Structural Properties as Proxy for Semantic Relevance in RDF Graph Sampling

Abstract

The Linked Data cloud has grown to become the largest knowledge base ever constructed. Its size is now turning into a major bottleneck for many applications. In order to facilitate access to this structured information, this paper proposes an automatic sampling method targeted at maximizing answer coverage for applications using SPARQL querying. The approach presented in this paper is novel: no similar RDF sampling approach exist. Additionally, the concept of creating a sample aimed at maximizing SPARQL answer coverage, is unique. We empirically show that the relevance of triples for sampling (a semantic notion) is influenced by the topology of the graph (purely structural), and can be determined without prior knowledge of the queries. Experiments show a significantly higher recall of topology based sampling methods over random and naive baseline approaches (e.g. up to 90% for Open-BioMed at a sample size of 6%).

Combining Vertex-Centric Graph Processing with SPARQL for Large-Scale RDF Data Analytics

“…SamplD [2] is a pipeline for sampling RDF graphs. Given an RDF graph, SamplD applies a set of graph operations using Apache Giraph [19] and PIG [20].…”

“…The third use case, SamplD [2] (see Section 2.2.3), starts by loading the RDF data into HDFS and transforming it into an unlabeled directed graph, which is then processed by Apache Pig and Giraph to evaluate PageRank and degree centrality. The results are stored in RDF format with each triple assigned a score that denotes its importance.…”

“…The emerging trend for RDF analytics [1], [2], [3], [4] goes beyond simple pattern matching, and requires complex pipelines that combine a variety of graph algorithms with SPARQL. For example, Qu et al [1] mine biological data to identify new targets for existing drugs.…”

“…On the results they execute graph centrality algorithms to identify key biological entities. Rietveld et al [2] focus on advanced graph sampling. First, they compute the degree centrality 1.…”

SPARTex

“…SPARQL is the standard RDF query language; it allows users to express subgraph pattern matching queries. On the other hand, there is an emerging new type of RDF analytics [9,10] that require the combination of generic graph search operations with SPARQL patterns. Examples for such operations are reachability queries, centrality analysis and community detection.…”