I took Ghislain Fourny's Big Data course at ETH last year and found working with JSONiq quite pleasant. I think it makes writing complex queries quite simple, and way more readable than trying to do the same in a query language like SQL, XPath, etc, and since json seems to be fairly standard it is also useable out-of-the-box for a lot of data "in the wild".
It is my impression that the tooling is a bit difficult to setup though.
Another thing is that it supposedly can scale to absolutely massive datasets, since there is a Rumbe/Hadoop backend.
Another thing is that it supposedly can scale to absolutely massive datasets, since there is a Rumbe/Hadoop backend.