Showing 1–1 of 1 results for author: Sklavos, A

Estimating the Cost of Executing Link Traversal based SPARQL Queries

Authors: Antonis Sklavos, Pavlos Fafalios, Yannis Tzitzikas

Abstract: An increasing number of organisations in almost all fields have started adopting semantic web technologies for publishing their data as open, linked and interoperable (RDF) datasets, queryable through the SPARQL language and protocol. Link traversal has emerged as a SPARQL query processing method that exploits the Linked Data principles and the dynamic nature of the Web to dynamically discover dat… ▽ More An increasing number of organisations in almost all fields have started adopting semantic web technologies for publishing their data as open, linked and interoperable (RDF) datasets, queryable through the SPARQL language and protocol. Link traversal has emerged as a SPARQL query processing method that exploits the Linked Data principles and the dynamic nature of the Web to dynamically discover data relevant for answering a query by resolving online resources (URIs) during query evaluation. However, the execution time of link traversal queries can become prohibitively high for certain query types due to the high number of resources that need to be accessed during query execution. In this paper we propose and evaluate baseline methods for estimating the evaluation cost of link traversal queries. Such methods can be very useful for deciding on-the-fly the query execution strategy to follow for a given query, thereby reducing the load of a SPARQL endpoint and increasing the overall reliability of the query service. To evaluate the performance of the proposed methods, we have created (and make publicly available) a ground truth dataset consisting of 2,425 queries. △ Less

Submitted 17 October, 2022; originally announced October 2022.