These engines present specific knowledge about subjects looked up through keyword queries, by in turn performing logical queries on knowledge bases such as Wikipedia, where encyclopaedic information is stored digitally. These knowledge bases, which artificial intelligence has been dreaming about for thirty years, have now become a reality. The rapid development of this concept and knowledge and database technologies have made it possible to store and query a much broader and heterogeneous range of content. Increasingly, a significant proportion of humanity’s knowledge is concerned, rather than simply company databases, for example. These knowledge bases have only been able to emerge as a result of the changing role of the user, who has become an active contributor to the web. This is what is referred to as the “crowd” aspect of the web, since the knowledge stored in these databases is entered in RDF format by a crowd of users. This means that we are dealing with a whole new generation of advanced databases including knowledge bases which, unlike those from 20 years ago, no longer require expert engineers to operate. These knowledge bases like Wikipedia are expanded by everyone, using their own web interfaces and can be easily queried via search engines.
Inria, an ideal environment
At Inria, the 13 members of the Links project team (including 8 permanent research professors) headed by Joachim Niehren, work in the field of logical queries, developing new research and information extraction techniques based on knowledge and databases. Given the increasingly massive volumes of data and knowledge on the web, the stakes are high. “The challenge is to find a smarter way of simultaneously querying and crossing as many databases and knowledge bases as possible, while taking the origin of the information into account in order to identify responses that are as comprehensive and relevant as possible”. We must also take into account new database formats, such as NoSQL, Graph or RDF knowledge bases ”, explains the researcher. After studying computer science and mathematics with artificial intelligence applications, in particular for programming languages and computational linguistics, Joachim Niehren began his research career in Saarbrucken, in his native country of Germany. It was upon arriving in Lille in 2003 that he decided to focus his research on databases for the web - from which knowledge bases have evolved, combining data representation methods using artificial intelligence. At Inria, Joachim Niehren found favourable working conditions for conducting his research and became head of the Links team in 2012. “Inria is a renowned international institution where we can work as a team in an ideal environment with fellow researchers and PhD and postdoc students ”.
The limits of artificial intelligence are constantly being expanded
Given the stakes of databases for companies, Links is regularly approached by the world of industry. “We have to work towards a long-term transfer. So it’s important for us to cooperate with industry players on issues they face today, while also preparing for the future through our more fundamental research work. Inria connects researchers with companies and organizes international and national meetings in Paris, and regional ones in Lille. I’m also very proud when the PhD researchers we train start working at innovative companies or start-ups that have grown out of the world of research, and spread our ideas”, adds Joachim Niehren. And in a field as exciting as IT research, even if researchers know how to further develop the foundations of databases, it is difficult to accurately anticipate upcoming application challenges. “Web advances correspond to advances in the field of databases but the limits of artificial intelligence are constantly being pushed back. Who could have predicted the applications we use today online?” Google, Facebook and Wikipedia are examples of huge database applications which were unimaginable a few decades ago.
- Project-Team Links - Linking Dynamic Data