You may find yourself wondering – just how is ODict so fast? ODict uses the marvelous FlatBuffers serialization library from Google, which allows us to store data in a well-formed, structured structured binary schema. In this schema, we can assign fields that acts as lookup keys – in this case, headwords.
A dictionary’s entries are sorted by their keys and rapidly retrieved using a binary search on the serialized buffer. This means we don’t even have to have to deserialize the entire dictionary and load it into memory before finding words!
For more information on the file specification, see here.