Towards Distributed Contextualized Knowledge Repositories for Analysis of Large-Scale Knowledge Graphs

Autoren
L. Bozzato, C. Schütz
Paper
Schu20a (2020)
Zitat
Proceedings of the 35th Italian Conference on Computational Logic (CILC 2020), October 13-15, 2020, Rende, Italy, CEUR-WS.org, Vol. 2710, pp. 82-90, 2020.
Ressourcen
Kopie  (Senden Sie ein Email mit  Schu20a  als Betreff an dke.win@jku.at um diese Kopie zu erhalten)

Kurzfassung (Englisch)

A knowledge graph (KG) represents real-world entities as well as their properties and relationships in a structured and often logic-based formalism. Given the large amount of information and the diversity of data stored in KGs, operations for analysis of such data akin to traditional OLAP operations are useful to understand the contents of KGs along different dimensions. In this direction, we recently proposed Knowledge Graph OLAP (KG-OLAP), a framework based on contextualized description logics that allows to organize knowledge graphs in a multi-dimensional structure -- a KG-OLAP cube. For KG-OLAP cubes, we defined operations for combination of knowledge from different cells and for abstraction of knowledge within cells. Experiments with a proof-of-concept prototype, however, revealed that the management of a centralized KG-OLAP cube is impractical for large KGs. In this paper, we extend KG-OLAP in order to formalize the case in which knowledge is distributed across different repositories. We hence formalize a distributed version of the multidimensional cube structure, and we show how the operations can be adapted to this scenario.