[OTDev] Superservice section of API documentation
Christoph Helma helma at in-silico.chMon Jan 31 15:09:03 CET 2011
- Previous message: [OTDev] Superservice section of API documentation
- Next message: [OTDev] Scheduled downtime of AA services on Thu
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Dear Roman, The supermodel section is currently a mess, because we do not have a common agreement yet. I will summarize email discussions and update the sextion until next week. Best regards, Christoph > Dear Christoph > > The Supermodel section of the documentation of the Model API makes things a > bit more confusing. > The first sentence says that a Supermodel is "A superservice for applying > complex models". A Superservice, however, is an algorithm (given that > Superservice is said to be an algorithm type in the documentation of the > algorithm API). > > Maybe we could rename the "Superservice" algorithm type to "Superalgorithm". > Then, both Supermodel and Superalgorithm are Superservices belonging > together (where a Supermodel is created by a Superalgorithm). > > Best regards, > Roman > > On Mon, Jan 31, 2011 at 1:45 PM, Roman Affentranger < > roman at douglasconnect.com> wrote: > > > Dear Christoph > > > > Few comments regarding the superservice documentation. > > 1) "Super service" is a part of "5. Algorithm Types", but the title has a > > different color than the other algorithm types. > > 2) Under the REST operations, there are links for Parameters, Result and > > Status Codes for GET ("as in Algorithm"). These links point to > > opentox.org/dev/Algorithm, which doesn't exist... > > 3) I find the "introduction" sentence a bit confusing: > > > > ===== > > Super Service > > Allows to create models and specify descriptor calculation , feature > > selection and possibly other services, necessary to generate the data, > > required by the model building algorithm > > A superservice for creating models is a specific instance of an algorithm, > > with input parameters defined below. The result of a super algorithm is a > > super model (see Model API). > > ===== > > > > I think this could be a bit more clear, e.g. > > ===== > > Superservice > > A Superservice is a specific instance of an algorithm that, like a regular > > algorithm, creates a model, but additionally allows for specific definition > > of the individual services (descriptor calculation, feature selection, and > > possibly other services) necessary to generate the data required by the > > model building algorithm. The result of a Superservice is a Supermodel (see > > Model API) > > ===== > > > > I don't know if the above sentences are even correct, but I think something > > like that would be more clear. In particular: > > - If you say "a superservice for creating models is a specific instance of > > an algorithm", this gives the impression that there are Superservices wih > > another purpose than creating models. Is that true? If so, it could be made > > more clear. If not, I'd suggest rephrasing as I suggested above. > > - The last sentence "The result of a super algorithm is a super model (see > > Model API)" should be rephrased. The term "super algorithm" appears out of > > nowhere. What's the difference between a Superservice and a Superalgorithm? > > - "(see Model API)" should be a link to the Model API. > > > > So, overall, I think it is fairly clear, however, could be made more clear > > ;) > > In particular, the terminology should be revisited (Superservice vs > > Superalgorithm. If both, both need to be defined). Also, the spelling should > > be unified (Super Service, super service, Superservice, etc.). That is: one > > word, two words? Capital letters/lower-case letters? Same for Supermodel, > > Superalgorithm (if we keep it). > > > > Since it clearly states that the output of a Superservice/Superalgorithm is > > a Supermodel, I'll look at that next.... > > > > Best regards, > > Roman > >
- Previous message: [OTDev] Superservice section of API documentation
- Next message: [OTDev] Scheduled downtime of AA services on Thu
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Development mailing list