Memory in the brain isn't super-well understood, so going to the level of "data-structures" isn't really possible with purely biological models. Not that a purely biolgocial description would be very useful anyways. When people ask how the brain works, they typically don't want to be told "molecule A interacts with molecule B, which triggers molecule C".
Since you're coming from computer science, I am going to assume that you're probably more interested in models or the "data-structures" (how to structure data, like a list or a tree) we assign to a brain. One way to build biologically plausible brain models is with the Neurological Engineering Framework (NEF) and there's been a couple data structures based on this that have been used. You can see more detail on this in my other answer "How is memory accounted for in the NEF?"
However, it's important to note that the NEF is a framework, thus much work is being done on implementing more complex data structures than the ones mentioned in that question. In particular, Aaron Voekler has implemented a dynamic hetero-associative memory using the framework, which functions somewhat in the same manner as a hash table.
I hope this answer gives you a good starting point for further investigation.