OATAO - Open Archive Toulouse Archive Ouverte Open Access Week

Knowledge hypergraph based-approach for multi-source data integration and querying : Application for Earth Observation domain

Masmoudi, Maroua. Knowledge hypergraph based-approach for multi-source data integration and querying : Application for Earth Observation domain. PhD, Informatique, Institut National Polytechnique de Toulouse, 2020

(Document in English)

PDF (Author's version) - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader


Early warning against natural disasters to save lives and decrease damages has drawn increasing interest to develop systems that observe, monitor, and assess the changes in the environment. Over the last years, numerous environmental monitoring systems and Earth Observation (EO) programs were implemented. Nevertheless, these systems generate a large amount of EO data while using different vocabularies and different conceptual schemas. Accordingly, data resides in many siloed systems and are mainly untapped for integrated operations, insights, and decision making situations. To overcome the insufficient exploitation of EO data, a data integration system is crucial to break down data silos and create a common information space where data will be semantically linked. Within this context, we propose a semantic data integration and querying approach, which aims to semantically integrate EO data and provide an enhanced query processing in terms of accuracy, completeness, and semantic richness of response. . To do so, we defined three main objectives. The first objective is to capture the knowledge of the environmental monitoring domain. To do so, we propose MEMOn, a domain ontology that provides a common vocabulary of the environmental monitoring domain in order to support the semantic interoperability of heterogeneous EO data. While creating MEMOn, we adopted a development methodology, including three fundamental principles. First, we used a modularization approach. The idea is to create separate modules, one for each context of the environment domain in order to ensure the clarity of the global ontology’s structure and guarantee the reusability of each module separately. Second, we used the upper-level ontology Basic Formal Ontology and the mid-level ontologies, the Common Core ontologies, to facilitate the integration of the ontological modules in order to build the global one. Third, we reused existing domain ontologies such as ENVO and SSN, to avoid creating the ontology from scratch, and this can improve its quality since the reused components have already been evaluated. MEMOn is then evaluated using real use case studies, according to the Sahara and Sahel Observatory experts’ requirements. The second objective of this work is to break down the data silos and provide a common environmental information space. Accordingly, we propose a knowledge hypergraphbased data integration approach to provide experts and software agents with a virtual integrated and linked view of data. This approach generates RML mappings between the developed ontology and metadata and then creates a knowledge hypergraph that semantically links these mappings to identify more complex relationships across data sources. One of the strengths of the proposed approach is it goes beyond the process of combining data retrieved from multiple and independent sources and allows the virtual data integration in a highly semantic and expressive way, using hypergraphs. The third objective of this thesis concerns the enhancement of query processing in terms of accuracy, completeness, and semantic richness of response in order to adapt the returned results and make them more relevant and richer in terms of relationships. Accordingly, we propose a knowledge-hypergraph based query processing that improves the selection of sources contributing to the final result of an input query. Indeed, the proposed approach moves beyond the discovery of simple one-to-one equivalence matches and relies on the identification of more complex relationships across data sources by referring to the knowledge hypergraph. This enhancement significantly showcases the increasing of answer completeness and semantic richness. The proposed approach was implemented in an open-source tool and has proved its effectiveness through a real use case in the environmental monitoring domain.

Item Type:PhD Thesis
Uncontrolled Keywords:
Institution:Université de Toulouse > Institut National Polytechnique de Toulouse - Toulouse INP (FRANCE)
Laboratory name:
Research Director:
Archimède, Bernard and Zghal Baazaoui, Hajer
Deposited On:01 Nov 2021 13:04

Repository Staff Only: item control page