cedat logo cedat logo

What's new

Understanding a dialogue through Machine Learning and NLP

December 3, 2021

According to several studies, the use of Artificial Intelligence and, in particular, Machine Learning and Natural Language Processing, can understand the structure of the conversation; highlight similarities and extract patterns that optimise communication according to the context and the shared information.

How many times have we asked ourselves if a different way of communicating could make a difference, for example, in terms of empathy or reduction of misunderstandings? Could science and technology help us? The doubt that they can play a key role also for the purposes of empathic communication is legitimate, but that’s what they do.

Today, we are approaching to this by leaps and bounds, thanks to Artificial Intelligence which contains different methodologies and families of algorithms and which, depending on its applications, is divided into different branches such as Machine Learning and Natural Language Processing. The areas of use can be many and for example, the improvement in the analysis of the stock market, the accuracy of weather forecasts or the detection of credit card frauds. But also, highly delicate contexts from an emotional point of view, such as health care.

According to the Biomed Cue blog, the Close-up Engineering Network area, ‘Machine Learning uses statistical methods to progressively improve the performance of an algorithm in identifying patterns in data and making predictions on them. The most widespread Machine Learning method is Neural Networks, inspired by the learning process of the brain. This method is used in the development of machine learning technologies, in which behaviour is modelled by a randomly connected switching network, following a reward and punishment-based learning routine (reinforcement learning).’

Basically, the idea would be to use Machine Learning and Natural Language Processing algorithms in order to analyse the most important or difficult conversations, with the ultimate goal of understanding their structure, highlighting their analogies and extracting patterns that allow communication to be optimised according to the context and the extent of the information to be shared.

Thinking machines? Surely yes, also because seventy years ago, someone else had imagined the way in which the world would have been transformed by machines capable of thinking.

We are talking about Alan Turing, the famous British computer scientist, who in his article ‘Computing machinery and intelligence’ of 1950 – re-proposed by the Financial Times and quoted by Internazionale – explained that ‘Computers could become so adept at imitating human beings that it would be impossible to distinguish them from people of flesh and blood.’ 

‘We can hope that one day machines will compete with humans in all intellectual fields.’ wrote Turing. Seventy years later the Financial Times explains that thanks to the rapid development of the internet and the exponential increase in computing power, we are entering to a world where the role of machines exceeds even Turing’s imagination. Thanks to the new software techniques, such as Neural Networks and Deep Learning, computer scientists have become much more adept at training machines.

The evolution of Machine Learning will give us more and more certainties also with regard to the human factor in sentiment analysis, but always in the context of collaboration and not competition with humans.