Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
August 1, 2025

Bidirectional Transformer-Based Neural Machine Translation for Amharic and Tigrinya: Bridging Morphological Complexity and Data Scarcity

View Full Paper
Ask AI
Bookmark
Share

Authors

MAMatiyas Gutema AngechaAddis Ababa UniversityMTMartha Yifiru TachbelieAddis Ababa University

Discussion

Loading...

Member takes

Implication

Bidirectional transformer enhances translation quality for Amharic and Tigrinya, indicating effective solutions for data scarcity.

Key Points

  • MAIN FINDING: A bidirectional neural machine translation system effectively addresses the complexities of Amharic and Tigrinya.
  • KEY EVIDENCE: The final system achieved BLEU scores of 44.32% for Amharic to Tigrinya and 44.10% for Tigrinya to Amharic.
  • APPROACH: The study used transformer architecture with data augmentation techniques like back translation and subword segmentation.
  • SIGNIFICANCE: These findings provide a pathway for improving machine translation for low-resource languages, enhancing communication capabilities.

Cite This Study

Angecha et al. (2025) studied this question.

synapsesocial.com/papers/689a0c6be6551bb0af8cff3fhttps://doi.org/10.21203/rs.3.rs-7040037/v1
View Full Paper
Ask AI
Bookmark
Share