Search not returning known matching results (let alone any results)

Submitted by James on Mon, 2012-10-08 05:58.Troubleshooting

I have an index called Breaking News. The SQL query I have that makes up the index returns over 8,000 results. Searching on that query (Select * From (select article.articleId,article.articleName,article.articleHeadline,(cast(article.articleCopy as varchar(max)) + ' ' + articleKeyword) as articleCopy,article.articleCreateDate, staff.staffName from article Left Join staff On (article.articleAuthorId = staff.staffId) where article.articleTypeId = 2 AND article.articleStatus='Prod') AS tempTbl Where articleName Like '%citi%' OR articleHeadline Like '%citi%' OR articleCopy Like '%citi%'), I get almost 500 results back. However, when I use dbsight to search 'citi', I get no results back. I tried reindexing and searching again with no luck. Here is my reindex log:
WARN 2012-09-25 10:16:30,484- IW 0 [main]: startCommit index=_0:C14 changeCount=4
WARN 2012-09-25 10:16:30,500- IW 0 [main]: done all syncs
WARN 2012-09-25 10:16:30,500- IW 0 [main]: commit: pendingCommit != null
WARN 2012-09-25 10:16:30,515- IW 0 [main]: commit: wrote segments file "segments_1"
WARN 2012-09-25 10:16:30,515- IFD [main]: now checkpoint "segments_1" [1 segments ; isCommit = true]
WARN 2012-09-25 10:16:30,515- IW 0 [main]: commit: done
WARN 2012-09-25 10:16:30,515- IW 0 [main]: at close: _0:C14
INFO 2012-09-25 10:16:30,531- Directory is marked ready: C:\Program Files\DBSight\webapps\dbsight\WEB-INF\data\indexes\Breaking\main1
INFO 2012-09-25 10:16:30,531- Indexing Context closed!
INFO 2012-09-25 10:16:30,531- Directory is marked ready as up-to-date: C:\Program Files\DBSight\webapps\dbsight\WEB-INF\data\indexes\Breaking\main1
INFO 2012-09-25 10:16:30,531- Directory is marked not ready: C:\Program Files\DBSight\webapps\dbsight\WEB-INF\data\indexes\Breaking\main
INFO 2012-09-25 10:16:30,531- Directory is marked ready as up-to-date: C:\Program Files\DBSight\webapps\dbsight\WEB-INF\data\indexes\Breaking\main1
INFO 2012-09-25 10:16:30,546- Directory is marked not ready: C:\Program Files\DBSight\webapps\dbsight\WEB-INF\data\indexes\Breaking\main
INFO 2012-09-25 10:16:44,671- ~~~~~~~ breaking:reBuildDictionary ...
INFO 2012-09-25 10:16:44,734- Opened index breaking
INFO 2012-09-25 10:16:44,859- Directory is marked ready: C:\Program Files\DBSight\webapps\dbsight\WEB-INF\data\indexes\Breaking\dictionary
INFO 2012-09-25 10:16:44,859- Complete Directory For breaking
INFO 2012-09-25 10:16:44,875- ~~~~~~~ breaking:buildDictionaryIfNeeded ...
INFO 2012-09-25 10:16:44,890- ~~~~~~~ breaking:ping-a-url ...
INFO 2012-09-25 10:16:44,890- ------- breaking:completed, Time used: 24 seconds
INFO 2012-09-25 10:16:44,890- Stop retrieving ...
WARN 2012-09-25 10:16:44,890- stopping indexing for breaking
INFO 2012-09-25 10:16:44,921- Period Table: [2007/02/01 21:20:39 EST ~ 2012/08/06 15:35:14 EDT]
INFO 2012-09-25 10:16:44,921- Indexing Context closed!
Monday, October 8, 2012 9:57:39 AM EDT

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
Submitted by James on Thu, 2012-10-11 04:21.

Thank you, I will try that. (I didn't even see the comments down here...)

Submitted by will on Wed, 2012-10-10 12:22.

Wild card matching %citi% is not the same as full text search by "citi".

If you want to search by "citi", you would need to have the analyzer analyze the text into a list of tokens, and "citi" should be one of the token.

You can experiment with the analyzer under the "Language" tab.

Submitted by James on Thu, 2012-10-11 04:48.

Thanks for the reply, but I don't really understand this. I've used the language tab to analyze the word "citi", as well as the word "citi" mixed in with other words to make sure it is being shown as a token correctly, and it is. Is that all you meant to do?

Submitted by will on Sun, 2012-10-14 18:01.

Can you show me the actual text coming out of database? With that, I can help you exactly what it should be.

Submitted by James on Mon, 2012-10-15 10:46.

An example would be:

With $78 billion in assets (at the end of 2010), Garanti doesn't come close to the size or scope of a Citi or American Express or JPMChase. When it comes to shaping powerful loyalty strategies, Garanti is second to none. The big boys should take note.

Submitted by will on Tue, 2012-10-16 13:10.

This should be trivial. The default analyzer,Number or LowerCase, or most other analyzer, should work. Make sure the field where the values comes from is not using analyzer "Case Insensitive Keyword", or "Comma-Semicolon".

Can you please send a screenshot of the "Data Source"->"Language" page to dbsight at gmail dot com, if you still have this problem?

Submitted by James on Wed, 2012-10-17 10:02.

I've sent the email with attachment and also a lot more data. :) Thanks for helping.