|
|
|
| e-Pub |
Previous | 
Home
Section: New Results
Vocabularies, Semantic Web and Linked Data Based Knowledge Representation and Artificial Intelligence Formalisms on the Web
Semantic Web for Biodiversity
Participants : Franck Michel, Catherine Faron Zucker, Antonia Ettorre.
The development of an activity related to biodiversity data sharing and integration is going on through the sustained collaboration with the ”Muséum National d'Histoire Naturelle” of Paris (MNHN).
First, at the very end of 2018, we published a journal paper about the SPARQL Micro-Services architecture and how this can be useful in the biodiversty domain [62]. Then, through the internship of a Ubinet master student, we explored how SPARQL Micro-Services can help biologists in editing taxonomic information by confronting multiple, heterogeneous biodiversity-related data sources. We presented some results of this work at the Biodiversty_Next conference 2019 [28].
Within the same internship we continued the work meant to publish biodiversity data as linked data (TAXREF-LD (http://agroportal.lirmm.fr/ontologies/TAXREF-LD)). The goal is to extend the dataset from simple taxonomic data to new types of data: species interactions, multi-lingual names, conservation and legal statuses. This work should lead to a publication in 2020.
During the last two years, we have lead the biodiversity task within the Bioschemas.org W3C community group that seeks the definition and adoption of common biology-related markup terms. We proposed the creation of the Taxon term (http://bioschemas.org/devSpecs/Taxon/) whose adoption in Schema.org is under discussion. The work now starts bearing fruits as 180.000+ webpages of the MNHN are now annotated with the Taxon term, paving the way to more biodiversity resources being published as structured data that search engines can process to provide more accurate search results.
Semantic Web for eEducation
Participants : Catherine Faron Zucker, Géraud Fokou Pelap.
In the framework of the EduMICS project we developed and populated an ontology to represent the students' activity on the Educlever learning platform.
Semantic Web for B2B applications
Participants : Molka Dhouib, Catherine Faron Zucker, Andrea Tettamanzi.
In the framework of the collaborative project with Silex France company aiming to model the social network of service providers and companies, as a preliminary step, we developed an ontology alignment approach combining word embedding and the radius measure to detect matching concepts and determining equivalence or hierarchical relations between them. We report and discuss the results of the evaluation of our approach on the OAEI complex alignment benchmark and on the SILEX use case: aligning reference vocabularies to annotate B2B services (ESCO to Cigref, ESCO to ROME, NAF to kompass and NAF to Silex activity domains) [35].
Integration of Heterogeneous Data Sources
Participants : Franck Michel, Catherine Faron Zucker, Fabien Gandon.
With the incentive of fostering the integration of Linked Data and non RDF data sources, we continued the work initiated around the SPARQL Micro-Service architecture that harnesses the Semantic Web standards to enable automatic combination of Linked Data and data residing in Web APIs. We published a paper at the LDOW workshop of the Web Conference that explores how we can leverage Schema.org to enable web-scale discovery and querying of Web APIs using SPARQL micro-services [27].
Uncertainty in the Semantic Web
Participants : Ahmed El Amine Djebri, Fabien Gandon, Andrea Tettamanzi.
In the framework of Ahmed El Amine Djebri's thesis, we proposed an approach to publishing uncertainty on the Semantic Web [15] and to link and negote uncertainty theories [14].
Uncertainty in Human Geography
Participant : Andrea Tettamanzi.
In the framework of the Incertimmo collaborative project between Université Côte d'Azur and Kinaxia, we applied machine learning and urban morphology theory to the investigation of the influence of the urban environment on the value of residential real estate [6].
Ontology Design Rule
Participants : Olivier Corby, Catherine Faron Zucker, Philippe Martin.
We worked on the topic of Ontology Design Rules with Philippe Martin, from université de la Réunion, during his visit to the Wimmics team. This work resulted in a publication at Semantics [25].
Suggestion of Data Sources for SPARQL Queries over Linked Open Data
Participants : Hai Huang, Fabien Gandon.
For querying processing over Linked Open Data, suggestion of relevant data sources with respect to a SPARQL query is crucial since it highly affects the performance of querying. In this work, we focus on the problem of suggesting k relevant data sources with respect to a SPARQL query. We propose a summarization method which models the RDF graph of linked data sources and query graphs as sets of feature paths (star, sink and chain paths) and an effective algorithm to extract these feature paths for data sources and query graphs. To obtain candidate data sources we propose a time and space efficient search algorithm based on locality sensitive hashing. We perform a large-scale experiment on real world linked datasets which shows that our algorithm outperforms existing baselines.