[Babase] MPI data testing
Lacey Maryott Roerish
lroerish4 at gmail.com
Fri Jan 8 14:40:58 EST 2010
On Fri, Jan 8, 2010 at 2:35 PM, Karl O. Pinc <kop at meme.com> wrote:
> On 01/08/2010 01:27:48 PM, Lacey Maryott Roerish wrote:
> > On Fri, Jan 8, 2010 at 2:22 PM, Karl O. Pinc <kop at meme.com> wrote:
> >
> > > On 01/08/2010 10:56:06 AM, Lacey Maryott Roerish wrote:
> > > > Karl,
> > > >
> > > > Susan and I did a little poking together and found some
> > > > problems.
> > > > One
> > > > is that the 'decided' column doesn't seem to be computed
> > correctly.
> > > > It
> > > > is
> > > > TRUE for everything... So that will need to be fixed.
> > >
> > > This is because MPIACTS.Decided is true for every row
> > > that ever appears as the first interaction in the
> > > multi-party interaction collection. This needs
> > > to be fixed in the data - you and I got it wrong
> > > somehow.
> > >
> > This doesn't make sense. I thought that 'decided' was only true if
> > the
> > interaction was initiated by a decided interaction (Only AS OS or
> > DS).
> > and
> > that for any MPI that was initiated by a VS or anything else, this
> > would be
> > false. is that what you are saying here and I am just dense?
>
> That logic is fine, but that's not what the MPIACTS.Decided
> column values say. The data needs to be fixed.
>
Right, the data in the DB is definitely wrong.. This isn't a value we
manually enter, it is computed. So will that just get fixed before we do
the real upload into babase, or should we empty and reload test once to make
sure the program is properly interpreting the data and calculating the
correct values?
> >
> > >
> > > I do see one problem. Do we have a rule that when the
> > > sequence number is one the MPIACTS code must all be
> > > the same? I think not, and maybe we don't care.
> > > But, if any one of the sequence number 1 MPIACTS codes
> > > have TRUE in the Decided column then the corresponding
> > > MPI_EVENTS.Decided column will also be true.
> > >
> > > Please think about if we want such a rule and if
> > > we want MPI_EVENTS.Decided to behave this way.
> > >
> > I guess I am really confused about the 'decided' information.
> > In terms of the MPIACTS codes matching for sequence 1... I Am also
> > confused.
> > If there are two lines sharing sequence one, the act would be the
> > same by
> > default since the program splits out those lines and not the person
> > entering
> > the data right? So the only way to have different acts in two
> > sequence 1
> > rows would be if the program screwed up... I think?
>
> The upmpi program is not the only way to get the data into the
> database. Data can be entered manually in ppa, or with SQL statements,
> or changed once entered.
>
Gotcha! It seems that we should definitely have the rule that lines sharing
a sequence number must have the same act. If it isn't the same thing
happening, it doesn't seem to warrant a shared sequence number. Good Call
:)
Lacey
>
> >
> > >
> > > > Also, (you may already know this but just in case) the links
> > > > built
> > > > into
> > > > the MPI_events documentation page (ie context_types) are broken.
> > They
> > > > currently don't go anywhere.
> > >
> > > Goes to the docs for the MPIS.Context_types column for me.
> > > Maybe you need to reload the page to clear the browser cache?
> > >
> >
> > Hmm went fine on my computer. I Wonder if this is another Mac
> > thing.
> > Susan, want to re-try on your computer?
>
> It's a browser thing.
>
> Karl <kop at meme.com>
> Free Software: "You don't pay back, you pay forward."
> -- Robert A. Heinlein
>
>
> _______________________________________________
> Babase mailing list
> Babase at www.eco.princeton.edu
> http://www.eco.princeton.edu/mailman/listinfo/babase
>
--
- -
Lacey K. Maryott Roerish
Alberts Lab
Department of Biology
Duke University
ph: 919-660-7306
fax: 919-660-7293
Lacey.Maryott at duke.edu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://eeblistserv.Princeton.EDU/pipermail/babase/attachments/20100108/e43df5a6/attachment-0001.html
More information about the Babase
mailing list