Schedule

We've completed the first two bullet points of the project and have reached our goal of choosing a word-prediction model by this milestone. Slightly ahead of schedule, we have also begun to analyze prediction performance.

• Collect, parse, and split email bodies into $n$ -grams, and compute cooccurrence statistics for use in the prediction model.
• Create the Markov-based, word prediction model. [Completed at the milestone.]
• Analyze model performance and productivity increase in writing emails.
• Complete analyzing bigram model performance.
• Analyze trigram model performance.
• Time permitting, investigate performance improvements associated with context such as subject, recipient set, and next-letter typed.
• Time permitting, determine the reply-to address based on the message recipient set.