Which of the following SPARQL queries are similar? why?
Renata Dividino, Gerd Gröner · 2013
Abstract. Linked data on the Web can be accessed by SPARQL queries. Previously executed queries to SPARQL endpoints are valuable information sources about the underlying data structure and schema of data sources. These queries reveal how resources are related to each other and they reflect the user interests on the data. Therefore, methods for query logs analysis provides a basis for extracting relevant and qualitative data from the LOD cloud. These query logs analysis rely on similarity measures for SPARQL queries in order to determine whether a query is related to another query. However, there is no commonly agreed notion of similarity for SPARQL queries. Instead, there are several measures that not only vary computational complexity and output values, but also on their purpose. In this paper, we present a comparison and a discussion of the existing similarity measures for SPARQL queries in the literature. Our results are guidelines designed to provide guidance on the application of similarity measures for comparing SPARQL queries. 1