Mirella Lapata
Mirella Lapata, Ratish Puduppully and Li Dong proposed a neural architecture for data-to-text generation. Their paper separates content selection and ordering from document generation, explicitly planning which information to express before producing text. Lapata and Yang Liu’s Text Summarization with Pretrained Encoders develops a BERT-based framework for extractive and abstractive summaries. The paper studies document representations and training schedules for encoder-decoder models, with evaluations on three datasets. At Edinburgh, Lapata works on compositional generalization, cross-lingual transfer and long-context understanding. Her research interests also include verifiable responses with supporting evidence and agent-based collaborative writing, extending her work on language generation.
- Now at
- University of Edinburgh
- Works
- 373
- Citations
- 27,320
- h-index
- 80
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
2018 preprint; revised 2019
Planning what to say
Mirella Lapata, Ratish Puduppully and Li Dong proposed a neural architecture for data-to-text generation. Their paper separates content selection and ordering from document generation, explicitly planning which information to express before producing text.
-
2019
Pretrained summarization models
Lapata and Yang Liu’s Text Summarization with Pretrained Encoders develops a BERT-based framework for extractive and abstractive summaries. The paper studies document representations and training schedules for encoder-decoder models, with evaluations on three datasets.
-
As of October 2026
Generalization and evidence
At Edinburgh, Lapata works on compositional generalization, cross-lingual transfer and long-context understanding. Her research interests also include verifiable responses with supporting evidence and agent-based collaborative writing, extending her work on language generation.
Sources and credits
Photo: ELLIS profile; creator not stated. License: Not stated by source. Source: ellis.eu. Cropped to a square and resized. Ask us to remove or replace this photo.