Attendees:
Updates from contributors
-
Soham Banerjee
- Worked on obligations end-point.
- Created endpoints to list all obligations and obligation by ID.
- Had questions about IPRA, Email/URL/Author, Keyword and ECC endpoints.
- Will work on these endpoints after finishing the obligations endpoints.
- Will go through v2 documentations and contact on Slack if needed.
-
Abdelrahman jamal
- Experimented with different algorithms. Best one so far is with 0.16% error on 16,377 entities.
- More aggrisive optimization can achieve 0.03%
- Received new data from mentors last week, 26K unique rows. 0.5% error on new test data from best model.
- Most errors are due to wrong data annotation.
- 2 copyrights wrongly labled on copyrights from Linux kernel, taken from a paper.
-
Kavya Shukla
- Working on Audit for license changes. There is a separate table to hold the changes and separate to gerenrate the changelogs.
- Audit logs can be fetched using a GET query and changes can be fetched using the changelog id.
- Had question about candidate licenses. Mentors explained in brief, will continue the discussion in project specific call later.