| Election type | Approval |
|---|---|
| Culture | Euclidean 2D |
| Candidates | [4-15] |
| Voters | [20-600] |
| Preflib data: 246 profiles with at most 25 candidates and 2000 voters | |
| Instances | 315500 |
| Each plot is obtained from 315500 profiles from various sources | |
| Parameters | [0,1]x[0,1] square with three points placed uniformly at random, describing three districs. Each district is a center of a Gaussian distribution (with std. dev. 0.2 in both dimensions). “For each district, we then sample a number of points representing voters and alternatives according to this distribution.” Each voter approves all candidates within distance 0.4 |
| Notes | The conference paper contains very little details about the experiment. The data was taken from the full version (the conference paper mentions using the Mallows model, but this does not appear in the full version) |
| Motivation | - intro talks about different settings, e.g., search engines, but experiments don’t really pick this up. |