In previous versions of CorpusSearch, subqueries had to be appended one at a time,
like this:
query: ((((A function B)
AND (C function D))
AND (E function F))
AND (G function H))
In CorpusSearch 2, subqueries can be appended with any logical combination of
parentheses. The above query is more easily written:
query: (A function B)
AND (C function D)
AND (E function F)
AND (G function H)
Old-style queries still work.
In previous versions of CS, the only query conjunction was
"AND".
CorpusSearch 2 has added
"OR" and
"NOT".
There is a new node-boundary option, $ROOT,
which is a variable that stands for the root node of a
matrix sentence (token), whatever its label may be. Even in cases where the
root node has no label, it can be referred to with $ROOT.
new search functions include:
-
Dominates
iDomsMod
hasSister
isRoot
sameIndex
These search functions have new algorithms:
-
iPrecedes
Precedes
Also, search functions that take an integer argument no longer have the
argument jammed onto the end of the function name, but are written with
a space before the argument. For example, instead of:
(WRONG!) query: (CODING column2 q)
you should now write this:
query: (CODING column 2 q)
These functions are affected:
-
column
domsWords
domsWords<
domsWords>
iDomsNumber
iDomsTotal
iDomsTotal<
iDomsTotal>
-
reformat_corpus
print_only
Coding is now performed once per boundary node, instead of once per sentence (token).
Also, the command "coding_query:" must now appear before the coding query.
CorpusSearch 2 includes
lexicon building,
automated corpus revision, and
local-frames production.
-
iDoms_conj_switch: see iDomsMod
anyPrecedes: see precedes
print_complement: see NOT