Last year I gave two presentations, one to the DLM Forum Triennial and one to the IRMS conference, in which I developed a fictional case study of an organisation that decides to apply machine learning and analytics to email.
In my case study a public sector organisation:
- is concerned about the low capture of email into its record system (SharePoint) and embarks on a programme to apply machine learning to remedy the shortfall;
- uses machine learning to apply its existing policy of moving important email into a records system;
- seeks to apply the machine learning capability on all email accounts on a corporate wide basis.
In reality I think that the attitude of public sector organisations to the application of analytics and machine learning to email will be rather different to the attitude taken by the organisation in my case study. My predictions are that public sector organisations in the UK:
- will be reluctant to apply machine learning to email accounts because of the risks involved;
- will be just as concerned about the prospect that the application of machine learning might result in very large volumes of email being captured into their record system as they would be about the existing under-capture of emails as records;
- would use machine learning (or an analytics capability) to look for certain specific types of correspondence that are valuable to the organisation in certain specific accounts rather than applying machine learning/analytics to all accounts across the business;
- would not move emails identified as important or valuable into their corporate records system, but would instead leave them within email accounts and either place them under a hold to prevent deletion, or move them to an email archive.
Here is the video of my monologue explaining how the organisation applying machine learning to all of its email accounts got on: