Proportional Participatory Budgeting with Additive Utilities
Authors: Grzegorz Pierczyński, Piotr Skowron, Dominik Peters
NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In this section we evaluate different voting rules on data from real-world participatory budgeting elections carried out in several major cities in Poland. The data we use is taken from Pabulib and was collected by [Stolicki et al., 2020]. In our analysis we evaluated the following metrics: Total utility (UTIL). The total utility of the voters from the selected set W: P. Distribution of projects (PROJ-DIS). For each election instance we look at the projects selected from each district. We compute their cost and divide it by the fraction of the budget that is proportional to the population of the district. From those ratios we take a variance. Distribution of utilities (UTIL-DIS). For each election instance and each voter i we compute her normalised utility from the set of selected projects W, which we define as P c2W ui(c) divided by n P c2W ui(c). |
| Researcher Affiliation | Academia | Dominik Peters University of Toronto Toronto, ON, Canada dominik@cs.toronto.edu Grzegorz Pierczyński University of Warsaw Warsaw, Poland g.pierczynski@mimuw.edu.pl Piotr Skowron University of Warsaw Warsaw, Poland p.skowron@mimuw.edu.pl |
| Pseudocode | Yes | Definition 1 (Method of Equal Shares (MES)). Each voter is initially given an equal fraction of the budget, i.e., each voter is given 1/n dollars. We start with an empty outcome W = ; and sequentially add candidates to W. To add a candidate c to W, we need the voters to pay for c. Write pi(c) for the amount that voter i pays for c; we will need that Pi2N pi(c) = cost(c). We write pi(W) = P c2W pi(c) 1 n for the total amount i has paid so far. For 0, we say that a candidate c 62 W is -affordable if ... and Definition 11 (Greedy Cohesive Rule (GCR)). The Greedy Cohesive Rule (GCR) is defined sequentially as follows: we start with an empty outcome W = ;. At each step, we search for a weakly (β, T)-cohesive group S. If such a group exists, we find one where β 1 is maximum,7 add all the candidates from T to W, remove all voters in S from the election and repeat the search. If no such group exists, we stop and return W. |
| Open Source Code | No | The paper does not provide a specific link or explicit statement about the availability of the source code for the described methodology. |
| Open Datasets | Yes | The data we use is taken from Pabulib and was collected by [Stolicki et al., 2020]. The data is publically available at pabulib.org. |
| Dataset Splits | No | The paper does not specify exact training, validation, or test dataset splits (e.g., percentages or absolute counts). |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., exact GPU/CPU models or computer specifications) used for running experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details, such as library names with version numbers. |
| Experiment Setup | No | The paper describes the types of voter preferences and evaluation metrics but does not provide specific experimental setup details such as hyperparameter values or training configurations. |