Thesis defense by Lise KIM

• The members of the jury are:

Mr. Sebti FOUFOU – Professor at the University of Burgundy

Mr. Hervé PANETTO – University Professor, University of Lorraine

Mr. Frédéric NOEL – PU Genroble INP

Ms. Nadège TROUSSIER – Professor, University of Troyes

Mr. Philippe Véron – Professor at LISPEN Arts et Métiers

Ms. Esma YAHIA – Lecturer, LISPEN Arts et Métiers

Mr. Frédéric SEGONDS – Senior Lecturer HDR LCPI Arts et Métiers

Mr. Benjamin DEGUILHEM – Doctor CAPGEMINI

Mr. Victor FAU – Engineer, CAPGEMINI

Mr. Nicolas CROUE – Engineer, CAPGEMINI

• Abstract: The valorization of information assets in manufacturing companies is an important issue. It enables informed decision-making and the identification of new value-added opportunities. When digitally transcribed, these information assets consist of heterogeneous data distributed across different silos within the company, making it difficult to obtain a holistic view of the information. This thesis proposes accessing the company's heterogeneous and distributed information through an information retrieval system. The originality of the proposal lies in considering and modeling all of the company's structured and unstructured data in a single graph. Applying this approach to a case study has identified a list of key issues that need to be addressed in order to improve the usual performance criteria in information retrieval. The four issues considered are: (i) processing the syntactic specificities of the data, (ii) semantically extending the terms used in the search, (iii) filtering out irrelevant results, and (iv) detecting implicit links between data. Finally, the approach enriched by the proposals for these issues is tested on a second case study in order to validate the proposal.

• Keywords: Information retrieval, Graph-oriented database, Manufacturing industry, Semantic query extension