Reading Time: 3 minutes
Photo by Woody Kelly on Unsplash

Everything ends happily … the idea of a dating site that matches researchers, funders, and policy makers is a great one. So I was excited to try out the new  UK Government-funded Areas of Research Interest (ARI) database. Unfortunately, it seems to let itself down at the very first hurdle: you can’t find things on it.

The idea behind the site is simple. There are several government departments who would like to have topics of interest researched and reported on. But keeping in touch with 10 or more different government departments is no fun. Instead, there is now a single portal that lists a lot of UK government areas of research interest, and institutions and researchers can conveniently look through one list rather than several.

But does it work? I searched for “generative ai” (as you would, since this is probably the hottest topic in IT during the last twelve months), and saw the following results – just three results, by the way, and none of them actually “generative AI”:

Is that all that the funding councils are offering on generative AI, i.e. nothing at all?

Worse, I realised the site provides a (literally) brainless string-matching search: I searched again, to make it easier for the system to find things, by searching  just  for “ai” – only to find hits for “aided”, “air”, and so on. In other words, the most primitive type of string search is provided, with no sorting (so that whole words comprising “ai” appear before “ai” contained within other words).

If I were building a new search database in 2023 I wouldn’t be so keen to have my name appear so prominently on the list of who is responsible for the website,  displayed on a very large footer:

Nice logos, but lousy search.

Moral of the story: if you provide a search box nowadays, people will expect it to perform the way that Google does – which means, with some intelligence behind it.

Leaving the defects of the search system aside, some of the problems of findability stem from the rather limited content provided. The research topics themselves have a very limited description. To make search easier, some attempt has been made to add subject tags, but these seem to reveal more about the people writing it than about the topic. Here is an example of one research topic:

I wonder if the MOD has found anyone to carry out this research, which is helpfully tagged “defence”, “technology”, and “allies”. It looks like several possible research areas, none of them described very clearly, and written in the language of a soldier who finds the real (non-military) world alien. Working with your allies is a very different topic to identifying “adversaries” who are “hiding within the civilian population”, which is different again to “alternatives to the use of hard/physical power”. With such a blunt, wide-ranging problem statement, it sounds unlikely the MOD will find what they are looking for. Perhaps more work could have been done to amplify the research interest and to state more clearly and accurately what is required.

It is commendable that the database is public and browsable. However, the combination of the wording of the requirements and the functionality of the site makes me feel that more work is required to reach the perfect date between researcher and government department. I suspect there will be many intermediate stages before the altar of happiness is reached.