Multilingual Student Queries for Educational Chatbot, 10,000 Samples

A dataset of 10,000 query-intent pairs for training a multilingual educational chatbot. The data was constructed by Divya Meena through primary collection from middle and secondary school students in the Vellore and Gudiyatham districts of Tamil Nadu, India. The primary dataset of 1,200 pairs was expanded using augmentation techniques.

Use Cases

• Train intent classification models for educational chatbots based on the described query-intent pairs.
• Study student communication patterns and question phrasing in a rural Indian context.
• Benchmark data augmentation techniques for expanding small, manually annotated text datasets.
• Develop multilingual NLP models for educational support, as the description mentions a multilingual system.

Strengths

• Contains 10,000 samples, expanded from a primary manually constructed set of 1,200 query-intent pairs.
• Data was collected directly from the target demographic (students in government schools) to ensure alignment with the deployment environment.
• Collection used multiple mediums (voice recordings and structured questionnaires) to capture genuine communication patterns.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the primary data is known (1,200), but the final augmented sample structure is not detailed.
• Data may reflect geographic and demographic bias inherent to its collection in specific districts of Tamil Nadu, India.

Provenance

Source

Primary data collection by author Divya Meena.

Collection Method

Data gathered through voice recordings and structured questionnaires from students, followed by transcription and annotation into intent categories.

Freshness

Last updated 2026-04-26 12:00:35; freshness should be verified.

Geography

Vellore and Gudiyatham districts, Tamil Nadu, India.