Can anyone explain delayed-choice entanglement swapping

Here is the article.

As they explain it, you create 2 pairs of entangled photons. Grab one from the first pair and send it to Victor. Take the other one and send it to Alice.

Now grab one from the second pair and also send it to Victor. Grab the other one and send it to Bob.

It seems that depending upon how Victor measures his photons determines whether or not the ones that Bob and Alice get are entangled. Moreover, this is the case even if Bob and Alice conduct their measurements first.

The part that’s not clear is how does Victor do his measurements to create or not create an entangled state?

I found this paper which seems to indicate that the difference is that in once case Victor make Bell-type measurements and otherwise makes not Bell-type measurements.

If this is correct, could anyone elaborate?

Does this mean that to create entanglement, Victor would make orthogonal measurements and otherwise parallel measurements?