Basic requirements to get started:

• english -> any one of a language for now
• each word has only one meaning for now
• high availabilty and low latency
• around 100m translations every day
• around 100k words for each language, how would u store it?

My approach was to use a trie, so i implemented addword() and searchWord(), but i was stuck at a point how to store it in a persistent store. I was not sure how to do it but time was limited, so I came up with using NoSQL to store the prefix as key, the next character as child(e.g. one node as a parent, followed by 26 children, for english) in the NoSQL Database....

Actually, how should we store a trie in a database? I did some research and it seems like there is no general practice:(

Actually, I didn't get to the followups, but I guess these would be the potential questions:

• show different tenses(e.g. search drive , show drove, driven. search driven, show drive, drove)
• support more languages