On 2013-01-31, at 03:53, Ben Coman <[email protected]> wrote:

> 
> Camillo Bruni wrote:
>> I was just waiting for the google issue tracker to come back to life...
>> 
>> https://code.google.com/p/pharo/issues/detail?id=7373
>> 
>> => Slice in Inbox
>> 
>> the major problem I have now, is that wrong categorizations might easily 
>> spread :P, the sepll-fleu and the space - vi rus, are very common...
>> 
>>  
> 
> Can the auto-categorization mark such methods as "auto-categorized" ?  So 
> that later if it doesn't seem quite right, a human can make a better judgment 
> rather than assuming someone else had already made such a judgment for 
> some-good-reason.  In reality perhaps such re-categorization wouldn't happen 
> a lot, but "being at least able to" make such judgment increases confidence 
> in the system, which can be subtly undermined by automated tools (and anyway 
> I hate throwing away information that is easily available at the time).  This 
> wouldn't need any tool support straight away, but at least the data is 
> captured for later use.

the categorization will happen only upon user interaction.
hence I think such a control is not needed directly.
Feel free to extend the current classifier to support it, the code is
fairly simple..

Reply via email to