After reading the excellent article by Ashwin, Now i am thinking of implementing the dictionary database as a ternary search tree. Here is an another link to an atricle on ternary search tree. http://www.javaworld.com/javaworld/jw-02-2001/jw-0216-ternary.html I like this approach as you could search for partial matches also. As i started to write scripts to load various dictionaries i found that, each of them have their own unique attributes. For example, the elements dictionary needs to be searched by element name, symbol or atomic number and the english dictionary has word definitions that belong to a type(noun, adjective ..) .This will create unique problems for this type of algorithm, although each of the attributes could be used a key.