Work

Knowledge and experience management integration: A graph-based learning approach

Year: 2025

Type: article

Abstract: In order to efficiently and easily manage knowledge within Small and Medium-sized Enterprises (SMEs), Personal Knowledge Management systems (PKMs) can be used. They enable individuals to formalize their experiences and knowledge using knowledge graphs called Personal Knowledge Graphs (PKG). To facilitate the sharing and dissemination of these individual knowledge repositories, an Organizational Knowledge Management system (OKMs) enables to gather several pieces of individual experiences or knowledge within Organizational Knowledge Graphs (OKG). This article proposes an approach for knowledge formalization and exploitation where graph nodes are transformed into vector embeddings and a learning mechanism enables to enrich them by the knowledge of their neighbors. Based on knowledge formalized by vectors, a second mechanism enables to identify relevant pieces of knowledge according to a new request. We applied our proposals in industrial problem-solving domain to recommend actors likely to work on new problems based on their fields of competence and their experiences. Our experiments show that it is possible to make better actor recommendations for new problems by reinforcing the knowledge of these actors through learning with knowledge that are semantically similar to that of the new problems.

Source: IFAC-PapersOnLine

DOI: 10.1016/j.ifacol.2025.09.429

Authors: Ismaël Tankeu, Adama Arama, Laurent Geneste, Thierry Coudert

Institutions: Laboratoire Génie de Production, Université de Toulouse

Language: English

Topic: Advanced Graph Neural Networks

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: diamond

Funder: Agence Nationale de la Recherche

Award: Experience and Knowledge Management for Continuous Improvement in the Enterprise 4.0