A Unified Framework with Novel Metrics for Evaluating the Effectiveness of XAI Techniques in LLMs
Proposes a unified evaluation framework and novel metrics for assessing explainable AI techniques in large language models.
View PublicationPUBLICATIONS
Journal articles, conference papers, and preprints in NLP, machine learning, computer vision and low-resource language technologies.
Explainable AI: XAI-Guided Context-Aware Data Augmentation
Introduces an XAI-guided augmentation framework that preserves important task-relevant features while improving performance on low-resource NLP tasks.
View PublicationEvaluating the Effectiveness of XAI Techniques for Encoder-Based Language Models
Evaluates XAI methods for encoder-based language models using metrics such as human-reasoning agreement, robustness, consistency, and contrastivity.
View PublicationBilingual Word Level Language Identification for Omotic Languages
Presents a bilingual word-level language identification approach for Omotic languages, contributing to low-resource language processing.
View PublicationOdio-BERT: Evaluating Domain Task Impact in Hate Speech Detection
Introduces Odio-BERT, a domain-adapted BERT model for hate speech detection, showing the benefits of domain-specific modeling.
View PublicationEthioLLM: Multilingual Large Language Models for Ethiopian Languages with Task Evaluation
Introduces EthioLLM and benchmark-based evaluation for Ethiopian languages, expanding multilingual LLM research for low-resource settings.
View PublicationEthio-Fake: Cutting-Edge Approaches to Combat Fake News in Under-Resourced Languages Using Explainable AI
Proposes explainable AI methods for fake news detection in under-resourced languages, integrating content and social context features.
View PublicationSemantic-Driven Topic Modeling Using Transformer-Based Embeddings and Clustering Algorithms
Presents a semantic-driven topic modeling approach that combines transformer embeddings with clustering to improve topic coherence.
View PublicationExtractive Text Summarization for Wolaytta Language Using Recurrent Neural Network
Develops an extractive summarization approach for Wolaytta using recurrent neural networks, supporting NLP for Ethiopian languages.
View PublicationDesign and Develop Automatic Parts of Speech Tagger for Siltigna Language
Presents an automatic part-of-speech tagging system for Siltigna, contributing foundational NLP resources for a low-resource Ethiopian language.
View PublicationMasakhaNEWS: News Topic Classification for African Languages
Introduces MasakhaNEWS, a multilingual news topic classification benchmark for African languages.
View PublicationMultilingual Hope Speech Detection using Machine Learning
Applies machine learning techniques to multilingual hope speech detection, focusing on English and Spanish social media data.
View PublicationTransformer-Based Hate Speech Detection for Multi-Class and Multi-Label Classification
Investigates transformer-based hate speech detection for both multi-class and multi-label classification settings.
View PublicationHabesha@DravidianLangTech: Utilizing Deep and Transfer Learning Approaches for Sentiment Analysis
Explores deep learning and transfer learning methods for sentiment analysis in Dravidian language shared-task settings.
View PublicationHabesha@DravidianLangTech: Abusive Comment Detection using Deep Learning Approach
Presents a deep learning system for abusive comment detection in shared-task settings for Dravidian languages.
View PublicationEvaluating the Effectiveness of Hybrid Features in Fake News Detection on Social Media
Examines hybrid features for fake news detection on social media, combining multiple information sources to improve classification.
View PublicationWord Level Language Identification in Code-mixed Kannada-English Texts Using Traditional Machine Learning Algorithms
Studies traditional machine learning approaches for word-level language identification in code-mixed Kannada-English text.
View PublicationWord Level Language Identification in Code-mixed Kannada-English Texts using Deep Learning Approach
Proposes a deep learning approach for word-level language identification in code-mixed Kannada-English texts.
View PublicationEarly Ginger Disease Detection Using Deep Learning Approach
Applies deep learning to early ginger disease detection, demonstrating practical computer vision methods for agriculture.
View PublicationA Parallel Corpora for bi-directional Neural Machine Translation for Low Resourced Ethiopian Languages
Introduces a parallel corpus for bidirectional neural machine translation involving low-resourced Ethiopian languages.
View Publication