Star Pattern Fragments: Accessing Knowledge Graphs through Star Patterns

Abstract

The Semantic Web offers access to a vast Web of interlinked information accessible via SPARQL endpoints. Such endpoints offer a well-defined interface to retrieve results for complex SPARQL queries. The computational load for processing such queries, however, lies entirely with the server hosting the SPARQL endpoint, which can easily become overloaded and in the worst case not only become slow in responding but even crash so that the data becomes temporarily unavailable. Recently proposed interfaces, such as Triple Pattern Fragments, have therefore shifted the query processing load from the server to the client. For queries involving triple patterns with low selectivity, this can easily result in high network traffic and slow execution times. In this paper, we therefore present a novel interface, Star Pattern Fragments (SPF), which decomposes SPARQL queries into star-shaped subqueries and can combine a lower network load with a higher query throughput and a comparatively low server load. Our experimental results show that our approach does not only significantly reduce network traffic but is also at least an order of magnitude faster in comparison to the state-of-the-art interfaces under high query processing load.

Optimizing SPARQL queries over decentralized knowledge graphs

“…To further limit the network overhead, Star Pattern Fragments (SPF) [3] clients send conjunctive subqueries in the shape of stars (star patterns) to the server and process more complex patterns locally on the client. Such conjunctive subqueries can be processed relatively efficiently by the server [58], which results in the transfer of significantly fewer intermediate results than in systems like TPF and brTPF.…”

“…However, as shown in our experimental evaluation in Section 7, LOTHBROK actually decreases the number of intermediate results to be enumerated by up to two orders of magnitude compared to triple pattern-based query executors using the optimization techniques explained above. Furthermore, we apply pagination of large result sets, which related studies [32,68] have already shown can effectively limit the effects of a large number of solution mappings, even for star pattern-based query execution [3].…”

“…As discussed in Section 1, star-shaped subqueries can be processed relatively efficiently over a fragment [58], thus they can also help achieving a better balance between reducing the communication overhead and distributing the query processing load [3,13,14]. To facilitate processing such star patterns on single nodes, we propose to fragment the uploaded knowledge graphs based on characteristic sets [13,14,55].…”

“…The intermediate results from previous steps are used as input to subqueries at a later stage in the query execution plan. In case of a distributed join, the intermediate results are transferred along with the partial query to use local bind joins similar to [3,32]. To formalize how star patterns in the query execution plan are processed over the fragments, we define a so-called selector function in line with related work [3,6,32].…”

“…In recent years, several decentralized systems [3,4,6,13,45,68] have been proposed to alleviate the aforementioned burden from the data providers by reducing the computational load required to keep the data available, albeit using different methods to do so. For instance, Linked Data Fragments (LDF)-based approaches [3,13,14,34,68] reduce the computational load on the server by distributing some of the query processing effort to the client, ensuring that the server only processes requests with low time complexity. On the other hand, Peer-to-Peer (P2P) more accurate cardinality estimations.…”