The upgrade did not correct the problem.
_____ From: general-boun...@developer.marklogic.com [mailto:general-boun...@developer.marklogic.com] On Behalf Of Tim Meagher Sent: Tuesday, June 23, 2009 11:16 AM To: 'General Mark Logic Developer Discussion' Subject: [MarkLogic Dev General] RE: Can someone explain what the wildcardsearch problem was that was fixed in 4.0.5? Hi Doug, et al, Somehow I missed your post regarding this subject until a colleague pointed it out to me, so I recovered your post from markmail and included it below. Yes, I do realize that the two search strings are mutually exclusive. I should have made that clearer in my post. I tried the query with stemming enabled and disabled, but that does not fix the wildcard problem. The main problem is that supplying a wildcarded query of "Personality*" returns results that do not include any form of the word "Personality". For example, I get "Abstract" as one of the results. I expect to get only one result, that being "Personality". I've noticed this anomaly before and kind of chalked it up to problems related to an unfiltered search, but even so I haven't fully understood why this is happening. Case-insensitive searches work fine and changing the request to all lower case does not solve the wildcard problem. The support staff at MarkLogic opened up a support case for this issue and recommended that I upgrade form 4.0.4 to 4.0.5 as a known wildcard bug was found in 4.0.3, so I'm curious to see the results after the upgrade today. Thank you! Tim Meagher Tim, The particular issue that was being discussed in that thread involved a bug only in cts:element-attribute-value-queries with wildcards, so I don't think it applies to your situation. A couple of notes on your situation: * Just to make sure, you realize that your two search strings are mutually exclusive, right? The search for "Personality*" will return only results where the Title is a single word that starts with "Personality". The search for "Personality *" will return only results where the Title is multiple words, the first of which is "Personality". To match both of those, you would need to search for "Personality* *". * I'm not sure (maybe there's somebody else on the list who has a better understanding and could explain it) how using a trailing wildcard and stemming in the same search acts. I'd assume you should also be getting results for "Personalities*", and maybe "Personalize*" or similar words (don't know how much variation the stemming engine allows). * Even though you explicitly specified case-insensitive, have you tried putting the search string in all lower-case? Just an idea to try to narrow down the scope. Otherwise, could you perhaps include some examples of results that are being included that shouldn't be? Doug Glidden _____ From: Tim Meagher [mailto:t...@aaom.net] Sent: Thursday, June 18, 2009 11:14 AM To: 'General Mark Logic Developer Discussion' Subject: Can someone explain what the wildcard search problem was that was fixed in 4.0.5? Hi Folks, I noticed a comment by Danny Sokolsky at one point. I'm running into some interesting results performing a wildcarded lexicon search using cts:element-values() with a cts:element-value-query() subsearch: cts:element-values(xs:QName("Title"), (), ("item-frequency", "item-order", "descending"), cts:and-query(( cts:element-value-query(xs:QName("Title"), string("Personality*"), ("case-insensitive", "diacritic-insensitive", "punctuation-insensitive", "stemmed", "wildcarded")), cts:directory-query("/Root/SubDir/", "infinity") )) ) Some of the results do not contain "Personality". If I add a space before the asterisk in the search ("Personality *"), then the results only include titles containing "Personality". Note that I have set my fragment root so that there should be only one occurrence of the element "Title" within a fragment. Thank you! Tim Meagher - AAOM Consulting
_______________________________________________ General mailing list General@developer.marklogic.com http://xqzone.com/mailman/listinfo/general
