Open Brend-Smits opened 4 years ago
In GitLab by @Brend-Smits on Mar 29, 2020, 19:55
changed the description
Currently we are working on a implementation where we use collaborative filtering. Similar users are found by comparing project likes. It is recommended to find out if the user is not interested in the project so it can be excluded from recommendations. Also for the cases when collaborative filtering does not work it is recommended to either look into content based filtering using tags / categories. This could also be done by tracking how popular a project is with project views or project likes. There is no hard evidence yet what the best approach is and these are just ideas on how to solve it.
More can be found in research below:
https://docs.google.com/document/d/1B1u_j5jQXmv3GVPG-HrXYJLR8rVKbtAhEkLjbac6KSs/edit?usp=sharing
If we would like to track the amount of times a project is viewed: It is important to not show who viewed the project as this could result into a GDPR breach. https://docs.google.com/document/d/1Sv6OTqp-wov658a2oc2bznVzR4W6lZH1iAauVBEJRVc/edit?usp=sharing
In GitLab by @Brend-Smits on Mar 29, 2020, 19:52
What ways are applicable to show interesting projects from the community on top of the search results or recommendations (page views, likes, forks, contributions as an example) How do we gather this data?
Related Epic: https://git.fhict.nl/dex/dex-epics/-/issues/3