Week 15: Summarization, translation, and recommenders

DATA 441/641 · Week 15

Notes

• 15.1 Text Summarization(slides)
• 15.2 Machine Translation(slides)
• 15.3 Recommender Systems(slides)

Additional

• Seq2seq models and loss(PDF)
• Evaluating sequence models(PDF)
• Forecasting with sequential models(PDF)
• Recommender notes(PDF)
• Self-supervised learning(PDF)
• One-shot learning(PDF)

Additional readings

• Practical NLP, Ch. 8–10 (summarization, MT, recommenders)

Discussion

• When is extractive summarization safer than an abstractive rewrite?
• What signal would you use besides text to rank items for a user?