I was asked the following question during the phase of an interview.
There is a text file which stores different words. I need to find the first three words with the most number of occurrences. I first thought that I would use a map with each word as the key and the number of occurrences as the value. If any word was already present in the map I would increment the value (count). Then I realized I would require an iteration through the map keys to find the keys with the highest value associated with it. Further if I stored them in a SortedMap implementation (e.g. TreeMap) how would I be able to sort them on the values and not the keys?
If I were to have a Comparator implementation which sorts a Map based on the values instead of the keys and I pass this Comparator while creating the new instance of the TreeMap, would the Map be sorted on the values?
No it wouldn't. In maps the keys cannot change (in regards to equals/hashCode or compareTo for TreeMap). In your example the same applies to the values. If you change a value, the result of compare will change and the order will effectively change. But because the internal structure is not updated objects can never be found anymore.
You have two distinct problems here:
1) collect the counts
2) collect the top three
You already have part 1 finished, so all you need is part two. You could use the iteration, or you could copy all Map.Entry objects into a new SortedSet. Then take the first three elements. Take care though - what will you do if two words have the same number of occurrences? Your comparison should probably also use the word itself, if the two words occur equally often.