Read out

Guest Talk: Sage: Web Preemption for Public SPARQL Query Services

Dr. Hala Skaf-­Molli 

Date/Time: 12.12.2018, 15:30 

Loca­tion: TC.4.18 


After a short repre­sen­ta­tion of the rese­arch topics of GDD team of the Univer­sity of Nantes. I will present our ongoing rese­arch on Sage: Web Preemp­tion for Public SPARQL Query Services. To preserve the avail­a­bi­lity of SPARQL query services, data provi­ders enforce quotas on server usage. Queries which exceed these quotas are inter­rupted and deli­vers partial results. Such inter­rup­tion is not an issue if it is possible to resume the query execu­tion after­ward. Unfor­t­u­n­a­tely, there is no preemp­tion model for the web that allows suspen­ding and resume SPARQL queries. In this talk, I will present a model for Web preemp­tion and Sage, a SPARQL query service that supports such preemp­tion. SPARQL queries are suspended by the web server after a fixed time quantum and resumed upon client request. Web preemp­tion is trac­table only if its cost in time is negli­gible compared to the time quantum. Thus, the chal­lenge is to support full SPARQL query language while keeping the cost of preemp­tion negli­gible. Exten­sive expe­ri­men­ta­tions demons­trate that sage outper­forms exis­ting approa­ches by several orders of magni­tude in term of average total query execu­tion time, time for first results and data trans­ferred per query. 


Dr. Skaf-­Molli asso­ciate professor, HDR [1] at the Univer­sity of Nantes and member of the GDD Team [2] at LS2N [3]. Dr. Skaf-­Molli works on decen­tra­lized semantic data manage­ment: semantic data synchro­niza­tion, federated SPARQL query proces­sing, and Servers/ Clients colla­bo­ra­tion for execu­ting SPARQL

[1] http://pages­



Back to overview