Java - Is it common practice to use a hashtable (eg HashMap) to map objects to themselves?

Question

I'm making a java application that is going to be storing a bunch of random words (which can be added to or deleted from the application at any time). I want fast lookups to see whether a given word is in the dictionary or not. What would be the best java data structure to use for this? As of now, I was thinking about using a hashMap, and using the same word as both a value and the key for that value. Is this common practice? Using the same string for both the key and value in a (key,value) pair seems weird to me so I wanted to make sure that there wasn't some better idea that I was overlooking.

I was also thinking about alternatively using a treeMap to keep the words sorted, giving me an O(lgn) lookup time, but the hashMap should give an expected O(1) lookup time as I understand it, so I figured that would be better.

So basically I just want to make sure the hashMap idea with the strings doubling as both key and value in each (key,value) pair would be a good decision. Thanks.

Answer 1

I want fast lookups to see whether a given word is in the dictionary or not. What would be the best java data structure to use for this?

This is the textbook usecase of a Set . You can use a HashSet . The naive implementation for Set<T> uses a corresponding Map<T, Object> to simply mark whether the entry exists or not.

Answer 2

If you're storing it as a collection of words in a dictionary, I'd suggest taking a look at Tries. They require less memory than a Set and have quick lookup times of worst case O(string length) .

Answer 3

Any class that is a Set should help your purpose. However, Do note that Set will not allow for duplicates. For that matter, even a Map won't allow duplicate keys. I would suggest on using a an ArrayList (assuming synchronization is not needed) if you need to add duplicate entries and treat them as separate.

Answer 4

My only concern would be memory, if you use the HashSet and if you have a very large collection of words... Then you will have to load the entire collection in the memory... If that's not a problem.... (And your collection must be very large for this to be a problem)... Then the HashSet should be fine... If you indeed have a very large collection of words, then you can try to use a tree, and only load in memory the parts that you are interested in.

Also keep in mind that insertion is fast, but not as fast as in a tree, remember that for this to work, Java is going to insert every element sorted. Again, nothing major, but if you add a lot of words at a time, you may consider using a tree...