This may not be as straightforward as it seems. If the implementation is to parallel that for movies, then every episode an actor every appeared in would be downloaded and recorded as a "movie." That might be what you want—so the actor and the episodes are properly cross-linked. But this also means your database could to end up with hundreds of thousands of invisible episode records you're not interested in. This would slow things down, and probably increase the likelihood of problems with duplicate records and other such things.
So, given the current database structure, it might be preferable to just record the a list of the episode appearances in a separate field associated with the filmography, rather than recording them all relationally. On the other hand, if the implementation were deferred until the next major change in the database structure (I have no idea when that might be, if ever), then perhaps this could be handled much more effectively by introducing a separate table for episode details.
