Data Modeling - Parent Child

I am curious on how I should approach an issue I am facing. I have a data source that collects data at two different granularities. One is a "parent" record that highlights total budget. In the same table I have a column that you can enter if it has a parent budget. So, in one source I have data for "child" records that allows users to enter in if it is related to a "parent" budget. My challenege is do I leave in the parent records and when building measures include filter context to exclude the parent record? or do I create a sepearte table for parent records? and force a physical relationship of 1:*? 

With that how do I overcome ambiguity? Here is what I mean. I have a table that captures all comments and they're written to a different dat source. If I create a seperate table for parent and one for child it would create ambiguity between all three tables and I am unsure of how I can cirumvent that. I don't think it would be good to bloat the model by creating another table to filter to only parent comments and one for child comments. I hope that makes sense.