[
https://issues.apache.org/jira/browse/SPARK-4036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15083604#comment-15083604
]
Joseph K. Bradley commented on SPARK-4036:
------------------------------------------
[~hujiayin] Thanks very much for your work on this, but I think we need to
discuss this more before putting it into MLlib. The primary reasons are:
* We have limited review bandwidth, and we need to focus on non-feature items
currently (API improvements and completeness, bugs, etc.).
* For a big new feature like this, we would need to do a proper design document
and discussion before a PR. CRFs in particular are a very broad field, so it
would be important to discuss scope and generality (linear vs general CRFs,
applications such as NLP, vision, etc., or even a more general graphical model
framework).
In the meantime, I'd recommend you create a Spark package based on your work.
That will let users take advantage of it, and you can encourage them to post
feedback on the package site or here to continue the discussion.
I'd like to close this JIRA for now, but I'll continue to watch the discussion
on it.
> Add Conditional Random Fields (CRF) algorithm to Spark MLlib
> ------------------------------------------------------------
>
> Key: SPARK-4036
> URL: https://issues.apache.org/jira/browse/SPARK-4036
> Project: Spark
> Issue Type: New Feature
> Components: MLlib
> Reporter: Guoqiang Li
> Assignee: Kai Sasaki
> Attachments: CRF_design.1.pdf, dig-hair-eye-train.model,
> features.hair-eye, sample-input, sample-output
>
>
> Conditional random fields (CRFs) are a class of statistical modelling method
> often applied in pattern recognition and machine learning, where they are
> used for structured prediction.
> The paper:
> http://www.seas.upenn.edu/~strctlrn/bib/PDF/crf.pdf
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
