General Idea:
By using the existing data we can attach an arbitrary trust or reputation score to an address. This serves two primary purposes. “Positive” or high reputation accounts could be recommended to readers, feature on the site or gain increased visibility with other measures since the system knows that its likely quality content.
On the other hand submissions from “negative” or low reputation accounts could be lower ranked, not displayed or completely removed automatically.
The optimal composition of this score requires a fair amount of fine tuning and data processing.
Early stage logic present in the backend. Such as caluclating and rating tips based on the value, tags and number of tips in this category
Scope
- Review module in the backend that processes the available information and stores the result in a database (3 days)
- Add the reputation in the scored tip list and blacklist (2 days)
- Build a debugging system to verify the health of the system and detect possible abuses (1 week)
- Change the formula for recommendation of tips (1 day)
- Add tip comments in forumula (TDB)
Set of parameters for the formula
- Quantity and length of .chain names attached to the public key
- How much did the public key spend?
- How much spend on average per tip
- How much did the public key receive?
- How much receive on average per tip
- Weights about # of comments to tip
Future reputation parameters
- Prevent fraudulent use building a transaction graph and detecting circles
- Retips send
- Retips received
- Comments created on own tips
- Comments created on other’s tips
- Check network interactions and score other addresses based on reputation, similar to page rank
- Show feed entries to the user with recommended tips
- Set the user’s reputation based on user’s tips
Proposed Team
Time Estimations
- Building the general scope, 2 weeks with absolute minimum 1 week
- First reputation parameters 1 week
- Future reputation parameters - TBD after prioritization