Semantifynd

Identified challenge

Efficient and relevant information retrieval is a major challenge in a digital environment with an abundance of data. The main challenge lies in managing the large volume of pages to be indexed or searched on the various websites of the Luxembourg State, in particular guichet.lu. This complexity is amplified by the diversity of heterogeneous content, whether structured, unstructured or semi-structured (such as JSON format), and by the diversity of hosts.

Currently, a search can be carried out in two main ways:

• Via the site structure, where ease of search is conditioned by the UX (User Experience) and clarity of the site.
• Via the internal search engine (Exalead), which remains rudimentary and is based mainly on keyword frequency.

These traditional methods come up against limitations, particularly in terms of the relevance of results, understanding the user context, managing typos, synonym searches, multilingualism, and interpreting ambiguous or complex queries. Furthermore, searching via external search engines imposes an additional constraint linked to the technical and semantic optimisation of the site to ensure good visibility on these platforms, with no guarantee of an optimal user experience.

The central objective of this innovation partnership is to introduce a third method of accessing information on websites, using semantic search based on vectorised documents. This innovative approach aims to overcome the limitations of existing methods and provide an enhanced user experience, facilitating access to accurate and relevant information through an intuitive and powerful search interface.