Attendees:
Google/OpenLMIS Artificial Intelligence Collaboration Background
Sandya is on Google's machine learning team- their team looks into stockouts, demand forecasting
Came across Openlmis – wants to build machine learning solutions on top of OpenLMIS’s data
Wants to be able to publish papers and present results on how effective this solution can be. They would contribute this to OpenLMIS for free
We are in the exploratory phase to see if/how this can work. Licensing issues regarding the software and the data are being worked out
- This wouldn’t actually be making modifications to the OpenLMIS codebase; they would be building a separate tool
- people should be able to pick up the tool and make it work for themselves
- Michael Downey is providing technical/legal support to this initiative on this
Summary of legal issues we are looking into:
- Google will not need to do any coding under our AGPL license, rather they would just need to use the data
- Google would do work under the Apache2 license
- Michael mentioned that there are few licenses that are often used for access to data: OpenData, Creative Commons, among others. Not sure what licenses Google prefers for this type of work.
- Is it possible to anonymize some of the individuals before licensing the data? Our data is at the facility level and doesnt have names or individual patient information, but some of it might need to be scrubbed
- If Google did want to make any code contributions (core or as a new service), would it be possible to have a specific license with them to get around the AGPL issues?
Next Steps:
- Set up another meeting in early April- Wesley Brown will send out a google poll
- Sandyha to get someone from the Google legal team. Hilary is their legal person
- Michael to send an email to one of his colleagues to learn from them re: these issues, to triangulate opinions. They are hoping to hear experience from previous projects to lean on