6533b831fe1ef96bd12980b3

RESEARCH PRODUCT

Knowledge engineering and semantic formalism of symbolic relations. From medieval network to web 3.0

Djibril Diarra

subject

Semantic Relation / Influencial RelationRéseau SociauxRelation Sémantique / Relation d'influence[INFO.INFO-TI] Computer Science [cs]/Image Processing [eess.IV]OntologyIngénierie des ConnaissancesEnluminure MédiévaleNumérisation du Patrimoine CulturelKnowledge EngineeringSocial NetworkCultural heritage DigitalisationOntologieMedieval Illumination

description

A search has always been at the center of human concerns, the one of conditions and means to represent, to control well, and to communicate effectively with his environment. While technological advances and electronic gadgets such as smartphones, the Web, and online social networks currently allow addressing this concern, it has not always been the case. Our works in this thesis treat medieval documents considered as a medium of communication and as a model of the social network in the Middle Ages. These documents are called medieval illuminations. They were luxurious paintings used in that time to represent the environment and ideal world for elites such as prince, duke, etc. Those paintings contain tremendous metaphoric and symbolical items such as ornaments, fabulous animals, etc. The semantic ambiguity of those items, some of their combination, and the contexts of realisation of illuminations themselves make that their interpretations and their comprehension are currently reserved exclusively to the medievists. In this thesis, we propose an approach that reduces the semantic heterogeneity of the items contained in the illuminations and allows the medievists to transmit their know-how on these medieval paintings through a formalisation of the knowledge they contain. The goal is to allow a more global understanding of illuminations and their items' interpretation. To achieve that, we use techniques and methods of knowledge engineering that we apply to the medieval illuminations. The contents of those laters are extracted and represented in a formal model. That model serves in a Web application that allows any user to describe illuminations and detect automatically some of their contained items. This automatic detection is done thanks to machine learning methods implemented in the Web application. At the same time, a comparative study between the principles of the spread of social influence through illuminations as a model of social network and the ones through social media is done. That comparison provides food for thought on the measures by which questions relating to this notion of social influence, such as its evaluation, quality, sustainability, and its maximisation, could be tackled better.

https://theses.hal.science/tel-03601884