Seoul's Machine-Learning Pipeline for Screening Participatory Budgeting Proposals
South Korea
A peer-reviewed study applied topic modelling and a Random Forest classifier to 21,833 real proposals from Seoul's participatory budget (2013–2021), …
United Kingdom · Edinburgh · See the United Kingdom profile · See the Edinburgh profile
Evidence: Observational / pre–post Top 6% 87/100 · Ask Evidence Copilot about this practice
Under a Scottish Government–COSLA framework requiring at least 1% of council budgets to go through participatory budgeting, researchers evaluated NLP tools on the Consul platform across 32 local authorities; by 2023 councils had exceeded the target, channelling £154m through PB.
In 2017 Scotland's local authorities and the Scottish Government agreed a framework committing at least 1% of local government budgets to participatory budgeting (PB), reaffirmed in a revised 2021 Framework Agreement between the Scottish Government and COSLA (Convention of Scottish Local Authorities).
To help PB scale consistently across all 32 Scottish local authorities at once, a research team from the University of Warwick, the Alan Turing Institute and the University of Oxford evaluated NLP tools on the Consul digital platform used for mainstream Scottish PB.
The team used a qualitative longitudinal design combining interviews, process observation and platform-data analysis across live PB study sites to assess whether NLP could ease the scaling challenges of running PB nationwide.
By the milestone reported in 2023, more than 110,000 people had taken part and directly decided how £154 million of council budgets were spent, reaching 1.4% of eligible budgets — above the original 1% target — across all 32 local authorities. The researchers' initial results suggest scaling challenges 'may be tackled using NLP technology,' which had previously shown improved citizen-participation effectiveness in a controlled use-case evaluation.
The NLP findings are explicitly framed by the authors as preliminary; the strongest, independently verified evidence concerns the PB programme's national scale and mainstreaming (COSLA's published spend and participation figures), while the AI layer's effectiveness is still being established through ongoing academic study rather than a completed government evaluation.
National / regional programmes
Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.
Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.
Where this practice's information was retrieved from, and when.
South Korea
A peer-reviewed study applied topic modelling and a Random Forest classifier to 21,833 real proposals from Seoul's participatory budget (2013–2021), …
Switzerland
Aarau's first city-wide participatory budget (2023-24) used the Method of Equal Shares to allocate CHF 50,000, funding 17 of 33 …
Poland
Wieliczka's 2023 'Green Million' pilot allocated a PLN 1,000,000 citizen budget using the Method of Equal Shares, a proportional algorithm …
Spain
Warwick, QMUL and Alan Turing Institute researchers built four NLP/ML tools — recommendation, interest grouping, summarisation, aggregation — for Madrid's …
Open full copilot Grounded in cited practices — always check the sources.