[xwiki-users] RE: Questions about the Lucene-plugin
I think it's impossible to do the paging properly under the conditions mentioned below. So I have a flat results view but I've done some efforts for a compact and easily customizable search results view, which has sorting and sort direction functionality. If you want to have it you should also have my lucene-patches to support sorting directions and to do AND-searches by default. If you want I can send these all to you but maybe somebody of the project team could shortly comment these issues before. If there is a roadmap with the plugin I could also offer myself to be a contributor. Lucene-plugin works 100X (well...almost) faster than regular xwiki-searches and is at least at the moment the only possible choice in my xwiki-environment. And by the way, many thanks for Ludovic and the whole team for great work! Regards, Petteri ________________________________ Lähettäjä: Michael Guntli [mailto:michael.guntli@leica-geosystems.com] Lähetetty: ke 9.5.2007 11:12 Vastaanottaja: xwiki-users@objectweb.org Aihe: [xwiki-users] RE: Questions about the Lucene-plugin I had the same problem with post-filtering the search results. (there were not all results displayed, and that caused a problem in a paged view (ex. just 10 out of 30 on page1, 15 of 30 on page2..) Maybe you can improve my lucene search page macro? http://www.xwiki.org/xwiki/bin/view/Code/LuceneSearchPage petteri.karttunen wrote:
About the "confusing" results: of course I can do some velocity- or groovy -post-processing on the page to filter out the duplicate document names from results. This is no problem if I want to show the results as one flat list but it becomes quite complicated to do efficiently if I want to show f.ex. 30 hits per page with paging. Basically I mean that all the index data holded by one page (page content, objects, attachments...) should be regarded as this page's properties and indexed under one lucenedoc. If I operationally have understood something incorrectly, please correct me.
What do you think?
-Petteri
I'm testing the version RC1 and have some thoughts about the Lucene-plugin:
Sorting ------- At the moment there is no method in the API to choose between the sorting directions. I think there should be one.
Boolean Queries --------------- Is the default query-type going to stay an ORed one? Should it better be an AND-query, like Google? I think most of the surfers expect that nowadays. At least there could be method to choose between them.
Results ------- Search results are a bit confusing now. When I search for a word that appears both in a document, in it's attachment and in it's object's properties I get three hits. One for each type. There a special cases, of course, but I think it would be more usable to get in my example case by default just one hit. There would still remain a method ("type: object") to distinguish between the types in a query.
If you like, I can provide patches for the two first (minor) issues.
Regards, Petteri
-- You receive this message as a subscriber of the xwiki-users@objectweb.org mailing list. To unsubscribe: mailto:xwiki-users-unsubscribe@objectweb.org For general help: mailto:sympa@objectweb.org?subject=help ObjectWeb mailing lists service home page: http://www.objectweb.org/wws
-- View this message in context: http://www.nabble.com/Questions-about-the-Lucene-plugin-tf3704984.html#a1039... Sent from the XWiki- Users mailing list archive at Nabble.com.
I hope xwiki will use the lucene search in future as default search engine! At least just because of the result score.. petteri.karttunen wrote:
I think it's impossible to do the paging properly under the conditions mentioned below. So I have a flat results view but I've done some efforts for a compact and easily customizable search results view, which has sorting and sort direction functionality. If you want to have it you should also have my lucene-patches to support sorting directions and to do AND-searches by default.
If you want I can send these all to you but maybe somebody of the project team could shortly comment these issues before. If there is a roadmap with the plugin I could also offer myself to be a contributor. Lucene-plugin works 100X (well...almost) faster than regular xwiki-searches and is at least at the moment the only possible choice in my xwiki-environment.
And by the way, many thanks for Ludovic and the whole team for great work!
Regards, Petteri
________________________________
Lähettäjä: Michael Guntli [mailto:michael.guntli@leica-geosystems.com] Lähetetty: ke 9.5.2007 11:12 Vastaanottaja: xwiki-users@objectweb.org Aihe: [xwiki-users] RE: Questions about the Lucene-plugin
I had the same problem with post-filtering the search results. (there were not all results displayed, and that caused a problem in a paged view (ex. just 10 out of 30 on page1, 15 of 30 on page2..)
Maybe you can improve my lucene search page macro? http://www.xwiki.org/xwiki/bin/view/Code/LuceneSearchPage
petteri.karttunen wrote:
About the "confusing" results: of course I can do some velocity- or groovy -post-processing on the page to filter out the duplicate document names from results. This is no problem if I want to show the results as one flat list but it becomes quite complicated to do efficiently if I want to show f.ex. 30 hits per page with paging. Basically I mean that all the index data holded by one page (page content, objects, attachments...) should be regarded as this page's properties and indexed under one lucenedoc. If I operationally have understood something incorrectly, please correct me.
What do you think?
-Petteri
I'm testing the version RC1 and have some thoughts about the Lucene-plugin:
Sorting ------- At the moment there is no method in the API to choose between the sorting directions. I think there should be one.
Boolean Queries --------------- Is the default query-type going to stay an ORed one? Should it better be an AND-query, like Google? I think most of the surfers expect that nowadays. At least there could be method to choose between them.
Results ------- Search results are a bit confusing now. When I search for a word that appears both in a document, in it's attachment and in it's object's properties I get three hits. One for each type. There a special cases, of course, but I think it would be more usable to get in my example case by default just one hit. There would still remain a method ("type: object") to distinguish between the types in a query.
If you like, I can provide patches for the two first (minor) issues.
Regards, Petteri
-- You receive this message as a subscriber of the xwiki-users@objectweb.org mailing list. To unsubscribe: mailto:xwiki-users-unsubscribe@objectweb.org For general help: mailto:sympa@objectweb.org?subject=help ObjectWeb mailing lists service home page: http://www.objectweb.org/wws
-- View this message in context: http://www.nabble.com/Questions-about-the-Lucene-plugin-tf3704984.html#a1039... Sent from the XWiki- Users mailing list archive at Nabble.com.
-- You receive this message as a subscriber of the xwiki-users@objectweb.org mailing list. To unsubscribe: mailto:xwiki-users-unsubscribe@objectweb.org For general help: mailto:sympa@objectweb.org?subject=help ObjectWeb mailing lists service home page: http://www.objectweb.org/wws
-- View this message in context: http://www.nabble.com/Questions-about-the-Lucene-plugin-tf3704984.html#a1040... Sent from the XWiki- Users mailing list archive at Nabble.com.
participants (2)
-
Karttunen Petteri -
Michael Guntli