evidoria

← Back to browse

Good practice Imported

GovText — Singapore's whole-of-government NLP platform for policy text analytics

Singapore · Singapore · See the Singapore profile · See the Singapore profile

Evidence: Observational / pre–post Top 66% 53/100 · Ask Evidence Copilot about this practice

Singapore's whole-of-government NLP platform: topic modelling, text summarisation, Hansard Analysis and IM8 policy Q&A. Logged 1,680 users in six months (satisfaction 4.85/5); powers the Municipal Services Office chatbot. OECD OPSI-listed government innovation (2023).

1,680 users
Users logged in (6-month window)
4.85 out of 5
User satisfaction score

Details

Maturity
Established
Promoter
Government Technology Agency of Singapore (GovTech), Data Science and AI Division
Period
2019–present
Keywords
NLP, text analytics, parliamentary analysis, policy compliance, government productivity

Context

GovText is a Natural Language Processing (NLP) platform built by GovTech Singapore's Data Science and Artificial Intelligence Division (DSAID) and deployed as a standard component of the Singapore Government Tech Stack, giving public officers across all agencies whole-of-government (WOG) credentialed access to text-analytics tools without specialist data-science expertise.

Objectives

Enable public officers to analyse large volumes of unstructured text — parliamentary transcripts, citizen feedback, policy documents — without needing data-science skills, and to support common civil-service workflows such as drafting replies to Parliamentary Questions and querying ICT governance standards.

Activities

Core capabilities include topic modelling with interactive dashboards, extractive and abstractive text summarisation, sentiment analysis, feedback clustering, and custom NLP model development and hosting for agencies. Two government-specific modules — Hansard Analysis (parliamentary transcript search) and IM8 Q&A (natural-language querying of ICT governance standards) — address specific civil-service tasks, and GovText APIs integrate with agency CRM systems. The Municipal Services Office (MSO) integrated GovText's text classifier and key information extractor into its citizen-feedback CRM.

Results

Adoption data published on the Singapore Government Developer Portal and in the OECD Observatory of Public Sector Innovation (OPSI) case study (November 2024) show 1,680 users logged in over a six-month window with a satisfaction score of 4.85 out of 5. OECD OPSI nominated GovText as a public-sector innovation in 2023. No published data quantifies throughput volume, processing-time savings or cost reduction relative to baseline.

Conclusions

GovText shows credible cross-agency adoption and high user satisfaction as a shared government NLP utility, but the public record lacks quantified efficiency or cost-saving evidence, and the platform is designed exclusively for Singapore's WOG credential environment with no documented cross-country replication.

Implementation
Indicative cost
Medium (€50k–€500k) — Built and maintained by GovTech's Data Science and AI Division as a shared platform service across all government agencies; no published unit-cost or total-cost figures.
Time to results
Long (> 3 years) — Operating as a standard Government Tech Stack component since 2019, with continued OECD OPSI recognition through 2023–2024.
Staffing & skills
GovTech Data Science and Artificial Intelligence Division (DSAID) engineers, public officers across agencies as end users (no specialist data-science expertise required)

Conditions for success

  • whole-of-government (WOG) credential access enabling cross-agency use
  • API integration with agency CRM and back-end systems
  • a centralised platform maintained by a dedicated government AI/data-science unit

Where it fits

Governance type
centralised national government technology agency
Scale
whole-of-government (all Singapore public agencies)
Income level
high-income (Singapore)

Commonly funded by

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.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful