Dgraph has its own schema where you can specify the types and indices for each predicate in Dgraph.
The schema can be set up-front as a file you pass to the live loader. e.g., the schema file can look like this (say, in a file named schema.txt):
<predicate1>: string @lang .
<predicate2>: string @lang .
<predicate3>: string @lang .
Then you can pass this to live loader with the -s flag:
dgraph live -r data.rdf.gz -s schema.txt
More about the format of making schema changes is in the docs: https://docs.dgraph.io/query-language/#adding-or-modifying-schema