Profiling of Services and Queries in SeCo

The efficient execution of Search Computing queries requires accurate statistics about invoked services (e.g. response time and results characterization) and query usage (e.g. which queries are executed most). The firsts are necessary for query planning and optimization, the latter for adapting the system by allocating resources to make the execution of the most frequently used queries more efficient.
The aim of the proposed project/thesis is to develop a solution to efficiently acquire and monitor statistics about services and queries, by: ul

• identifying the statistics to be collected and the sources where statistics can be gathered from, which include both the log files collected by the system during query execution (e.g. measured response times) and external sources available on the Web (e.g. the Seekda! portal);
• developing a system to efficiently collect statistics and keep them updated;
• developing a mechanism to monitor services for deviations from previously collected statistics (e.g. a sudden increase in response time) in order for the system to quickly react.

Previous knowledge of Web service technologies and database systems is useful, although not required.