Process security after deployment

vvella

Hi,

we just started investigating Rapidminer for a possible deployment at our client - on premise setup. The ML models are our IP, so obviously we prefer if the models (processes) are hidden to the client and just accessible via a web service (pretty much a black box). Does rapidminer server provide such facility? (for example client administrator is still able to monitor the environment from rapidminer server, however the process and model details remain hidden). 

Pavithra_Rao

edin_Hi @vvella,

 

The underlying model is secured as the model results are consumed/made available to as Web services REST APIs using RapidMiner Server.

But if you intend to deploy this process at our client - on-premise setup and wish to hide the process to the client and just make it accessible via a web service (pretty much a black box), then this is not something feasible at this point.

 

another possibility is, you could build out the solution as an extension to hide/black box the process.

 

I would defer it to our R&D Data scientist @Edin_Klapic and Product Manager @jpuente to share their opinion here.

 

Cheers,

 

jpuente

Hi,

In RapidMiner Server, you can publish web services associated to the processes (see https://docs.rapidminer.com/latest/server/advanced-topics/qlik-integration/expose-process.html). If you create a user and add it only to the "report viewer" role, he'll be able to call the WS (if you set the permissions of that process to read/execute for him), but he won't be able to see the process, connect through Studio or anything else. To him, it will be only a web page he can access. This is, by the way, ideal if you have a portal for the users and you just embed the results from RapidMiner there. 

 

I hope that helps.

Jesús

 

Edin_Klapic

Hi @vvella,

 

More information about User rights and Permissions can be found on the following pages on docs.rapidminer.com:

• User Management
• Assign access rights
• Predefined groups

 

Happy Mining,

Edin