Optimisation of Link Traversal Query Processing over Distributed Linked Data through Adaptive Techniques

Abstract

An increasing amount of distributed Linked Data is being made available at different locations, with varying formats, structures, interfaces and availability. Making use of that data through declarative query languages such as SPARQL requires query engines capable of executing queries over it. Efficiently executing queries over the data requires efficient query plans, yet prior access to the information for producing such plans may not be possible due to the distributed and dynamic nature of the data. Furthermore, the inability to be aware of all data sources at a given time, following links to discover data in the form of link traversal may be needed. Consequently, query planning and optimisation may need to be performed with limited information, and the initial plan may no longer be optimal. Discovering additional information and data sources during query execution and adjusting the execution based on such discoveries using adaptive query processing techniques therefore could help perform queries more efficiently. The aim of this work is to explore a variety of existing or potential new techniques and their combinations for query-relevant information acquisition and query plan adaptation within the context of distributed Linked Data. Already prior results from multiple studies have demonstrated the benefits of various such techniques within or beyond Linked Data and Link Traversal Query Processing, and this work seeks to build upon such results to realise the benefits of various techniques in practice to tackle performance-related challenges.