You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@opennlp.apache.org by "Boris Galitsky (Updated) (JIRA)" <ji...@apache.org> on 2011/11/18 00:21:54 UTC

[jira] [Updated] (OPENNLP-387) Demonstration on how similarity component improves search accuracy. A query is run through Bing search API, and less relevant hits are sorted out by similarity measure between the query and the snippet

     [ https://issues.apache.org/jira/browse/OPENNLP-387?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Boris Galitsky updated OPENNLP-387:
-----------------------------------

    Attachment: patch.OPENNLP-387.txt

It is now illustrated how to build search engine based on similarity. 3 new tests are added, including the one which goes to Bing for search and then re-sorts results based on similarity. Some fixes for the cases of problematic parsings are done.
                
> Demonstration on how similarity component improves search accuracy. A query is run through Bing search API, and less relevant hits are sorted out by similarity measure between the query and the snippet
> ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: OPENNLP-387
>                 URL: https://issues.apache.org/jira/browse/OPENNLP-387
>             Project: OpenNLP
>          Issue Type: Improvement
>          Components: Similarity
>    Affects Versions: 1.6.0
>            Reporter: Boris Galitsky
>            Assignee: Boris Galitsky
>         Attachments: patch.OPENNLP-387.txt
>
>   Original Estimate: 48h
>  Remaining Estimate: 48h
>
> Right now we use a fancy domain to demonstrate the usability of similarity component, such as content generation based on web mining. I believe it also makes sense to use a simpler case such as search relevance improvement by use of similarity component. We get search candidates by TF*IDF and then apply OpenNLP parsing and then Similarity component to assess relevance of a snippet to an answer.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira