The first one is just a combination of some computer modeling guessing at possibly useful existing drugs and a look at the existing patient data showing that people who were on these seemed to have Alzheimer's less. It's very clear that it's just an initial step that doesn't actually show there's a benefit.
The second is more of that same basic kind of research, but weighing in the opposite direction: it showed no association with Alzheimer's for people with a specific other condition who got these drugs vs. ones with the same condition that didn't (also they did some cell culture testing outside the body that didn't show any effect/improvement).
This new one is an actual double-blind study giving people the drug. Equally important, it's not claiming an effect on Alzheimer's, but vascular dementia, a different form of dementia that's directly related to blood flow and therefore to Viagra's known effects. Of course it's still just really a marker for future research on if this is broadly useful.