ibm bluemix - IBM Watson Knowledge Studio 2.0 - deploying a rule-based model is experimental. What does that mean? -
when building rule-based model in watson knowledge studio, reminder appears saying:-
the rule-based model build here can deployed , used other watson services experimental purposes only.
in online documentation, @ beginning of chapter "deploying rule-based annotator ibm watson discovery", similar mention appears:-
attention: experimental feature of service.
i started building application based on wks custom rule-based model deployed in discovery , want sure solution not jeopardized @ point in future release of either wks or discovery.
what experimental means in context? there chance in future version of wks or discovery rule-based modeling won't available anymore?
i'm offering manager wks.
an example of official definition of experimental/beta ibm can found here: http://www.ibm.com/software/sla/sladb.nsf/pdf/6605-11/$file/i126-6605-11_06-2017_en_us.pdf
my interpretation of official ibm policy follows:
- experimental/beta provides no warranty
- experimental/beta not guarantee performance , not suitable production
- experimental/beta not provide migration ga
- experimental/beta not save client data @ service/feature termination - client has save whatever data want retain
- experimental/beta can terminate without prior notice
- any feedback experimental/beta client can used ibm , third parties
in context of wks, decided keep rules feature experimental time being because still experimenting around how combine rules machine-learning models. (currently, separate, decreases value wks users.) rule-based technology stable whole , can assure not going away. want improve ux around it. we'll make our best effort protect client investments in rule-based model, although reserve right introduce breaking changes if have to. (hence, "experimental" state of feature.)
there 2 aspects of using rules in wks:
- as pre-annotator machine-learning model annotation
- as stand-alone (deployable natural language understanding , watson discovery services) rule-based model.
because rules still in experimental state, benefit using in production @ no charge when deploy rules models paid bluemix services.
i encourage continue using rules in wks , provide feedback on ways improve feature. mentioned, rules here stay , working on improving them.
thanks question , using wks.
kind regards,
stefan
Comments
Post a Comment