11209 2010-11-03 15:57:30 +0000 [FT] Wildcard and thesaurus options in combination 2010-11-30 05:37:06 +0000 1 1 1 Unclassified XPath / XQuery / XSLT Full Text 1.0 Candidate Recommendation PC Windows NT CLOSED FIXED P2 normal --- 1 tim jim.melton jmdyck public-qt-comments oldest_to_newest 42042 0 tim 2010-11-03 15:57:30 +0000 Suppose an implementation's default thesaurus expands each of the tokens "three" and "3.0" to the two query items "three" and "3.0". Consider queries such as: "version 3.0" contains text "three" using wildcards using default thesaurus and "version three" contains text "3.0" using wildcards using default thesaurus It seems that thesaurus expansion can occur even when the wildcard option is enabled. However, in: fts:lookupThesaurus( $queryTokens, $matchOption/fts:thesaurusName, $matchOption/@language, $matchOption/fts:relationship, $matchOption/fts:range ) it appears that the thesaurus won't know whether characters in the query tokens should be interpreted as wildcards or not. Also the thesaurus might need to escape its results ("three" expanding to "3\.0") to avoid triggering wildcard matching. 42875 1 jmdyck 2010-11-30 05:36:43 +0000 At their joint 2010-11-23 meeting, the XQuery/XSL WGs approved the following fix: -- In 4.2.5.2, have fts:applyQueryTokensAsPhrase() pass $noThesaurusOptions to fts:applyThesaurusOption(). -- In 4.2.5.7, have fts:applyThesaurusOption() accept $noThesaurusOptions and pass it on to fts:lookupThesaurus(). -- In 4.2.5.3, have fts:lookupThesaurus() [an external (implementation-defined) function] accept $noThesaurusOptions, and in the prose, say something to the effect that by giving it access to the other match options, we allow the implementation to apply any of those match options (whichever it deems relevant) to the input or output of the "actual" thesaurus lookup. (They also approved the subsequent cleanup of eliminating the now-redundant $thesaurusLanguage parameter from fts:lookupThesaurus(), and the now-unnecessary "language" attribute from complexType ftThesaurusOption.) I have made these changes to the editors' draft of the Full Text spec, and so am marking this issue resolved-FIXED. Since you were at the meeting and indicated that these changes would be satisfactory, I will also mark the issue CLOSED.