Leyu

Leyu is an open-source crowdsourcing platform designed to advance AI development by providing open datasets for low-resource African languages. The platform addresses the need for high-quality, representative language data by enabling the collection of real-world text and speech datasets, supporting linguistic diversity and inclusive AI. Its offerings are positioned for both commercial and research use, with a focus on transparency and ethical data practices.

The platform features scalable workflows that manage contributor and reviewer pipelines, allowing for efficient data collection and quality assurance. Leyu supports versatile input types, enabling the seamless capture of both text and speech data in real time. To ensure the integrity of its datasets, the platform incorporates built-in task review and approval systems, facilitating robust quality control throughout the data collection process. Each dataset is accompanied by comprehensive documentation, including a Data Dictionary and README, to aid users in understanding and utilizing the data effectively.

Leyu's approach to data collection is community-driven, leveraging crowdsourcing from local annotators to ensure cultural accuracy and diversity in the datasets. The platform employs a hybrid labeling strategy, combining human and automated methods to enhance the accuracy of its data. Ethical considerations are emphasized, with attention to data privacy and transparent practices in dataset creation.

The service is accessible through an open-source mobile application, which is the same tool used for professional-grade data collection by the platform itself. Leyu's open offerings are freely available, providing immediate access to datasets and tools for researchers and developers. This open access model is intended to support the development of ethical and representative AI systems, particularly for underserved communities and low-resource languages.