Hospitality Domain Embedding Model
This model is a domain-specific embedding model trained on real hotel operational text, including:
- Guest complaints
- Housekeeping procedures
- Maintenance logs
- SOPs and service standards
- Training references
- Operational guidelines
The goal is to provide high-quality semantic embeddings for hospitality RAG systems, search, clustering, and similarity matching.
🔧 Base Model
This model is fine-tuned from:
sentence-transformers/all-MiniLM-L6-v2
Chosen for:
- Fast CPU inference
- Small size (HF Free Tier compatible)
- Strong semantic performance
📚 Training Data
The model was trained on two combined datasets:
Dataset 1: Hotel Complaints
- Column:
complaint
- Contains maintenance and service-related guest issues
Dataset 2: Hospitality Operational Text
- Column:
text
- Includes SOPs, housekeeping standards, maintenance procedures, and training content
All text was cleaned, normalized, and deduplicated before training.
🎯 Intended Use
This embedding model is designed for:
- RAG (Retrieval-Augmented Generation)
- Semantic search
- Similarity matching
- Clustering of complaints or reviews
- Maintenance triage systems
- Housekeeping and SOP retrieval
- Hotel operations AI agents
🚫 Not Intended For
- General-purpose text generation
- Sentiment classification (use a classifier instead)
- Legal or medical decision-making
🧪 Evaluation
The model was evaluated on:
- Retrieval quality (Recall@k)
- Semantic similarity
- Domain relevance
📦 Deployment
This model is optimized for:
- Hugging Face Spaces (CPU)
- ChromaDB vector stores
- Llama-based RAG pipelines
🙌 Author
Created by Amey Tillu
Hospitality & Tourism Data Analyst & AI/ML Hobbyist
