Machine learning

Machine Learning is as old as time. It is hard to imagine, but before calculators were on the market, calculations were done by people, and there was a profession called 'calculator'. In recent years, we mainly see it in the field of language comprehension, but computers are also becoming increasingly adept at handling images and sound.

Language models: teaching a machine to read

In many studies, we gain insights based on statistics. Sometimes, the necessary figures are not available in a structured form, but in unstructured texts. For example, we can collect data on the skills demanded in the job market based on job vacancy texts. Text classification can automate this process and enable a computer to assign a 'label' automatically to a piece of text. The algorithms we have developed are reliable and resilient to (for example) spelling mistakes.

In the example below, we demonstrate how easy it is to teach a computer to distinguish between Dutch boys' and girls' names. We utilise the software and machine learning models developed by Dialogic themselves.

Text classification allows for the analysis of large amounts of text. An example is analysing large volumes of literature to determine the extent to which articles from a university relate to sustainability goals (SDGs). The images below show the results of this analysis for a Dutch university.