Understand a dictionary as an abstract data type of key-value pairs, its operations, how it is typically implemented using a hash table, and when a dictionary is appropriate.

What this dot point is asking

AQA wants you to describe a dictionary as an abstract data type that stores key-value pairs, list its operations, explain that it is usually implemented with a hash table, and say when a dictionary is the right choice.

The dictionary abstract data type

Typical operations are: add or update a (key, value) pair, look up the value for a key, delete a pair, and test whether a key is present. Because a dictionary is an abstract data type, the specification describes only this behaviour, not the internal storage; the same operations could in principle be backed by different structures, though a hash table is by far the most common.

phonebook = {"Alice": "0123", "Bob": "0456"}
phonebook["Carol"] = "0789"   # add
number = phonebook["Alice"]   # look up by key

The keys give the data meaning that a numeric index cannot. In a phonebook the key "Alice" tells you what the value represents, whereas an array index such as 7 is arbitrary. This is the conceptual heart of why dictionaries exist: they let you store and retrieve data by something humans care about.

Implementation with a hash table

It is worth keeping the two ideas separate in your mind: the dictionary is the interface (what you can do, expressed as operations on keys and values), while the hash table is one mechanism that delivers that interface efficiently. Exam questions often probe exactly this distinction.

When to use a dictionary

A dictionary is ideal when data is naturally accessed by a label rather than a position: looking up a customer record by account number, counting word frequencies, caching computed results so they are not recalculated, or representing a sparse mapping where most possible keys are absent. If you only need ordered data accessed by position, an array or list is simpler; if you need the data kept in sorted order, a balanced search tree may be more suitable than a hash-backed dictionary.

The sparse-mapping use is worth dwelling on because it shows the dictionary's memory advantage. Suppose you needed to record which of a billion possible product codes are in stock. An array indexed by product code would need a billion slots, almost all empty. A dictionary stores only the codes that are actually present, so its size grows with the data, not with the range of possible keys. This is why dictionaries are the natural representation for sparse data, configuration settings keyed by name, and any mapping where the set of keys is unpredictable or far smaller than the set of all possible keys.

Most modern programming languages provide a dictionary as a built-in type (Python dict, the maps and hash maps of other languages), and being a built-in does not change its status as an abstract data type: the language documents the operations and guarantees, and the runtime supplies an efficient hash-table implementation behind the scenes. When you answer an exam question you should describe the dictionary in terms of its operations on keys and values, and only mention the hash table when asked how it is implemented or why it is fast.