[Paper Review] Reciptor: An Effective Pretrained Model for Recipe Representation Learning

Title: Reciptor: An Effective Pretrained Model for Recipe Representation Learning

Authors: Diya Li, Mohammed J. Zaki

---

Summary

• a joint approach for learning effective pretrained recipe embeddings using both the ingredients and cooking instructions
• a novel set transformer-based joint model to learn recipe representations that preserve permutation-invariance

 

Framework 

1. Tags data: Model Validation에 사용하는 데이터, Recipe의 Label, Node Classification으로 검증 시 활용
2. Ingredients data: Set Transformer로 embedding, input의 순서에 상관없이 학습시키기 위함
3. Instructions: LSTM으로 embedding, instruction의 sequence를 학습시키기 위함

---

Paper URL: https://dl.acm.org/doi/10.1145/3394486.3403223

RECIPTOR: An Effective Pretrained Model for Recipe Representation Learning | Proceedings of the 26th ACM SIGKDD International Co