Dictionaries in Minion: An Introduction
As I've mentioned previously, our design philosophy for Minion is that everything is a dictionary and postings.
To recap, in Minion, a dictionary is a map from entry names to entries. The entry name may be a term from a document, a value from a saved field, or a bigram that will be used for doing wildcard matching. The entry pointed to by the name contains information about that name, including frequency of occurrence, the dictionary that the entry was drawn from, and (most importantly) a pointer to the postings associated with that entry.
There are actually two kinds of dictionaries in Minion. The first kind is a MemoryDictionary, which is used to hold a dictionary at indexing time. The second kind is a DiskDictionary, which is used to hold a dictionary at query time. Both of these dictionary types are implementations of the Dictionary interface that describes the capabilities that all dictionaries share.
Over the next few posts I'll be describing the entries that you'll find in dictionaries, and how the MemoryDictionary and DiskDictionary classes are used in Minion.
Posts in this series:
