Back to overview

Text-to-text Generation Methods

Type of publication Not peer-reviewed
Publikationsform Other publication (non peer-review)
Project An Integer Linear Programming Approach to Text-to-Text Generation
Show all

Other publication (non peer-review)

Book Text-to-text Generation Methods


Text-to-text generation has recently emerged as a research area in computational linguistics, which focuses on the re-use of existing texts in order to robustly generate a new text, by fusing relevant phrases and discarding irrelevant ones, rather than generating a new text from scratch. In this paper, I review existing approaches to text-to-text generation, discussing their specific aims as well as the underlying methodology. I focus on approaches based on Integer Linear Programming, an optimisation technique, and on the specific task of text simplification, which has a high practical impact on a broad category of users, from children to language- impaired people, second-language learners, poor-literacy readers, and which also benefits other language applications.