Detecting Email Components with Constraints: Expressive and Extensible Models in Answer Set Programming

Gromit Yeuk-Yin Chan, Shunan Guo, Caroline Kim, Cole Connelly, Michelle Lee, Andrew Thomson, Eunyee Koh

Abstract: Detecting structures and components in business emails is vital for editor software to convert third-party emails so that designers can edit them without needing to know how HTML works. In a production environment, the challenge is to make the model easy to be understood and maintained by different stakeholders. We propose detecting email components with a collection of constraints written in Answer Set Programming (ASP). Hard constraints can detect well-defined components like email layouts, and soft constraints can incorporate ML to detect custom components like buttons and titles in emails. Using constraints, developers can apply their domain knowledge to the model and express them in a concrete, extensible, and deterministic form. We demonstrate the effectiveness with a prototype and evaluations from real datasets.