Pabulib: A Participatory Budgeting Library
aa r X i v : . [ c s . D C ] D ec Pabulib: A Participatory Budgeting Library ∗ Dariusz Stolicki
Jagiellonian University in Krak´ow [email protected]
Stanis law Szufa
Jagiellonian University in Krak´ow [email protected]
Nimrod Talmon
Ben-Gurion University [email protected]
December 14, 2020
Abstract
We describe the
PArticipatory BUdgeting LIBrary website (in short,
Pabulib ), which can be accessed via http://pabulib.org/ , and which isa library of participatory budgeting data. In particular, we describe thefile format ( .pb ) that is used for instances of participatory budgeting.
Since it was initiated by the Brazil workers’ party [7] in the 90s, Participatorybudgeting (PB) [3] has been gaining increased attention all over the world.Essentially, the idea behind PB is a direct democracy approach in which theway to utilize a common budget (most usually a municipality budget) is beingdecided upon by the stakeholders themselves (most usually city residents). Inparticular, given a set of proposed projects with their costs, and a designatedtotal budget to be used, voters express their preferences over the projects andthen an aggregation method takes the votes and decides upon a subset of theprojects to be implemented.As research on PB from the perspective of computational social choice isaccordingly increasing (see, e.g., the survey of Aziz and Shah [2]; as well assome specific recent papers on PB [6, 5, 4, 1]), there is a need to have publicly-available datasets; this is the goal behind the
PArticipatory BUdgeting LIBrary (in short,
Pabulib ), that is available in http://pabulib.org .The main aim of this document is to define a data format that is used inPabulib. ∗ Please cite this paper when using Pabulib. The .pb
File Format
The data concerning one instance of participatory budgeting is to be stored ina single UTF-8 text file with the extension .pb . The content of the file is to bedivided into three sections: • META section with general metadata like the country, budget, numberof votes. • PROJECTS section with projects costs and possibly some other meta-data regarding projects like category, target etc. • VOTES section with votes, that can be in one of the four types: approval,ordinal, cumulative, scoring; and optionally with metadata regarding vot-ers like age, sex etc.
METAkey; valuedescription; Municipal PB in Wieliczkacountry; Polandunit; Wieliczkainstance; 2020num_projects; 5num_votes; 10budget; 2500rule; greedyvote_type; approvalmin_length; 1max_length; 3PROJECTSproject_id; cost; category1; 600; culture, education2; 800; sport4; 1400; culture5; 1000; health, sport7; 1200; educationVOTESvoter_id; age; sex; vote1; 34; f; 1,2,42; 51; m; 1,23; 23; m; 2,4,54; 19; f; 5,75; 62; f; 1,4,76; 54; m; 1,77; 49; m; 58; 27; f; 49; 39; f; 2,4,510; 44; m; 4,5 Detailed Description
The bold part is obligatory. • key ∗ description ∗ country ∗ unit – name of the municipality, region, organization, etc., holdingthe PB process ∗ subunit – name of the sub-jurisdiction or category within which thepreferences are aggregated and funds are allocated– Example : in Paris, there are 21 PBs – a city-wide budgets and 20district-wide budgets. For the city-wide budget, unit is Paris,and subunit is undefined, while for the district-wide budgets, unit is also Paris, and subunit is the name of the district (e.g.,IIIe arrondissement).–
Example : before 2019, in Warsaw there have been district-wideand neighborhood-wide PBs. For all of them, unit is Warsaw,while subunit is the name of the district for district-wide bud-gets, and the name of the neighborhood for neighborhood-widebudgets. To associate neighborhoods with districts (if desired),an additional property district can be used.–
