View Full Version : Exclusion/negative searches
01-16-2009, 08:46 PM
I am trying to verify that the Exclusion/negative search function on our intranet is working correctly and am having some trouble.
If I search for 'restaurant chicken' with the Any Search Words selected I get 3 pages with both and 86 pages with one or the other. If I want to search for all pages that contain restaurant but not chicken I can search for 'restaurant -chicken' and get 80 results which seems right. If I enclose it in double quotes "restaurant -chicken" I get no results. That doesn't seem right.
If I search for Latin I get 36 results. If I search for Kings I get 191 results. If I search for 'latin kings' I get 26 containing both and 175 containing one or the other. But, if I search for 'latin -kings' I still get 36 and can see the word 'kings' in the results. That doesn't seem right. If I search for "latin -kings" I get no results which doesn't seem right especially since I have found at least one document which contains the word 'Latin' but not the word 'King'.
Am I doing something wrong or is this search not working correctly?
01-17-2009, 12:07 AM
The minus sign can be used for negative searches in multiple word searches. However if you use double quotes, you are then doing an exact phrase search, and having a word excluded within an exact phrase doesn't make much sense, and is not supported search syntax (http://www.wrensoft.com/zoom/support/searchtips.html).
The 2nd case of searching latin -kings (without quotes) is more interesting. What is the URL for the search function?
01-19-2009, 04:49 PM
Thank you for your response. Since the example on the Search Tips page was in double-quotes I assumed that it meant that you were supposed to use double-quotes.
As for the url with the 'Latin- Kings' search, and my other searches, it is an internal intranet site. Now that I know that the exclusion/negative search is to be done with the use of quotes my results are better. The Latin -Kings search is still coming back wrong but all of the documents returned are pdf files and maybe there is something in how those pdfs were created. I have tried several other searches without using the double quotes and are have worked successfully.
01-19-2009, 06:35 PM
Now that I know that the exclusion/negative search is to be done with the use of quotes...
The only time you should use quotes is when you want an exact phrase search.
01-19-2009, 08:27 PM
Sorry, I meant WITHOUT the use of quotes. Thanks again.
If you are still seeing a problem, ZIP up your search files and e-mail them to us (http://www.wrensoft.com/contactus.html). Give us the exact search scenario we need to reproduce the issue and include any other files needed to demonstrate the problem.
01-20-2009, 07:18 PM
I will zip up my search files and send them to you. I will have to separate the search files in to two separate zip files because the 'zoom_pagetext.zdat' file is 29MB.
The exact search that I am performing is 'latin -kings' with no quotes. From my search I get 36 results, nearly all of which contain the word 'kings' even though it shouldn't. All but three of four of the files that are found are pdf files which may end up totalling several MB. Do you need all of the pdf files or will a sample do?
Just a sample will do (i.e. one PDF file which is included in the index and demonstratably contains the word "kings" but is not excluded with the negative search). And yes, please send us the index files.
01-27-2009, 01:58 PM
Sorry, I got sidetracked on a different problem. I will get them sent today.
UPDATE: Messages have been sent. Had to send three emails due to file size restrictions. I included all of the search files plus 4 of the affected pdf files.
Thanks again for your assistance.
We have confirmed that there is a bug with exclusion searches, when Stemming is enabled (and when the search word being excluded is affected by stemming).
This will be fixed in the next release (V6.0 build 1009). Disabling stemming (on the Languages panel) will workaround the issue but it might be best to wait for the next release.
01-29-2009, 08:12 PM
Thank you for your help with this. At this time, stemming is more important to us than the exclusion/negative searches are. We will leave stemming enabled and wait for the updated version.
Powered by vBulletin® Version 4.1.12 Copyright © 2013 vBulletin Solutions, Inc. All rights reserved.