Big Data in the public sector

Big Data in the public sector

20/08/18 | by: Jack Blogg

The public sector has a number of distinctive points that separate it from private business. It generally has less money and is accountable for all of it to every individual paying into the tax system; it is rightly awash with rules on confidentiality and its data points overlap.

It’s potentially complicated and regulated to the Nth degree, but with the right partnership it can be made to hook up like any other instance of Big Data.

Take one example: death through domestic violence. It’s obviously an appalling thing, and in data terms a profile might already have been emerging had all of the data points been taken into account to build the complete picture. Humans tend to see the links with hindsight, whereas artificial intelligence can make the links dispassionately. Other equally troubling occurrences, such as modern slavery, identifying reoffenders and troubled families that can be led into further difficulties, also traverse the data points through financial, social, law enforcement, health and others.

The thing is, the better the ability to report on these elements and link them up, the more efficient prevention is likely to be. It’s often a matter of breaking a report into its different elements, which is where something like IBM’s Watson cognitive computing system can help. Say an ambulance crew picks up John Smith who has passed out after a night drinking at home. They report the incident and due to language and pattern analysis, Watson can spot that this is likely to be the same person as Jack Smythe, known to the police and wanted in connection with knife crime. Watson then joins the dots and informs the police.

En masse, breaking down a lot of data like that, structuring it and redistributing it (subject to regulations) among the right agencies will lead to more accurate forecasting of what’s going to go wrong for people and where. Health agency A might note that child B is ill frequently; Watson will spot coincidences but also cases in which there might be a pattern, and be able to check other criteria as well before presenting the probability of the need for intervention to the social services. Cross-matching of data is also possible; sticking with the fictional person above, he may easily present himself as Johnny Smith, change his car, be found somewhere else and be accused of a mugging. The more data that’s been extracted about him and attached to a profile, the better chance there is of ascertaining whether there are two or three John Smiths involved or just the one.

As long as all the data is held by the right authorities in a secure environment there is no reason not to take full advantage of the technology available. Entity extraction, predictive analytics and flagging can exist at agency level with a central pool of shareable data so that people can be more easily identified by the services that can offer them the support they need.

It works in Brent, for example, where eleven risk factors have been identified and correlated by the IT system. It’s early, but it should be possible very quickly to identify difficulties, some of which have not yet happened, and completely revolutionise the public sector and maybe prevent a few tragedies. It starts with a look at cloud computing and the opportunities that offers to save some money.

Contact us to find out more.

Comments are turned off for this article.