Analysis
Bandai Namco is exploring innovative ways to enhance anime recommendations using contrastive learning! This research showcases the potential of using embeddings generated from anime summaries and series relationships to provide more relevant suggestions to users. The project highlights a proactive approach to solving the cold-start problem in recommendation systems.
Key Takeaways
- •Contrastive learning with a Japanese BERT model was used for fine-tuning.
- •The project aimed to solve the cold-start problem in their recommendation system.
- •Embeddings were created from anime synopsis data to measure similarity.
Reference / Citation
View Original"This Fine-tuning model was used to create embedding expressions from the synopsis data of each anime."