Reading about data stream clustering I met the next terms:
- landmark window model,
- sliding window model,
- damped window.
As to sliding window it's clear - oldest data escape the scope, the new data go inside. But what are the consepts of the other two?
I can suppose that damped window it's like a buffer, that is flushed after filling but some explanation, that I had found, states that dumped uses decreasing weights of data as function of time.