The team began by researching and collecting a large dataset of Pitman shorthand strokes and their corresponding text translations. They also studied existing shorthand translation systems and analyzed the challenges of developing an accurate and efficient translation app.