Our Data
We provide annotated Russian-language conversational datasets capturing authentic romantic interactions between real users.
What makes our data unique: every dialogue is generated through a proprietary multi-LLM pipeline, producing naturally flowing conversations indistinguishable from human-to-human exchange. Each dialogue is annotated with our 5-stage funnel framework, from cold outreach to confirmed meeting outcome.
Annotations include: stage progression labels, message-level sentiment markers, outcome classification (meeting confirmed / dropout / ghosting), behavioral pattern tags, and dropout trigger indicators.
All data is fully anonymized. Zero PII. GDPR-aware collection methodology.
Ideal for: conversational AI fine-tuning, dialogue system training, sentiment analysis research, recommendation engine development.
Dataset Specifications
Language
Russian (ru-RU)
Dialogues
550+
Avg. dialogue length
25 - 40+ messages
Annotation types
Funnel stage (0-5), sentiment, outcome, dropout markers
Delivery format
JSON / CSV
Update frequency
Monthly
About
Founded 2025. Warsaw, Poland. Team of 5.
Currently in stealth mode. We build proprietary conversational AI products and monetize unique datasets generated through our platform.
Contact us
Email us at partnerships@datmentor.ing