Uploaded image for project: 'Jira Data Center'
  1. Jira Data Center
  2. JRASERVER-13672

Better searching when stemming is in place. Improve Lucene QueryParser to perform analysis on prefixed queries.

XMLWordPrintable

    • We collect Jira feedback from various sources, and we evaluate what we've collected when planning our product roadmap. To understand how this piece of feedback will be reviewed, see our Implementation of New Features Policy.

      NOTE: This suggestion is for JIRA Server. Using JIRA Cloud? See the corresponding suggestion.

      Currently the default using of "Porter" stemming means that certain search phrases do not return results that the user might expect.

      For example if the issue summary is "Moose problems" then the default stemming will cause "moos" to be stored in the index.

      If the user then enters a search request for "Moose" then no match will be made. The reason is that the Lucene QueryParser code does NOT perform analysis on "Prefix" queries by default.

      Hence a PrefixQuery for the term "moose" is submitted and the index doent contain this terms and hence no issue is returned.

      The improvement would be to tweak QueryParser so that it does perform "analysis" on prefix queries. So the term "moose" would be stemmed to "moos" and then the Prefix Query would return the expected results.

        1. Lucene-JIRA+search.doc
          45 kB

            [JRASERVER-13672] Better searching when stemming is in place. Improve Lucene QueryParser to perform analysis on prefixed queries.

            Kerrod Williams (Inactive) added a comment -

            Thanks for taking the time to raise this issue.

            Due to the large volume of JIRA feature suggestions, we have to prioritise our development efforts. In part, that means concentrating on those issues that resonate the most with our users.

            I am writing this note to advise you, that we have decided to close your Suggestion as it has not gained traction on jira.atlassian.com. We believe being upfront and direct with you will assist you in your decision making rather than believing Atlassian will eventually address this issue.

            Thank you again for your suggestion and if you have any concerns or question, please don’t hesitate to email me.
            Kind Regards,
            Kerrod Williams
            JIRA Product Management
            kerrod.williams at atlassian dot com

            Kerrod Williams (Inactive) added a comment - Thanks for taking the time to raise this issue. Due to the large volume of JIRA feature suggestions, we have to prioritise our development efforts . In part, that means concentrating on those issues that resonate the most with our users. I am writing this note to advise you, that we have decided to close your Suggestion as it has not gained traction on jira.atlassian.com. We believe being upfront and direct with you will assist you in your decision making rather than believing Atlassian will eventually address this issue. Thank you again for your suggestion and if you have any concerns or question, please don’t hesitate to email me. Kind Regards, Kerrod Williams JIRA Product Management kerrod.williams at atlassian dot com

            bain added a comment -

            Just wanted to add a pointer to this: http://wiki.apache.org/lucene-java/LuceneFAQ#head-133cf44dd3dff3680c96c1316a663e881eeac35a.

            It explains why lucene does not stem the query string by default. Not saying whether this is good or bad, just adding the document.

            bain added a comment - Just wanted to add a pointer to this: http://wiki.apache.org/lucene-java/LuceneFAQ#head-133cf44dd3dff3680c96c1316a663e881eeac35a . It explains why lucene does not stem the query string by default. Not saying whether this is good or bad, just adding the document.

            G B added a comment -


            Brian, great document. I laughed, I cried... I submitted JRA-14641.

            G B added a comment - Brian, great document. I laughed, I cried... I submitted JRA-14641 .

            Tom Clarkson added a comment -

            I agree that this is a defect and the item should be marked as a bug not an improvement (we're using 3.12). I have some issues which I am unable to search for at all... I just have to remember the key... lame!

            Tom Clarkson added a comment - I agree that this is a defect and the item should be marked as a bug not an improvement (we're using 3.12). I have some issues which I am unable to search for at all... I just have to remember the key... lame!

            Brian Riedinger added a comment -

            The attached document includes the findings of searches we ran against JIRA 3.11.

            Brian Riedinger added a comment - The attached document includes the findings of searches we ran against JIRA 3.11.

            Brian Riedinger added a comment -

            This is not an improvement, its a defect. I understand that there is a logical explaination as to why this happens, but from the users' perspective, search is unreliable.

            Our customers are having a hard time getting over this. So much so that we're considering building our own external search service.

            Brian Riedinger added a comment - This is not an improvement, its a defect. I understand that there is a logical explaination as to why this happens, but from the users' perspective, search is unreliable. Our customers are having a hard time getting over this. So much so that we're considering building our own external search service.

            Jens Wilke added a comment -

            This is even worse here!

            Here is the behaviour of the search for the term "Affiliate-ID" on v3.12:

            "affiliate" -> result correct
            "aff*" -> result correct
            "affi*" -> result correct
            "affili*" -> result correct
            "affilia*" -> no result!

            Jens Wilke added a comment - This is even worse here! Here is the behaviour of the search for the term "Affiliate-ID" on v3.12: "affiliate" -> result correct "aff*" -> result correct "affi*" -> result correct "affili*" -> result correct "affilia*" -> no result!

            ɹǝʞɐq pɐɹq added a comment -

            Some of the issues raised in JRA-5567 would be addressed here.

            ɹǝʞɐq pɐɹq added a comment - Some of the issues raised in JRA-5567 would be addressed here.

              Unassigned Unassigned
              bbaker ɹǝʞɐq pɐɹq
              Votes:
              6 Vote for this issue
              Watchers:
              5 Start watching this issue

                Created:
                Updated:
                Resolved: