A Supervised Approach To The Interpretation Of Imperative To-Do Lists

Interesting machine analysis of "to do" lists. It surprises me that more research hasn't been done in this area, but understandably there is not a lot of data to pull out. To Do's are very short, so understanding intent could be hard.

To-do lists are a popular medium for personal information management. As to-do tasks are increasingly tracked in electronic form with mobile and desk- top organizers, so does the potential for software support for the corresponding tasks by means of intelligent agents. While there has been work in the area of personal assistants for to-do tasks, no work has focused on classifying user intention and information extraction as we do. We show that our methods per- form well across two corpora that span sub-domains, one of which we released.

This approach would also benefit from tuning over time. People tend to develop "shorthands" that they use in their To Do's and a system could learn those and tune to each person over time.