NASS AI: Revolutionizing access to Nigeria’s legislative bills

In Nigeria, it is a herculean task to get access to legislative bills. It is typically the duty of the National Assembly of Nigeria to provide Nigerians with proposed bills, order papers and other related documents This duty, especially in recent years, is not taken seriously and has been left to third-party websites to handle.

The situation is changing gradually. The principles of open government data state that for government-related documents to be viewed as “open” to the public, they must be complete, accessible, and machine-processable, amongst other requirements.

The good news is that numerous organizations have been set up to address legislative loopholes in the system where government-related documents are concerned. Citizens have pooled technological resources to solve some of the current problems of open access to legislative records.

For instance, there is NASS-AI, a data duality initiative, and others using artificial intelligence (AI) techniques to ensure that people have access to legislative bills.

Wale Akinfaderin – NASS-AI

Launched in 2019 by Adewale Akinfaderin and Olamilekan Wahab, both machine learning specialists, it has increasingly become established and sponsored, enabling both practitioners to host conferences in within and outside Africa.

Olamilekan Wahab

This was originally an AI side initiative aimed at making legislative information more available according to Social Development Goals (SDG) through means of machine learning and natural language processing (NLP) approaches.

The problem with parliamentary bills in Nigeria

So why should this solution be considered at all?

According to the Open Data Barometer, Nigeria scores 21 out of 100 in the worldwide metric for policy publication and the use of open data for transparency, creativity, and social impact.

The problem seems to be worse because NASS has stopped publishing bills on its website. To access public government documents, researchers and policy experts need to rely upon third-party websites.

The NASS-AI team felt this impact firsthand while going through legislative documents from 1998 to 2018. They found unexpected patterns in the way legislators function during the collection of 20 years of government data.

Most, if not all, of the bills provided by the national assembly, are not well provided. According to the second principle of open government data, documents provided by the government must be “published as collected at the source, with the finest possible level of granularity, not in aggregate or modified forms.”

This requirement was not met by the typical bill that used to appear on the NASS website. Documents were regularly arranged incorrectly and then checked, unclearly explained and were not in any manner highlighted or updated.

An astonishing number of bills over the years

The NASS-AI team estimates that approximately 2,376 bills were proposed by both the Senate and the House of Representatives in the last 20 years, from the timeline in which the bills were drawn up.

Moreover, in important sectors such as education, research, and technology, the number of bills introduced was largely done to create new institutions and nothing more. This gave the illusion that they were pushed by “political and regional undertones.”

Lack of reliable data and engagement of legislators after bills have been introduced

Clearly, with the exception of the interesting news outlet, Nigeria lacks any reliable data regarding the process a bill has passed after its introduction. Nonetheless, apart from the adoption of a bill and it becoming legislation, no evidence can be given for anything that occurs in the midst of the process.

According to the NASS-AI team, “If we have this data, we’ll like to build a predictive model to determine if a bill will pass each stage of reading and if it’ll be passed or not…”

Nigerian legislators are currently struggling to conform to the standards essential for open data governance. Amongst others, NASS has not rendered legislative documentation available to the ordinary Nigerian person, which is much less technological to allow documents to get processed further by the individual.

NASS-AI project on parliamentary documents

“We noticed that the quality of policy-making in developing countries like Nigeria is extremely important, so we decided to work on a project that ties into the United Nations SDGs (particularly, target 16.6 and 16.7),” Wahab says.

Essentially, AI is used by NASS-AI for the classification of legislative bills by the National Assembly in order to make them more available.

This project can be implemented in many cases. State legislators who are eager to enact or support legislation that offers free healthcare to people above the age of 65 could use the project details to search on old bills in order to see what has already been proposed. They could also curate and identify laws from other jurisdictions to adopt regardless of language, analyze policy proposals and discern trends of expenditure.

The purpose of the project NASS-AI is, according to the AI duo, to curate and provide precise metadata for the data supplied by the national assembly.

Work started on the project in February 2019, and by August of the same year, the project had won an Artificial Intelligence for Development (AI4D) Innovation Grant with Olubayo Adekanmbi. In October 2019, they were invited to talk at the NeurIPS  Machine Learning for the Developing World (ML4D) Workshop in Canada, and they also published a report on their experience.

The team is also developing a web-based application to ensure public access to the proposed Government policies.

Their goal is to do so in other African countries, including Nigeria, Kenya, and South Africa.

“We also want the average citizen and/or policymaker to have access to it for knowledge or research purpose,” says Akinfaderin.

It is hoped that their work will enhance the quality of parliamentary access across Africa.