refine search by categories miscounted

[maryjili]

I just found the number of results for each category in the "refine search by categories" section is not consistent with that in the individual category search. The number in the "refine" section is much smaller. Does it has anything to do with incremental indexing?

[David]

What version of the software are you using? There was a bug like this in an earlier patch release.

[maryjili]

I tested with 1016 and 1017, incremental indexing and continuous indexing, all got the same results.

[maryjili]

I have many categories, and the first a few sets being imported and indexed may not contain urls in all the categories.

[Ray]

Check that you are actually using the latest CGI file generated with the latest build. That is, you're not using a "search.cgi" file from a previous index and only updating the .zdat files on upload. If you are using ASP.NET, you'd need to uninstall and reinstall the server control.

I am not sure what you meant by your last post.

Again, it would help if we can see the problem and if you can send us the index files. If you are worried about data confidentiality, try to reproduce the problem with a smaller set of files (which are not confidential) and send us the index for that.

[maryjili]

I was using the latest cgi file. Please ignore my last post, I was just guessing the cause of the problem which now I don't think it was. After testing again with your site, I got the same miscounted categories, and two new files created and saved to the same index folders: zoom_plugin.in, zoom_plugin.out. Are they virus files or some intermediary zoom index files? Below is the zcfg file I just used to test with 1017.

__6_0
#STARTDIR:
#SPIDERURL:http://wrensoft.com
#BASEURL:http://wrensoft.com/
#OUTDIR:c:\test\
#SPIDERURLTYPE:0
#SPIDERURLUSELIMIT:0
#SPIDERURLLIMIT:0
#SPIDERURLBOOST:0
#USE-CRC:1
#CURRENTMODE:1

<Moderator: Removed 400 lines of config file>

#SITEMAP_USEPAGEBOOST:1
#SITEMAP_USEBASEURL:1
#SITEMAP_BASEURL:http://www.mywebsite.com/

[Ray]

After testing again with your site, I got the same miscounted categories, and two new files created and saved to the same index folders: zoom_plugin.in, zoom_plugin.out. Are they virus files or some intermediary zoom index files?

They are not virus files. They are intermediary files used during the plugin processing. However, they should not be left behind at the end of indexing. This is probably a sign that something went wrong, and there should be an error/warning message corresponding to this in the log.

I'm not sure why you insist on not sending files via e-mail despite our request and instructions, time and time again. Pasting the configuration file here just adds room for error and we have to manually fix things up just to test them (the forum adds spaces to lines which are too long for example).

Despite this, we were able to reproduce the problem here. And we have in fact confirmed that this is a bug in the current release. The problem is that when a search returns many results, some items are culled out of the result set (because it knows it will never display them, as they are very far back in the order of relevance). Unfortunately, the bug caused these items to also be culled out of the counting process for the category summary. This is why sometimes it reports the correct number of files in that category, and sometimes it reports less, depending on the search query.

This will be fixed in the next build (V6.1 build 101.