Dragoș-Mitruț Vasile, Elena-Simona Apostol, Ciprian-Octavian Truică. Awakened at HIPE-2026: Knowledge-Augmented Prompting and Logic-Constrained Fine-Tuning for Person-Place Relation Extraction, Working Notes of the Conference and Labs of the Evaluation Forum (CLEF 2026), September 2026
Anamaria Vișan, Ciprian-Octavian Truică, Adrian Paschke, Elena-Simona Apostol. VANGUARD at eRisk 2026 Task 1: Conversational Depression Detection with LLM Personas, Working Notes of the Conference and Labs of the Evaluation Forum (CLEF 2026), September 2026
Anamaria Vișan, Ciprian-Octavian Truică, Adrian Paschke, Elena-Simona Apostol. VANGUARD at eRisk 2026 Task 2: Contextualized Early Detection of Depression, Working Notes of the Conference and Labs of the Evaluation Forum (CLEF 2026), September 2026
Anamaria Vișan, Ciprian-Octavian Truică, Adrian Paschke, Elena-Simona Apostol. VANGUARD at eRisk 2026 Task 3: ADHD Symptom Sentence Ranking, Working Notes of the Conference and Labs of the Evaluation Forum (CLEF 2026), September 2026
Cosmin-Gabriel Becheru, Ciprian-Octavian Truică, Elena-Simona Apostol. Awakened at eRisk 2026: A Multi-Agent LLM Architecture for Automated BDI-II Depression Screening via Simulated Clinical Interviews, Working Notes of the Conference and Labs of the Evaluation Forum (CLEF 2026), September 2026
Cosmin-Gabriel Becheru, Ciprian-Octavian Truică, Elena-Simona Apostol. Awakened at eRisk 2026: Contextualized Early Detection of Depression using DeBERTa Embeddings and Temporal MLP Classification, Working Notes of the Conference and Labs of the Evaluation Forum (CLEF 2026), September 2026
Cosmin-Gabriel Becheru, Ciprian-Octavian Truică, Elena-Simona Apostol. Awakened at eRisk 2026: Sentence Embedding-Based ADHD Symptom Ranking from Reddit Posts, Working Notes of the Conference and Labs of the Evaluation Forum (CLEF 2026), September 2026
Bogdan Tamba, Alexandru Petrescu, Ciprian-Octavian Truică. SafetyNLP at EXIST 2026: Multimodal Sexism Detection in Memes using LLM-Driven OCR Enhancement and LeWiDi Target Modeling, Working Notes of the Conference and Labs of the Evaluation Forum (CLEF 2026), September 2026
Ana-Maria Luisa Mocanu, Sebastian Mocanu, Ciprian-Octavian Truică, Elena-Simona Apostol. Through the Eyes of the Beholder: Biometric and Demographic Conditioning for Multimodal Sexism Detection, Working Notes of the Conference and Labs of the Evaluation Forum (CLEF 2026), September 2026
Ana-Maria Luisa Mocanu, Ciprian-Octavian Truică, Elena-Simona Apostol. ACAT: A Collaborative Platform for Efficient Aspect-Based Sentiment Dataset Annotation, International Conference on Big Data Analytics and Knowledge Discovery (DAWAK), 38-43, August 2026
Noe Come Jacques Le Pollotec, Elena Simona Apostol, Ciprian Octavian Truică. NetGuardAI at MultiPRIDE: Multilingual Detection of Reclaimed Language. The 9th Evaluation Campaign of Natural Language Processing and Speech Tools for Italian, 2026. [link]
Ciprian-Octavian Truică, Elena-Simona Apostol, Alexandru-Gabriel Ilie, Adrian Paschke. HarmLLaMA: Harmful Language Detection with Large Language Models. International Conference on Intelligent Computer Communication and Processing (ICCP 2025), October 2025. DOI 10.1109/ICCP68926.2025.11427180 (Rank National: Romania Conference)
Daria-Elena Burghelea, Ciprian-Octavian Truică, Elena-Simona Apostol. VERIT-ALBERT: A Finetuned LLM Approach for Verifying Information Credibility. RoEduNet Conference Networking in Education and Research, September 2025. DOI: 10.1109/RoEduNet68395.2025.11208267
Alexandru Petrescu, Ciprian-Octavian Truică, Elena-Simona Apostol. Language-based Mixture of Transformers for Sexism Identification in Social Networks, Conference and Labs of the Evaluation Forum (CLEF 2025), p. 142-155, September 2025. DOI: 10.1007/978-3-032-04354-2_10
Alexandru Petrescu, Elena-Simona Apostol, Ciprian-Octavian Truică. Awakened at EXIST2025: Adaptive Mixture of Transformers, Working Notes of the Conference and Labs of the Evaluation Forum (CLEF 2025), p. 2112-2118, September 2025. [link]
Maria-Diana Cotelin, Ciprian-Octavian Truică, Elena-Simona Apostol. NetGuardAI at EXIST2025: Sexism Detection using mDeBERTa, Working Notes of the Conference. and Labs of the Evaluation Forum (CLEF 2025), p. 1897-1905, September 2025. [link]
ATESA-BÆRT: https://github.com/DS4AI-UPB/ATESA-BAERT
HarmLLaMA: https://github.com/DS4AI-UPB/HarmLLaMA
Relation Extraction: https://github.com/DS4AI-UPB/Awakened-CLEF2026-HIPE
Humor identification: https://github.com/DS4AI-UPB/VANGUARD-CLEF2026-JOKER
Sexism Identification: https://github.com/DS4AI-UPB/VANGUARD-CLEF2026-EXIST
Source Retrieval: https://github.com/DS4AI-UPB/Awakened-CLEF2026-CheckThat
Slur identification: https://github.com/DS4AI-UPB/NetGuardAI_MultiPride2026
Humor identification Qwen2.5-7B: https://huggingface.co/DS4AI-UPB/jokes-on-qwen2.5-7b
Humor identification Gemma4-31B: https://huggingface.co/DS4AI-UPB/jokes-on-gemma4-31b
Multimodal Sexism Identification: https://huggingface.co/DS4AI-UPB/vanguard-xlmr-clip-film-exist2026
Relation Extraction: https://huggingface.co/DS4AI-UPB/person-place-relation-xlmr