IBM Watson evaluation - is training data uploaded/saved to IBM servers?

2
votes

Reading the documentation at https://github.com/watson-developer-cloud/conversation-enhanced#local it states "Ensure that you have a Bluemix account. While you can do part of this deployment locally, you must still use Bluemix.". If I use a custom dataset will any of this data be sent to IBM's servers ?

Specifically I think im referring to step circled red ('Create or import training data') : enter image description here I plan to build a similar app to conversation-enhanced (https://github.com/watson-developer-cloud/conversation-enhanced) but want to ensure local data will not be sent to IBM.

1
ibm-cloudibm-watsonwatson-conversation
Both Conversation and Retrieve and rank require to upload training data. – Simon O'Doherty
@SimonO'Doherty uploading training data equates to uploading the data to IBM servers, even if deploying watson locally ? – blue-sky
There is no "local Watson" except Watson explorer. All the services require to see your data in order to train. – Simon O'Doherty
@SimonO'Doherty github.com/watson-developer-cloud/conversation-enhanced contains 'building locally' and 'running locally' sections but training data does not remain on local machine as part of training ? – blue-sky
I haven't used that demo, but both R&R and Conversation will require training data to be uploaded in order to work. – Simon O'Doherty

1 Answers

2
votes

Yes, you will need to send data to IBM.

If you are planning on using the IBM BlueMix services, then I recommend you read the terms of use, found here, to understand your full agreement:

http://www-03.ibm.com/software/sla/sladb.nsf/pdf/6606-08/$file/i126-6606-08_05-2016_en_US.pdf