Each field that is indexed also includes a special value called norms,
which includes three components:
-
Document Boost: a per-document boost provided at index-time
-
Field Boost: a per-field boost also provided at index-time
-
Length Norm: a normalization where shorter fields obtain a higher boost relative to longer fields. This prevents long fields from matching higher simply because they are longer.
By default, norms are calculated for any analyzed field, and not calculated
for not_analyzed field. But if you are using one of the normalization schemes
we just talked about (e.g. Keyword + lowercase), you may wish to disable norms
to save space because you don’t care about score at all.
Disabling norms is as simple as setting omit_norms: true in your mapping.
Similarly, you may wish to change the default index_options value. This
configuration controls the type of data that is stored about each field.
The options are:
-
docs: Indexes only the doc ID
-
freqs: Indexes doc IDs and term frequencies
-
positions: Indexes doc IDs, term frequencies and token positions
Analyzed fields default to "positions", while not_analyzed default to "freqs".
Again, if you are using Keyword + lowercase to make a case-insensitive not_analyzed,
it may make sense to change index_options to "freqs".
You only have a single token, so the position information is redundant and useless - it is just taking up space.