After reading this DBPedia discussion on storing empty nodes , the goal seems to be to avoid a collision with the WikiPedia URI.
This is apparently used for sites that do not have a corresponding WikiPedia article, but rather point to a related article on this subject. Because DBPedia is trying to create a URI for everything, this URI is built according to certain rules (more on this in the discussion above).
From the discussion:
Note that intermediate node URIs always contain double underscores, for example 1 . Wikipedia does not allow consecutive underlines on a page, so we can be sure that these URIs will not collide with the DBpedia URI for Wikipedia pages. We select the name from the arguments of the template from which the intermediate node is extracted, add two underscores, this is the name and number in the URI of the main page and use this as the URI for the intermediate node. If there are several intermediate nodes on the same page for which we select the same name, we use different numbers, for example. see 1 and [2].
source share