This dataset contains multi-turn customer service dialogues in Hausa, Yoruba, and Swahili. It is designed for instruction-tuning multilingual customer support language models using conversations that involve multiple exchanges.
Potential Use Cases
- Multilingual Customer Support Fine-Tuning: Multi-turn customer service dialogues in Hausa, Yoruba, and Swahili can support fine-tuning language models to understand customer inquiries, maintain conversational context across multiple turns, and generate relevant responses. This can support the development of conversational customer support applications for speakers of these three languages.
- Multilingual Customer Intent Classification: With additional intent labels, these customer service dialogues can be used to train multilingual models to classify customer requests and inquiries. They can also support analysis of shared intent categories and language-specific expressions, as well as research into cross-lingual transfer among Hausa, Yoruba, and Swahili.