Question

1. In general, how useful is the Dictionary/Hash Table data structure? Can you think of other uses aside from those?
2. If we didn’t have the Dictionary/Hash Table data structure can you think of another way to implement the dictionary concept?
3. Can the Dictionary/Hash Table ADT accept/handle null key entries? What about null values? Why or why not?
4. Given these behaviors, what conclusions can you draw about tuning a hash table for different data sets?

Answer 1

(i) A dictionary is a concept that maps keys to values. There are many ways to implement such mapping. One such method is the hash table. Using a hash function, a hash table calculates an index into an array of buckets/slots, which is then used to find the desired values.

An advantage of such a data structure is speed. This can be practically observed when there is a lot of data. When the maximum number of entries can be predicted in advance, hash tables can be particularly efficient because the bucket array can be allocated once with the optimum size.

Insertion is fast (assuming less collisions on an average) as the location of insert is calculated based on the hash function. A direct effect of this is that searching a hash table is considerably faster than a list-based structure, as the hash for the search item can be calculated, and the possible location(s) can be directly accessed to check whether the element is present.

(ii) Hash tables are used for database indexing, caching etc.

(iii) Another common way to implement dictionaries is red-black trees. Each method has its own advantages and disadvantages. A red-black tree, for example, can always perform a lookup in O(log N). A hashtable can perform a lookup in O(1) time, although that can degrade to O(N) depending on the hash function and the input.

(iv) Depending on the implementation, a dictionary ADT may or may not allow null keys. Even if they do, there can be only one null key, as keys must be unique. This can be thought of like a default case. A lot of implementations will simply throw a null pointer exception when attempting to insert an entry with a null key. Null values are allowed because it is possible for multiple keys to contain no data. Values can repeat.

(v) Depending on the characteristics of a dataset, it is important to decide a few things: the hash function, bucket size, implementation type (hash table, red-black trees...), whether to allow null key etc. These features can be decided based on the data AND the requirements and general access trends for the data by the application.

Hope this answers your question. All the best!