linen.dev
linen.dev copied to clipboard
content score - Negative attributes
We need some sort of algorithm to surface good content.
- page content - should take in to account that there are threads within the page - If it is all threads it shouldn't count towards the score
- thread content
Potential attributes:
- Admin respond
- emoji reactions
- text length
- Number of replies
- Number of repliers
Negative attributes
- swearing
- bank account information
- spam like behavior (Maybe simple NLP libraries)
- bot messages - automated messages
- links that seem malicious
Potentially look at other search engine - page score algorithm