[
https://issues.apache.org/jira/browse/NUTCH-2209?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15129637#comment-15129637
]
ASF GitHub Bot commented on NUTCH-2209:
---------------------------------------
Github user sujen1412 commented on a diff in the pull request:
https://github.com/apache/nutch/pull/87#discussion_r51670857
--- Diff: conf/nutch-default.xml ---
@@ -1390,7 +1390,7 @@ For more detailed information on the working of this
filter
visit https://wiki.apache.org/nutch/SimilarityScoringFilter-->
<property>
- <name>scoring.similarity.model.path</name>
+ <name>cosine.goldstandard.file</name>
<value>goldstandard.txt</value>
<description>Path to the gold standard file which contains all the
relevant text and terms,
--- End diff --
It is a goldstandard file now and the model containing the term vector gets
created at runtime. I felt the earlier name was misleading and given that we
could have more similarity metrics in the future, I renamed it.
Later on if we can incorporate the idea of a corpus of documents, we can
rename it to something like cosine.corpus.path ?
> Improved Tokenization for Similarity Scoring plugin
> ---------------------------------------------------
>
> Key: NUTCH-2209
> URL: https://issues.apache.org/jira/browse/NUTCH-2209
> Project: Nutch
> Issue Type: Improvement
> Components: scoring
> Reporter: Sujen Shah
> Assignee: Sujen Shah
> Labels: memex
> Fix For: 1.13
>
>
> This patch would add Lucene based tokenization to the cosine similarity
> plugin and clean up the code currently present.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
