Interview with Josiane MOTHE, optimizing information retrieval (IR)

We publish the fourth video in our series of interviews presenting research work from our different departments. Josiane MOTHE, teacher-researcher in the department Gestion de Données (GD) – GIS team, presents its research in the field of information retrieval (IR). What is at stake? Responding to strong and varied societal problems such as research and access to accurate information, cancer detection in the medical field, detection of abnormal phenomena by observing the Earth or searching in social networks. 

What is information research?

The work of Josiane MOTHE and her team focuses on information systems, whose central core is data. This consists in the design and development of models and tools allowing an efficient access to data, and by extension, to knowledge. The objective of information retrieval (IR) is to find, for the user, relevant documents in relation to a query he formulates. To illustrate this, searching the Web with a search engine is a good example. Researchers are not necessarily only interested in the number of clicks or the number of ads seen, which are now considered as efficiency criteria in the industrial world. On the contrary, they consider that user satisfaction, diversity of answers and plurality of viewpoints proposed following an information search are crucial criteria for citizens. A number of subfields of information retrieval are of interest to academics. Among them, we can mention information retrieval for specialists (e.g. medical IR); on specific media (social networks); interactive information retrieval (conversational agents); biases in IR; false information; reproducibility of results; diversity of sources and points of view to avoid bubbles, unique or sensational information that maximizes clicks or artificial intelligence for information retrieval.

What are the issues related to this area of research? 

“It should not be the search engines that, through their answers, dictate the decisions and points of view of citizens. It should be the tools that, instead, show different points of view, help in reading the world and each other’s environment.” Josiane MOTHE

The field of information retrieval is now dominated by GAFAM. However, these companies are interested in the masses, while the researchers focus on the specific aspects of search, the niches, which are smaller-scale issues. For example, answering all queries well and not just “well on average” is a key issue for the team. The team seeks to design methods to optimize each query individually, without user information, and thus in a non-intrusive way. Among these methods, the team has developed a multi-core engine, which ensures good results, not on average, but for each query. One of the short and medium term challenges is that the industrial world should adopt this principle of the multi-core search engine, which has great potential for users. More generally, the search for digital information has a very central place in the lives of citizens. It is important to offer search mechanisms that respond to all queries in an appropriate way. We can talk about fair and equitable information search.