Am 09.08.2011 14:58, schrieb Ahmet Arslan:
while searching with debug on I see strange query parsing:
<str
name="rawquerystring">identifier:"ub.uni-bielefeld.de"</str>
<str
name="querystring">identifier:"ub.uni-bielefeld.de"</str>
<str name="parsedquery">
+MultiPhraseQuery(identifier:"(ub.uni-bielefeld.de ub) uni
bielefeld de")
</str>
<str name="parsedquery_toString">
+identifier:"(ub.uni-bielefeld.de ub) uni bielefeld de"
</str>
It is a PhraseQuery, but
- why is the string split apart?
WordDelimiterFilterFactory breaks your text and generates token at the same
position. analysis.jsp displays step by step analysis phase.
Ahhh, I see my misunderstanding.
Thought that a PhraseQuery is not touched by WordDelimiterFilter.
- why is it grouped this way?
If there are multiple tokens at same position, MultiPhraseQuery is generated instead of PhraseQuery. Assume that
'quick' is synonym of 'fast'. "quick cat" becomes "(quick fast) cat". It accepts both
"quick cat" and fast cat"
OK, due to "preserveOriginal=1" the original query and the first token
are at the same position and this builds the first group.
And according to your explanation "quick cat" --> "(quick fast) cat"
the grouped tokens are ORed and the others are ANDed.
Is that right?
