Our goal is to provide a context sensitive preposition-checking utility to students who are learning English as a second language.
We want to identify incorrectly used prepositions, and suggest an appropriate correction.
For example,
The dog walked throughout the door
should be identified, and the correct preposition
through should be proposed.
Because this problem is completely contextual, we approach it as a
context-sensitive spelling correction problem. A machine learning program based on the SNoW learning architecture, reads text, learns the appropriate context for each preposition, and is then able to make suggestions on future sentences.
In this demo, this capability is implemented as a macro for Microsoft Word, which communicates with a server running SNoW, the machine learning program which is the heart of this utility.
