As the Information Retrieval (IR) field increasingly recognizes the importance of inclusivity, addressing the needs of low-resource languages remains a significant challenge. Transliteration between Urdu and its Romanized form, Roman Urdu, remains underexplored despite the widespread use of both scripts in South Asia. Prior work using RNNs on the Roman-Urdu-Parl dataset showed promising results but suffered from poor domain adaptability and limited evaluation. We propose a transformer-based approach using the m2m100 multilingual translation model, enhanced with masked language modeling (MLM) pretraining and fine-tuning on both Roman-Urdu-Parl and the domain diverse Dakshina dataset. To address previous evaluation flaws, we introduce rigorous dataset splits and assess performance using BLEU, character-level BLEU, and CHRF. Our model achieves strong transliteration performance, with Char-BLEU scores of 96.37 for Urdu→Roman-Urdu and 97.44 for Roman-Urdu→Urdu. These results outperform both RNN baselines and GPT-4o Mini and demonstrate the effectiveness of multilingual transfer learning for low-resource transliteration tasks.

Effective learning in digital environments depends on quick access to educational resources and timely support. We present AIDEN, an advanced, AI-driven virtual teaching assistant integrated into lectures, to provide meaningful support for students. AIDEN’s capabilities include reading lecture materials aloud, locating specific slides, automatic speaker notes generation, search through a video stream. Powered by state-of-the-art retrieval and text generation, AIDEN can be adapted to new lecture content with minimal manual adjustments, requiring only minor customization of data handling processes and model configurations. Through automated testing, we evaluated AIDEN’s performance across key metrics slide retrieval recall for questions, and alignment of generated speaker notes with ground-truth data. The evaluation underscores AIDEN’s potential to significantly enhance learning experiences by offering real-world application and rapid configurability to diverse learning materials.