Information Commissioner's Office - ML email classification model

Information Commissioner's Office - ML email classification model

Fined Facebook £500k

900,000 fee payers

Upholds GDPR and other data protection legislation

ICO: Email classification

The Information Commissioner’s Office (ICO) is the public body which is responsible for regulating data protection in the UK. The ICO’s mission is to “uphold information rights in the public interest, promoting openness by public bodies and data privacy for individuals”. Organisations and freelancers all over the UK must register and pay a monthly fee to the ICO if they are processing personal data.

Fast Data Science - London

Need a business solution?

NLP, ML and data science leader since 2016 - get in touch for an NLP consulting session.

The ICO receives a large volume of emails from customers, which could be large multinationals or small freelancers. The ICO had a need for a machine learning model to categorise the emails for their internal processes.

Fast Data Science used Microsoft Azure ML to train a machine learning model which categorises incoming emails to a number of categories, allowing the ICO to handle them more efficiently.

Data protection and privacy

The ICO project was challenging in particular because the machine learning model had to be trained on sensitive data, which could not be stored or duplicated. For this reason we had to handle all training data with the utmost care and make sure that it was used only for training the model, was not saved, and did not leave the isolated training environment.

You can read more about the approaches which we use at Fast Data Science to train machine learning models on sensitive data in this blog post.

Machine learning email classification basics

In general, machine learning email classification can have various applications, such as spam filtering, customer support, sentiment analysis, and information retrieval. There are many different ML email classification methods and techniques, such as feature extraction, feature selection, text preprocessing, and supervised or unsupervised learning algorithms.

What we can do for you

Transform Unstructured Data into Actionable Insights

Contact us