Rules or Frequent Items as Input?

Does CORELS consume rules as input or frequent items?

The paper talks about "rules", but then only mentions a support threshold (λ) and not a confidence. Also the sample input data as well as the Python code you shared, mine.py, seems to be doing frequent itemset mining rather than rule mining.

I am just trying to figure out whether I should give antecedents of rules as inputs or frequent itemsets themselves as input. Thanks!