In such case she will start a dictionary App.
Similarly, a user shopping online will probably start a payment App after hunting a desired commodity.
The last used App directly affects the next using App.
Therefore, we use a Markov chain to describe such memoryless situation in which the current App is only affected by the last used App but not all Apps in the history.
During the data collection process, we record the package name to identify different Apps.
Note that current mobile Apps are with frequent updates.
Only recording the package name may count the same App (both old version and new version) as different ones.