[ 
https://issues.apache.org/jira/browse/MADLIB-1199?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Frank McQuillan updated MADLIB-1199:
------------------------------------
    Description: 
As a follow up of https://issues.apache.org/jira/browse/MADLIB-1160

We found the current lda online documentation is not clear enough - more 
documentation may need to be done.

Besides, some of lda related functions may have confusing name or interface. - 
we are going to review the lda code and do some refactoring and rename. There 
might be some function interface change in this phase.

Possible improvements:

1) The parameter `voc_size` should be inferred and we don't need to ask the 
user to provide it as an in put param to the training function.

2) Currently all params in the training function are mandatory.  Does it make 
sense to add optional params with default values. e.g., for alpha and beta?

 

  was:
As a follow up of https://issues.apache.org/jira/browse/MADLIB-1160

We found the current lda online documentation is not clear enough - more 
documentation may need to be done.

Besides, some of lda related functions may have confusing name or interface. - 
we are going to review the lda code and do some refactoring and rename. There 
might be some function interface change in this phase.


> Usability changes for LDA - phase 2
> -----------------------------------
>
>                 Key: MADLIB-1199
>                 URL: https://issues.apache.org/jira/browse/MADLIB-1199
>             Project: Apache MADlib
>          Issue Type: Improvement
>          Components: Module: Utilities
>            Reporter: Jingyi Mei
>            Priority: Minor
>             Fix For: v2.0
>
>
> As a follow up of https://issues.apache.org/jira/browse/MADLIB-1160
> We found the current lda online documentation is not clear enough - more 
> documentation may need to be done.
> Besides, some of lda related functions may have confusing name or interface. 
> - we are going to review the lda code and do some refactoring and rename. 
> There might be some function interface change in this phase.
> Possible improvements:
> 1) The parameter `voc_size` should be inferred and we don't need to ask the 
> user to provide it as an in put param to the training function.
> 2) Currently all params in the training function are mandatory.  Does it make 
> sense to add optional params with default values. e.g., for alpha and beta?
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to