As a user I want to process Wikidata items about people in order to get relevant insights without having to download a full dump. Compared to other sets of instances, the set of instances of Q5 is quite homogeneous and easily distinguishable from other sets, since people are very similar to each other (well-known range of lifespan, similar skills, etc.) and we have much interesting knowledge about them.
Problem: Full Wikidata dumps are too big to be processed, sometimes even too big to be downloaded. When we just need to study people, that is, instances of Q5, having a specific dump with only people (so much smaller) would be valuable. Books, scholarly publications and other heavy sets of items would be excluded from this dump.