Home / Services / Data labelling
Text and NLP labelling
Entities, sentiment, topics and intent labelled in English, Hindi and Hinglish text, so your model understands what people mean.
Priya Sharma booked a flight from Delhi to Bengaluru on 12 March.
The ticket cost ₹6,450 through IndiGo.
AnnotatorSpans tagged
ReviewerChecked
QAApproved
What you get
- Named entity recognition, standard and custom
- Sentiment, topic and intent labels
- Multi-label classification
- Span-level and document-level labels
- English, Hindi and Hinglish
How we check quality
- Span boundaries checked word by word
- Agreement measured on a shared sample
- Disagreements discussed and logged
- Label balance reported per batch
How it works
A clear path from first conversation to delivered data.
- GuidelinesWe build an example sheet for every label.
- PilotA small set is labelled free for your review.
- LabelAnnotators label with tricky cases logged.
- Review and deliverAgreement measured, then delivered in your format.
Common questions
Can you handle Hinglish?
Yes. Mixed Hindi and English text is common in Indian data and we label it with clear rules for both scripts.
Which formats do you deliver?
JSONL spans, CoNLL BIO, spaCy JSON, CSV or Excel.
Can you create custom entity types?
Yes. We define them with you and add examples to the guide.
Related
Start with a free pilot
Send a small sample and your guidelines. No cost, no commitment.