EDAO · Example-Driven Analytics of Open Knowledge Graphs
Horizon 2020 — Marie Skłodowska-Curie Actions
- Duration
- 2019-09-15 → 2021-09-14
- EU contribution
- €207,312
- Participants
- 1
- Scheme
- MSCA-IF-EF-ST
Lines connect the coordinator with its partners.
Results in brief
Example-Driven Analytics of Open Knowledge Graphs
The Linked Open Data cloud (LOD) contains a very rich corpus of information that requires dedicated business analytics and information extractions technologies for the elicitation of valuable insights. These are usually modelled in Open Knowledge Graphs, which is a way to model information as entities linked by semantic relationships. Yet, to access this type of data and perform such analysis, the typical gateways are specialized query languages (e.g., SPARQL) that are usually challenging to use for non-expert users. This constitutes a major impediment to the successful exploitation of Linked Open Data. To support advanced LOD analytics we propose a novel data exploration system that allows users to extract insights within complex and unfamiliar datasets. We studied and proposed methods to help non-expert users to perform exploratory analysis of open knowledge graphs by applying the Exemplar Query paradigm to the case of Exploratory Online Analytical Processing (OLAP). Example-based methods have proven to be extremely valuable since they avoid complex query languages by using examples to represent the required information. Yet, they have never been studied in the OLAP/BI context. Therefore, we propose to study a new Example-Driven Exploration system to bridge the gap between example-based queries and BI methods. We study two important aspects of the problem: how to enable users to express their information need, and how to make this approach efficient for rich and large scale datasets.
Data: CORDIS, © European Union
Project objective
Linked Open Data (LOD) is a standard methodology especially adopted to implement Knowledge Graphs, i.e., networks of facts where entities are connected by predicates describing relationships among them (via RDF triples). LOD are adopted in many domains, and an enormous set of information is currently shared by the private and the public sector in this form (e.g., on the EU Open Data Portal). Therefore, the LOD cloud contains a very rich corpora of information that requires dedicated business analytics and information extractions technologies for the extraction of valuable insights. Yet, to access this data and perform such analysis, the typical gateway are specialized query languages (e.g., SPARQL) that are usually challenging to use to non-expert users. This constitutes a major impediment in their successful exploitation. To support advanced LOD analytics we propose a novel data exploration system which allows users to extract insights within complex and unfamiliar datasets. We plan to implement dedicated Business Intelligence (BI) operators enabled by the Exemplar Query paradigm for Exploratory Online Analytical Processing (OLAP). Example-based methods have proven to be extremely valuable since they avoid complex query languages by using examples to represent the required information. Yet, they have never been studied in the OLAP/BI context. Therefore, we propose to study a new Example-Driven Exploration system to bridge the gap between example-based queries and BI methods. The researcher has co-authored the first paper on Exemplar Queries for graphs. Moreover, the supervisor, prof. Torben Bach Pedersen at Aalborg University, is an expert on BI/OLAP methods for web and semi-structured data. The host of the secondment, prof. Ioana Manolescu, at INRIA Saclay, is expert in advanced RDF analytics operators. These high-profile collaborations will ensure both the successful outcome of the project as well as a platform for the development of the researcher’s career.
Original text from CORDIS.
Participants
- AALBORG UNIVERSITET · AalborgCoordinatorDenmark
Links
Data: CORDIS, © European Union