Predictive Text Entry Using Commonsense Reasoning

Tom Stocky, Alex Faaborg, Henry Lieberman

Predictive Text Entry Using Commonsense Reasoning
Text messaging on cell phones is often slow and difficult, yet it has emerged as an increasingly popular medium of communication. This project will study how Commonsense Reasoning algorithms can be used to enhance word completion on mobile devices. We have created a prototype application that uses ConceptNet to understand the context of a user’s text message, and uses this context to drive intelligent word completion. For instance, if the user types “let’s have a picnic in the pa…” the software suggests park, as picnic and park are related to each other in ConceptNet. Similarly the phrase, “I have arrived at the train st..” will result in a suggestion of station.

back to Software_Agents