Arguments Score = Sum of [Sub argument scores] * their [Linkage Scores] * their [Unique Scores]
myklob opened this issue · 0 comments
We should track each ideas score over time, similar to how we track a stock's performance.
Arguments will have scores, depending on the quantity and quality of sub-arguments that strengthen or weaken each argument.
Ultimately a conclusion's strength depends on the relative strength of supporting arguments and evidence.
I propose the following steps to determining an argument's score, and then playing with the features until we get the algorithm "right".
- Adding the number of pro arguments
- Subtracting the number of con arguments
Separation Distance, SD = 1
- Adding the scores of reasons that agree
- Subtracting the scores of reasons to disagree
Separation Distance, SD = 2
- Adding the score of reasons to agree with reasons to agree
- Subtracting the scores of reasons to disagree with reasons to disagree
- Adding the scores of reasons to disagree with reasons to disagree
- Subtracting the scores of reasons to agree with reasons to disagree.
Each sub-argument score would be multiplied by it's linkage and unique score, before being counted towards the conclusion score.
A multiplier could be used to reduce the strength of arguments the greater their separation distance, SD.