mumbleapi
mumbleapi copied to clipboard
divanov11
User Recommendation Enhancement
Is your feature request related to a problem? Please describe.
A clear and concise description of what the problem is.
- In the API endpoint of User recommender in the 'user' application, the users are being recommended based on their primary key that literally makes no sense.
Describe the solution you'd like
A clear and concise description of what you want to happen.
- A recommendation system can be used instead of this, which will be based on a machine learning model that recommends similar user based on certain user attributes.
- I want to work on this, if you are willing to have this feature in this project.
- I have worked on something similar previously that I can show you in the next discord meeting for better understanding of what I am trying to say
Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.
Additional context
Add any other context such as screenshots, schematics, about the feature request here.
Hi @patel-jay14144 !
Thanks for your suggestion. You actually beat me in making this issue! First off, we are returning for the first few users from the database for now. It's just a beta release after all.
Can you discuss how you are going to build a recommender system? Is it with the apriori algo?
It depends what kind of recommendation we want, I'll see if I get any article for types of recommendation systems and link that in the following comment so that we can discuss it further
I have made one recommendation system using K- Nearest Neighbour Algorithm, the choice of this algorithm was because of the situation
But we can definitely discuss and come out with a efficient algo based on what type of recommendations we want to have.
We have skills and interests, a feature that I am developing at the moment, that we can use as factors for the algorithm. We should use those as the base of recommending anything in the system.
Yes if we are maintaining those details about every user then using Apriori makes sense
like @abhivemp said, I'd maybe just start with randomly grabbing active users who have similar skills and interests? The mention of machine learning and Apriori sounds cool, but I'm worried that people new to mumble will be suppressed because the algorithm only promotes the "most well known and liked users" in the system.
