A5068105185- Misinformation and Its Impacts
- Topic Modeling
- Hate Speech and Cyberbullying Detection
- Natural Language Processing Techniques
- Spam and Phishing Detection
- Software Engineering Research
- Text Readability and Simplification
- Video Surveillance and Tracking Methods
- Advanced Image and Video Retrieval Techniques
- Data-Driven Disease Surveillance
- Authorship Attribution and Profiling
- Cybercrime and Law Enforcement Studies
- Adversarial Robustness in Machine Learning
- Vaccine Coverage and Hesitancy
- Web Data Mining and Analysis
- Wikis in Education and Collaboration
- Anomaly Detection Techniques and Applications
- Vehicle License Plate Recognition
- Personal Information Management and User Behavior
- Biomedical Text Mining and Ontologies
- Robotics and Sensor-Based Localization
- Artificial Intelligence in Law
- Complex Network Analysis Techniques
- Advanced Vision and Imaging
- Information and Cyber Security
Universitat Politècnica de València
2021-2023
Deutsche Welle
2022
University of Koblenz and Landau
2019-2021
Universität Koblenz
2019-2020
Koblenz University of Applied Sciences
2020
Aalto University
2020
Institute of Informatics and Telematics
2020
Izmir Institute of Technology
2016-2018
Equitable access to reliable health information is vital for public health, but the quality of online resources varies by language, raising concerns about inconsistencies in Large Language Models (LLMs) healthcare. In this study, we examine consistency responses provided LLMs health-related questions across English, German, Turkish, and Chinese. We largely expand HealthFC dataset categorizing disease type broadening its multilingual scope with Turkish Chinese translations. reveal significant...
This paper describes our submission to SemEval-2019 Task 7: RumourEval: Determining Rumor Veracity and Support for Rumors. We participated in both subtasks. The goal of subtask A is classify the type interaction between a rumorous social media post reply as support, query, deny, or comment. B predict veracity given rumor. For A, we implement CNN-based neural architecture using ELMo embeddings text combined with auxiliary features achieve F1-score 44.6%. B, employ MLP network leveraging...
The popularity of social media has created problems such as hate speech and sexism. identification classification sexism in are very relevant tasks, they would allow building a healthier environment. Nevertheless, these tasks considerably challenging. This work proposes system to use multilingual monolingual BERT data points translation ensemble strategies for English Spanish. It was conducted the context sEXism Identification Social neTworks shared 2021 (EXIST 2021) task, proposed by...
In a hybrid camera system combining an omnidirectional and Pan-Tilt-Zoom (PTZ) camera, the provides 360 degree horizontal field-of-view, whereas PTZ high resolution at certain direction. This results in wide field-of-view system. this paper, we exploit for real-time object classification tracking traffic scenes. The detects moving objects performs initial using shape-based features. Concurrently, classifies frames Histogram of Oriented Gradients (HOG) also high-resolution classified as...
Enabled by recent improvements in generation methodologies, DeepFakes have become mainstream due to their increasingly better visual quality, the increase easy-to-use tools and rapid dissemination through social media. This fact poses a severe threat our societies with potential erode cohesion influence democracies. To mitigate threat, numerous DeepFake detection schemes been introduced literature but very few provide web service that can be used wild. In this paper, we introduce MeVer...
This paper addresses debiasing in news editing and evaluates the effectiveness of conversational Large Language Models this task. We designed an evaluation checklist tailored to editors' perspectives, obtained generated texts from three popular models using a subset publicly available dataset media bias, evaluated according checklist. Furthermore, we examined as evaluator for checking quality debiased model outputs. Our findings indicate that none LLMs are perfect debiasing. Notably, some...
In the current era of social media and generative AI, an ability to automatically assess credibility online content is tremendous importance. Credibility assessment fundamentally based on aggregating signals, which refer small units information, such as factuality, bias, or a presence persuasion techniques, into overall score. signals provide more granular, easily explainable widely utilizable information in contrast currently predominant fake news detection, utilizes various (mostly latent)...
Combating misinformation is a challenging task due to the fact that evolves in content and strategy.We describe challenges of this propose git-based framework for collaborative open policymaking against ever-evolving misinformation.We present setup future test-runs where users receive tasks conduct core functions policy-making misinformation.
We propose a method for vehicle detection and classification in traffic scenes using an omnidirectional PTZ (pan-tilt-zoom) camera. The proposed controls the camera with respect to location of object detected by background subtraction camera, classifies frames also compare accuracy when is performed only types we worked on are motorcycle, car, van pedestrian.
This paper presents a unified user profiling framework to identify hate speech spreaders by processing their tweets regardless of the language. The encodes with sentence transformers and applies an attention mechanism select important for learning profiles. Furthermore, layer helps explain why is spreader producing weights at both token post level. Our proposed model outperformed state-of-the-art multilingual transformer models.
Health misinformation on search engines is a significant problem that could negatively affect individuals or public health. To mitigate the problem, TREC organizes health track. This paper presents our submissions to this We use BM25 and domain-specific semantic engine for retrieving initial documents. Later, we examine news schema quality assessment apply it re-rank merge scores from different components by using reciprocal rank fusion. Finally, discuss results conclude with future works.
Identifying check-worthy claims is often the first step of automated fact-checking systems. Tackling this task in a multilingual setting has been understudied. Encoding inputs with text representations could be one approach to solve check-worthiness detection. However, suffer if cultural bias exists within communities on determining what check-worthy.In paper, we propose language identification as an auxiliary mitigate unintended bias.With purpose, experiment joint training by using datasets...
Widespread and rapid dissemination of false news has made fact-checking an indispensable requirement. Given its time-consuming labor-intensive nature, the task calls for automated support to meet demand. In this paper, we propose leverage commonsense knowledge tasks classification check-worthy claim detection. Arguing that is a factor in human believability, fine-tune BERT language model with question answering aforementioned multi-task learning environment. For predicting fine-grained...
Check-worthiness detection is the task of identifying claims, worthy to be investigated by fact-checkers. Resource scarcity for non-world languages and model learning costs remain major challenges creation models supporting multilingual check-worthiness detection. This paper proposes cross-training adapters on a subset world languages, combined adapter fusion, detect claims emerging globally in multiple languages. (1) With vast number annotators available storage-efficient models, this...
This paper describes our submission for the subjectivity detection task at CheckThat! Lab. To tackle class imbalances in task, we have generated additional training materials with GPT-3 models using prompts of different styles from a checklist based on journalistic perspective. We used extended set to fine-tune language-specific transformer models. Our experiments English, German and Turkish demonstrate that subjective are effective across all languages. In addition, observe style-based...