Bnar Ismail

Student Information

• Name: Bnar Ismail
• Level: MSc
• University: University of Kurdistan Hewlêr
• Subject: Computer Science
• Research Area: Tokenization-free Kurdish Machine Translation
• Status: Completed
• Supervisor: Dr. Polla Fattah
• Duration: 2023 - 2024

Research Focus

Bnar’s research focused on developing tokenization-free machine translation systems for the Kurdish language using ByT5 (Byte-to-Byte) architectures. The work addressed the unique challenges of Kurdish language processing through innovative approaches.

Key Research Areas

Tokenization-free Translation

• Implemented ByT5-based translation models
• Developed specialized approaches for Kurdish text processing
• Created innovative solutions for Kurdish language challenges

Kurdish Language Processing

• Developed specialized approaches for Kurdish text
• Implemented cross-lingual training techniques
• Created comprehensive evaluation frameworks

Model Optimization

• Optimized translation quality and performance
• Implemented advanced evaluation metrics
• Developed robust translation systems

Research Outcomes

• Successfully implemented tokenization-free Kurdish translation
• Completed comprehensive evaluation of ByT5 models for Kurdish
• Published research on Kurdish machine translation methodologies
• Contributed to understanding of tokenization-free approaches

Technologies and Methodologies

• ByT5 (Byte-to-Byte) architectures
• Neural machine translation
• Tokenization-free approaches
• Translation evaluation metrics
• Cross-lingual training techniques

Impact

Bnar’s work has advanced Kurdish machine translation technology by demonstrating the effectiveness of tokenization-free approaches, making Kurdish translation more accessible and improving the quality of Kurdish language technology.

Publications and Presentations

• Tokenisation-Free Machine Translation for Central Kurdish using the ByT5 Model (2024) - Conference Paper
• Domain Adaptation in Low-Resource Kurdish–English Machine Translation (2024) - Journal Article
• Evaluation Metrics in Low-Resource MT: Insights from Using Automatic and Human Evaluation in Kurdish (2024) - Conference Paper
• Completed comprehensive thesis on Kurdish machine translation
• Published research findings on tokenization-free approaches
• Contributed to Kurdish language technology advancement

Academic Achievement

• Successfully completed MSc degree
• Developed innovative approaches to Kurdish language processing
• Contributed significantly to Kurdish language technology
• Thesis focused on practical applications of machine translation