0x2E
[FR] AI integreation for scoring
This is merely a brainstorming for a feature request that needs more thought into it and is meant to be more of a draft of an idea rather than an implementation suggestion. I'd like the idea of an automatically calculated score for feed entries, but I don't know if this could be of any appeal for others too.
What
As a user I would like to have feed entries with an "interest score" based on previous upvotes or downvotes on other feed entries.
Why
Regardless of the number of feeds a user might subscribe to, articles can be often divided in high, neutral, or low interest for the end user. It would be beneficial to the end user to have entries of a feed skimmed beforehand.
(Optional) How
To start with I'd propose an integration with OpenAI API (with more providers in the feature maybe):
- when an entry is added to the feed it gets parsed by the model through the use of a well thought prompt (to be defined)
- the model gives back a list of tuples keyword + weight, where weight calculation is to be defined.
- the list of keywords and weights is compared to previous upvotes/downvotes and an interest score is calculated
- optionally entries with a high / low score are highlighted/hidden or otherwise clearly more/less noticeable
I'll just drop a little note here that I'm using this app precisely because there is no "AI" involved and nothing is parsed, calculated, or recommended to me automatically.
