Question

I have a graph representation such that each nodes contains (int id, String categoryName). And it consist of approx 500,000 nodes. Right now say I would like to check if categoryName "computing" exist in the graph, would that mean I have to traverse through the whole graph like using BSF method? (correct me if I am wrong here). It is pretty slow now (takes up to about 3 minutes) to check if that categoryName exist, any advise on how can I improve the speed of searching and comparing of the String values?

Answer 1

If you have no other data – then yes, searching the graph will be needed.

However, you can pre-process and use a Map<String,Node> (Where Node is your node class) that maps from the strings to your actual nodes, and quickly find if a certain String has a relevant node.

You can get the node from the string by using: map.get(string)

You can use a TreeMap (which has O(logn) seek & insertion time and is ordered), and a HashMap (which implements a hash table and has O(1) average case seek & insertion time)

---

Note: this assumes no duplicates, i.e. no string has two different nodes representing it.

If this assumption is not correct, you can use something like Guava Multimap