Thank you for your answer.
What I would like to do is importing end of date state data in dgraph. To avoid importing each day data without any changes, I would like to use some bi-temporal abstraction like ibm db2 with four timestamps for each node (business_time_start, business_time_end, sys_time_start, sys_time_end) - take a look here for the implementation in db2. The graph implementation would be something like that:
When I query the graph, then I could use a filter for example on business_time… to return the state of the data on some point in time.