What is Automatic Classification?

Many tasks undertaken as part of the operations and business processes are focused on data organisation in order to simplify their management or automatize them. We are dealing with processes such as:

Choosing the most suitable policy

Choose the most suitable policy for a particular driver, according to the risk involved.

Determining fraud probability

Determine whether a banking transaction is likely to be fraudulent.

Classifying potential buyers

Classify potential buyers depending on the impact a marketing action might make on them..

Identifying the mood and motive of a client’s message

Identify the mood and motive of a client’s message or complaint written in free text, and send it to the appropriate person to be managed if necessary.

And many other processes where experts on the topic are required to take part. The Automatic Classification techniques support the work of experts and make it easier to analyse a higher volume of data at a lower cost.

How does it work?

Automatic Classification is strongly based on the scientific discipline known as supervised machine learning, which consists in presenting labelled training data from past classifications, provided by experts on the topic, to a system of classification. This system analyses the training data and, using statistical learning algorithms, infers the knowledge of the expert, thus imitating their work.

All this learning process results in a classification system with capacity to process a high volume of data and have shorter response times than those of a manual management. Besides, the system is alive and can be reloaded with new data to readjust the classification processes. This way, we can make sure it is updated to new business policies or future trends on the data.

Exploitation of these classification systems is made applying the business rules to the classifications generated by them. For each classification, the system generates a certainty score that indicates the confidence the system has in its own decision. This allows setting business rules such as automatically issuing an insurance policy classified as low-risk with a high score, or deriving others which generate more doubts to an expert analyst. This means reducing the workload of experts substantially, or increasing the volume of data processed counting on the same number of employees.

Benefits and value of IIC Automatic Classification solutions

Automatic Classification offers:

Automatic processes

Automatic processes of data classification at a lower cost.

Higher capacity of information treatment

Higher capacity of information treatment, involving the same personnel.

High speed of response

High speed of response, providing real-time services that would not be possible if the process was manual.

More solid business rules

More solid business rules based on the confidence of Automatic Classification.

Guaranteed efficiency of IIC Automatic Classification solutions

Over 25 years at the IIC developing highly valuable solutions such as those to detect fraud in payments vouch for our experience in Automatic Classification projects.

In addition, IIC has a team of highly qualified professionals in machine learning, who combine their activity at the Institute with works associated to research, teaching and collaboration with universities, thus keeping up-to-date with the latest technologies. In particular, IIC performs research studies in leading-edge sectors for the most advanced technological companies. This knowledge and practical experience in several Automatic Classification techniques allows us to develop customised solutions which fulfil the response times and efficiency required for each individual project.

The automatic classification techniques support the work of experts and make it easier to analyse a higher volume of data at a lower cost.