Example : assume that in a given city, there are distinct PBs foreach of n > unit is the city name, while subunit is the name of the category. ∗ instance – a unique identifier of the specific edition of the PB process(year, edition number, etc.) used by the organizers to identify thatedition; note that instance will not necessarily correspond to theyear in which the vote is actually held, as some organizers identifythe edition by the fiscal year in which the PB projects are to becarried out ∗ num projects ∗ num votes ∗ budget – the total amount of funds to be allocated ∗ vote type – approval – each vote is a vector of Boolean values, v ∈ B | P | ,where P is the set of all projects,– ordinal – each vote is a permutation of a subset of P suchthat | P | ∈ [ min length , max length ], corresponding to a strictpreference ordering,– cumulative – each vote is a vector v ∈ R | P | + such that k v k ≤ max sum points ∈ R + , 3 scoring – each vote is a vector v ∈ I | P | , where I ⊆ R . ∗ rule – greedy – projects are ordered decreasingly by the value of theaggregation function (i.e., the total score), and are funded untilfunds are exhausted or there are no more projects– other rules will be defined in future versions ∗ date begin – the date on which voting starts ∗ date end – the date on which voting ends ∗ language – language of the description texts (i.e., full project names) ∗ edition ∗ district ∗ comment ∗ if vote type = approval :– min length [default: 1]– max length [default: num projects]– min sum cost [default: 0]– max sum cost [default: ∞ ] ∗ if vote type = ordinal :– min length [default: 1]– max length [default: num projects]– scoring fn [default: Borda] ∗ if vote type = cumulative :– min length [default: 1]– max length [default: num projects]– min points [default: 0]– max points [default: max sum points]– min sum points [default: 0]– max sum points ∗ if vote type = scoring :– min length [default: 1]– max length [default: num projects]– min points [default: −∞ ]– max points [default: ∞ ]– default score [default: 0] ∗ non-standard fields • value .2 Section 2: PROJECTS • project id • cost • name – full project name • category – for example: education, sport, health, culture, environmentalprotection, public space, public transit and roads • target – for example: adults, seniors, children, youth, people with dis-abilities, families with children, animals • non-standard fields • voter id • age • sex • voting method (e.g., paper, Internet, mail) • if vote type = approval : ∗ vote – ids of the approved projects, separated by commas. • if vote type = ordinal : ∗ vote – ids of the selected projects, from the most preferred one tothe least preferred one, separated by commas. • if vote type = cumulative : ∗ vote – project ids, in the decreasing order induced by points , sep-arated by commas; projects not listed are assumed to be awarded 0points. ∗ points – points assigned to the selected projects, listed in the sameorder as project ids in vote . • if vote type = scoring : ∗ vote – project ids, in the decreasing order induced by points , sep-arated by commas; projects not listed are assumed to be awarded default score points. ∗ points – points assigned to the selected projects, listed in the sameorder as project ids in vote . • non-standard fields Outlook
We have introduced the PArticipatory BUdgeting LIBrary (Pabulib; availableat http://pabulib.org ), and have described the .pb file format that is usedin it.We hope that Pabulib will foster meaningful research on PB, in particu-larly helping the computational social choice community offer better aggregationmethods to be used in real-world instances of PB.
Acknowledgement
Nimrod Talmon has been supported by the Israel Science Foundation (grant No.630/19). Dariusz Stolicki and Stanis law Szufa have been supported under thePolish Ministry of Science and Higher Education grant no. 0395/DLG/2018/10.
References [1] Haris Aziz, Barton Lee, and Nimrod Talmon. Proportionally represen-tative participatory budgeting: Axioms and algorithms. arXiv preprintarXiv:1711.08226 , 2017.[2] Haris Aziz and Nisarg Shah. Participatory budgeting: Models and ap-proaches.
CoRR , abs/2003.00606, 2020.[3] Y. Cabannes. Participatory budgeting: a significant contribution to partic-ipatory democracy.
Environment and urbanization , 16(1):27–46, 2004.[4] A. Goel, A. K. Krishnaswamy, S. Sakshuwong, and T. Aitamurto. Knapsackvoting.
Collective Intelligence , 2015.[5] Pallavi Jain, Krzysztof Sornat, and Nimrod Talmon. Participatory budget-ing with project interactions. In
Proceedings of IJCAI’20 , pages 386–392,2020.[6] Nimrod Talmon and Piotr Faliszewski. A framework for approval-basedbudgeting methods. In
Proceedings of AAAI’19 , pages 2181–2188, 2019.[7] H. Wainwright. Making a people’s budget in Porto Alegre.