Leveraging Customer Reviews for Multi-Label Topic Classification in Tourism

Explore multi-label topic classification techniques for leveraging customer reviews data at Booking.com in this 38-minute conference talk by Moran Beladev. Learn how to identify popular activities per destination, detect key facilities for properties, and filter reviews by specific topics. Discover methods for detecting violence in reviews and summarizing the most discussed topics per property. Gain insights into fine-tuning BERT-like models on a small tourism dataset and compare their performance to other pre-trained models. Examine the data collection process using active learning and AWS Sagemaker Ground Truth, and witness a live demo showcasing model explainability with Streamlit. Perfect for those interested in machine learning, deep learning, NLP, data science, and AI applications in the tourism industry.

- Intro
- James Family Holiday Example
- Model
- Results
- Takeaways