Abstract
When users query a SPARQL endpoint, they normally face an empty text box in which they have to write the desired queries. This obstructs the process of obtaining the data they want, since users rarely have any assistance in querying a (possibly huge) RDF database. In this paper we report a deep analysis of the server log files that record the queries that users send to the SPARQL endpoints, focusing in the Bio2RDF cluster. This log analysis reveals the large number of repeated queries that users submit, and how they pursue a trial and error process by adding and removing operations from the submitted queries to obtain the desired results. We also show how users try to connect to other RDF datasets in the Linked Open Data cloud. Our results offer insight into the interaction between users and a schema-light RDF dataset, and secondly, suggest improvements to SPARQL server optimizations in terms of optimization and results caching.
| Original language | English |
|---|---|
| Pages (from-to) | 196-204 |
| Number of pages | 9 |
| Journal | CEUR Workshop Proceedings |
| Volume | 1378 |
| Publication status | Published - 2015 |
| Externally published | Yes |