FILTER is an optional part of the query used to specify which part of the corpus should be included in the reply set. FILTER can include full-text keywords, facet filtering, or others.
A query without a FILTER part returns a reply set containing the whole corpus. These blank queries can be used to retrieve statistical counters from the corpus thanks to facets. But they should be used with caution on large datasets because they require the processing of each document in the corpus.