Pre Merge pull request !249 from ZichengQu/datacompliance-meetings_1

sigs/datacompliance/meetings/003-20220420.md

@@ -0,0 +1,73 @@
+# Tuesday, April 19, 2022, at 20:30pm GMT+8
+
+## Meeting
+
+* Meeting link: https://welink.zhumu.com/j/635815200
+* Meeting time: Tuesday April 19, 2022 at 20:30pm GMT+8
+
+## Attendees
+
+* Li Zi, Gopi Krishnan Rajbahadur, Zicheng Qu, Liusong, Skyzc
+
+## Agenda
+
+* 1 . Issue about SPDX AI BOM and our improvements.
+* 2 . Next step to TAC Meeting
+* 3 . A report to CAICT, our contribution
+* 4 . Progress of toolchain
+* 5 . Sandbox process
+
+## Notes
+
+* 1 . LF-AI has four critical stages of a project, including "Sandbox Stage", "Incubation Stage", "Graduation Stage", "Archived Stage". We aim to sandbox this project, not incubate it.
+* 2 . If CBG wants to use the dataset license compliance, they should provide us with a workgroup like coach, experts, or lawyers, and we can give them some insights to help them, but we cannot do their work and are not responsible for maintaining their works.
+* 3 . There are four projects from the LF-AI landscape that are very close to our project, including "Datapractices.org", "OpenBytes", "OpenLineage", and "OpenDS4All" (Open data signs for all projects).
+
+> "Datapractices.org" and "OpenDS4All" are educational in nature, focusing on how to do data practices and how to open datasets.
+> "Datapractices.org": Tells how to best handle datasets, or how to collect datasets, set, ethics, and even provenance and lineage, but no content regarding compliance.
+> "OpenDS4All": A course taught in a UK college, they have material around overview foundation data practices and the whole pipeline.
+We can contribute to these projects, not compete, and enhance the projects as an LF-AI initiative.
+> "OpenBytes": This project has potentially the most overlap with our project.  Their goal is to enable sharing of open datasets, and how to collaborate data, and facilitate data exchange in companies, organizations, or researchers by the creation of data standards and formats. The difference between this project and our project is that we do dataset license compliance. They do how to create new licenses or create custom licenses and standards. We can contribute to them by helping them build their license and dataset sharing through our enhanced Montreal data license schema and data license analyzer tools.
+> "OpenLineage": A project for collecting license's metadata and lineage, open standard and very extendable. This standard can be used to document the lineage for dataset creation. We can use this standard to collect our lineage if needed.
+
+## TODO
+
+Gopi responsible:
+
+* 1 . Gopi will share with us the slides "Dataset License Compliance - A report for sandboxing our project to LF-AI".
+* 2 . After sandboxing, Gopi may spend a month or two writing some research papers about issues and how we are doing SPDX stuff and everything relevant.
+* 3 . Gopi will summarize a list of licenses for the 25 datasets, and send it to Clement first to review, then Gopi will send it to lawyers and ask them to decompose it to rights, limitations, and obligations.
+* 4 . Gopi attends AI BOM meeting and synchronizes information with Clement and Zev in time.
+
+LiZi (Clement) responsible:
+
+* 1 . Clement will make an initial draft to allocate people in the following areas, including requirements for sandbox, establishing governance structure (also for regulations and standards), making the GitHub repo, making core infrastructure best practice, making license file for each repo, making two factor auth on GitHub, identifying the security issues team, submitting a completed project contribution, LF-AI related activities, going through depth of the four projects above, etc.
+* 2 . Next week, Clement will join a CAICT meeting to report the dataset compliance and give a summary and feedback to Gopi.
+* 3 . Clement will let some designers design some icons to demonstrate the risk of the dataset used in the models of MindSpore.
+
+Zicheng (Zev) responsible:
+
+* 1 . Once Zev has any fields that want to contribute to the AI BOM, Gopi can push them to the AI BOM after internal discussion and agreement.
+
+Song responsible:
+
+* 1 . Continue to improve the front-end display of the license uploading webpage.
+
+Zhengcai responsible:
+
+* 1 . After the questions are completed, only licenses in DataLicense can be displayed on the webpage.
+
+Others:
+
+* 1 . Make our project ready for the sandbox.
+* 2 . Work on the proposal.
+* 3 . Talk to the attack about the governance structure.
+* 4 . Go through the depth of the four projects above, whatever contributions and collaborations can be to these projects, and what we can potentially expect from these projects.
+* 5 . Apply and ask Zhipeng to help push our project to the sandbox in May.
+* 6 . After meeting all the requirements of the sandbox, we can have a strategy to interact with the four projects.
+* 7 . After sandboxing, we can decide on a strategy for how we are going to contribute to these different projects.
+
+## Other resources shared in the meeting
+
+* 1 . [1chipML Governance](https://github.com/1chipML/1chipML/blob/main//GOVERNANCE.md)
+* 2 . [LF-AI TSC template](https://github.com/lfai/tsc-template)