Scholarlink

With the increasing availability of AI services and language models, many language-related tasks have been automated. However, most advancements remain focused primarily on high-resource languages such as English, leaving low-resource languages significantly underexplored. This study addresses that gap by designing and deploying a multilingual, web-based platform that provides 11 essential AI services: Text-to-Speech, Speech-to-Text, Question Answering, Automatic Summarization, Machine Translation, Named Entity Recognition, Optical Character Recognition, Image Captioning, a Large Language Model, KazCLIP, and VideoCLIP. The project included extensive research into existing datasets and state-of-the-art methods, and it optimized each model within computational and data-availability constraints. User experience evaluations indicated high satisfaction with the platform, achieving an average user rating of 9.40 out of 10. The platform demonstration is at https://zhadiger.vercel.app/ , and code is at https://github.com/fano2458/Zhadiger-Kazakh-Language-AI .

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

ZhadigerAI: Software-as-a-Service AI Platform for Kazakh and English

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Aidana Baglanova,
  • Nail Fakhrutdinov,
  • Umit Azirakhmet,
  • Ruslan Kalimzhanov,
  • Zilola Babakhojayeva,
  • Hakan Yekta Yatbaz,
  • Adnan Yazici

摘要

With the increasing availability of AI services and language models, many language-related tasks have been automated. However, most advancements remain focused primarily on high-resource languages such as English, leaving low-resource languages significantly underexplored. This study addresses that gap by designing and deploying a multilingual, web-based platform that provides 11 essential AI services: Text-to-Speech, Speech-to-Text, Question Answering, Automatic Summarization, Machine Translation, Named Entity Recognition, Optical Character Recognition, Image Captioning, a Large Language Model, KazCLIP, and VideoCLIP. The project included extensive research into existing datasets and state-of-the-art methods, and it optimized each model within computational and data-availability constraints. User experience evaluations indicated high satisfaction with the platform, achieving an average user rating of 9.40 out of 10. The platform demonstration is at https://zhadiger.vercel.app/ , and code is at https://github.com/fano2458/Zhadiger-Kazakh-Language-AI .