No index configuration setup is strictly necessary to enable spelling correction.
By default, all words contained in searchable dimensions and properties will be considered as possible spell correction recommendations. But in practice, to achieve the best possible spelling correction behavior and performance, it is typically necessary to configure bounds on the list of words available for spelling correction, commonly known as the dictionary.
The application-specific spelling dictionary is created by Dgidx. As Dgidx creates search indexes of property and dimension value text, it accumulates lists of words available for spelling correction by the Aspell module into the following file:
<db_prefix>.worddat(for the Aspell module only)
where
is the output index
prefix.
<db_prefix>
Note
The
<db_prefix>.worddat file for the Aspell
module is reloaded into the MDEX Engine each time you run the
admin?op=updateaspell administrative command. This
command enables you to make updates to the Aspell spelling dictionary without
stopping and restarting the dgraph.
This file contains application-specific dictionary words separated by new-line characters. Duplicate words listed in these files are ignored.
This file is automatically compiled by the Dgidx during the indexing operation.
