Ph.D
Group : Large-scale Heterogeneous DAta and Knowledge

Maintenance of semantic annotations

Starts on 18/11/2015
Advisor : REYNAUD, Chantal

Funding : Bourse pour étudiant étranger
Affiliation : Université Paris-Saclay
Laboratory : LIST et LRI

Defended on 07/12/2018, committee :

Research activities :

Abstract :
Semantic annotations are often used in a wide range of applications ranging from information retrieval to decision support. Annotations are produced through the association of concept labels from Knowledge Organization System (KOS), i.e. ontology, thesaurus, dictionaries, with pieces of digital information, e.g. images or texts. Annotations enable machines to interpret, link, and use a vast amount of data. However, the dynamic nature of KOS may affect annotations each time a new version of a KOS is released. New concepts can be added, obsolete ones removed and the definition of existing concepts may be refined through the modification of their labels/properties. As a result, many annotations can lose their relevance, thus hindering the intended use and exploitation of annotated data. To solve this problem, methods to maintain the annotations up-to-date are required. In this thesis we propose a framework called MAISA to tackle the problem of adapting outdated annotations when the KOS utilized to create them change. We distinguish two different cases. In the first one we consider that annotations are directly modifiable. In this case, we proposed a rule-based approach implementing information derived from the evolution of KOS as well as external knowledge from the Web. In the second case, we consider that the annotations are not modifiable. The goal is then to keep the annotated documents searchable even if the annotation are produced with a given KOS version but the user used another version to query them. In this case, we designed a knowledge graph that represents a KOS and its successive evolution and propose a method to extract the history of a concept and add the gained label to the initial query allowing to deal with annotation evolution. We experimentally evaluated MAISA on realistic cases-studies built from four well-known biomedical KOS: ICD-9-CM, MeSH, NCIt and SNOMED CT. We show that the proposed maintenance method allow to maintain semantic annotations using standard metrics.


CAUSAL LEARNING FOR DIAGNOSTIC SUPPORT


CAUSAL UNCERTAINTY QUANTIFICATION UNDER PARTIAL KNOWLEDGE AND LOW DATA REGIMES


MICRO VISUALIZATIONS: DESIGN AND ANALYSIS OF VISUALIZATIONS FOR SMALL DISPLAY SPACES
The topic of this habilitation is the study of very small data visualizations, micro visualizations, in display contexts that can only dedicate minimal rendering space for data representations. For several years, together with my collaborators, I have been studying human perception, interaction, and analysis with micro visualizations in multiple contexts. In this document I bring together three of my research streams related to micro visualizations: data glyphs, where my joint research focused on studying the perception of small-multiple micro visualizations, word-scale visualizations, where my joint research focused on small visualizations embedded in text-documents, and small mobile data visualizations for smartwatches or fitness trackers. I consider these types of small visualizations together under the umbrella term ``micro visualizations.'' Micro visualizations are useful in multiple visualization contexts and I have been working towards a better understanding of the complexities involved in designing and using micro visualizations. Here, I define the term micro visualization, summarize my own and other past research and design guidelines and outline several design spaces for different types of micro visualizations based on some of the work I was involved in since my PhD.