After NLP , here is through a use case, NLU or Natural Language Understanding , a sub-part of language processing in AI which has the specificity of “Morocco Email List unstructured text (documents, emails, etc.) in structured information in a sufficiently clear and unambiguous manner ”.
To better understand this technology and what it can be used for, I interviewed Thomas Solignac, CEO and co-founder of the company Golem.ai, which he co-created in 2016 following four years of research and development on the subject of artificial intelligence applied to automatic reading of texts. What we particularly appreciated in this testimony is the dual competence of Thomas, technical and philosophical, which gives his vision a particularly interesting perspective, both from the point of view of the design of the solution as of its application and of its possible societal implications.
The Importance Of Philosophy In Artificial Intelligence
Thomas has a dual education, he first comes from programming via Epitech where he specialized very early on in artificial intelligence, and at the same time did a university of philosophy in Nanterre. Philosophy is important to him in artificial intelligence. This indeed affects many areas. When we build AI, we build intelligence, and it helps a lot to ask ourselves the question of how intelligence was built, he emphasizes. All these thinkers of the human sciences who have worked on how intelligence manifests itself, what is its medium, are a tremendous source of inspiration to create AI.
At golem.ai, we are deeply inspired by Chomsky’s linguistic theories of how our AI analyzes language, with the same mechanics regardless of the language. If only to manufacture AI, the philosophy is inspiring
Underlying Technologies That Make It Possible To Achieve This Result
The technologies put in place by golem.ai to analyze a call for tenders from A to Z are based on several stages: A first technological brick will transform any type of document into text;
Then, this text will be subjected to a semantic analysis. The crux of the AI difficulty is there. It’s about having categorized text. This semantic analysis will make it possible to resolve ambivalences, for example for the word orange designating the color, the city, or the company. This disambiguation will give rise to categories and information extracted in a very precise way;
Thirdly, the text will be synthesized, in order to transform the semantic analysis into actionable data, for example the amount of the budget, or all the technical characteristics necessary for this call for tenders. The report produced allows questions to be answered immediately and a decision to be taken immediately. This third phase transforms massive semantic data into specific business information;