S-CUBE - Knowledge extraction from service usage

Silvestri F., Nardini F. M., Tolomei G.

Process Mining  Service Log Mining  Sequential Pattern Mining 

Data is everywhere. Computer systems keep track of activities of users in the form of log files. Ranging from system logs on Web servers to logs collected by large-scale service based applications, this type of data represents a goldmine of knowledge that, once extracted, can help the stakeholders of the whole system to understand better if, and how, the application can be improved. To this aim, data mining consist of a set of techniques aiming at extracting patterns from large data sets by combining methods from statistics and artificial intelligence with database management. With recent tremendous technical advances in processing power, storage capacity, and inter-connectivity of computer technology, data mining is seen as an increasingly important tool by modern business to transform unprecedented quantities of digital data into business intelligence giving an informational advantage. Service-centric systems are said to be flexible and dynamic. To support this flexibility, event processing mechanisms can be used to record which events occur within the system. This includes both basic "service events" (e.g., service is created) and complex events regarding QoS (e.g., average response time of service X has changed) and invocations (e.g., service X has been invoked), supporting complex event processing. Users can subscribe to various events of interest, and get notified either via email or Web service notifications (e.g., WS-Eventing). Such notifications may trigger adaptive behavior (e.g., rebinding to other services). Service Oriented Architectures (SOAs) are thus complex infrastructures consisting of thousands or millions of service interacting together in order to achieve complex operations (tasks). Service invocation logs are file tracing the interactions between services. As in other contexts, data mining techniques can be thus applied in order to derive useful knowledge. Such knowledge can be spent in order to enhance both effectiveness and efficiency of the overall infrastructure. The same approach within other fields like, for example, the Web domain is proven to be effective. The knowledge extracted by means of data mining techniques from query logs (files containing the interactions of the users with the search engine) is the first way a search engine improve its performances in terms of effectiveness and efficiency. In this deliverable we thus investigate how useful knowledge can be extracted from service logs and possible ways of applications within the SOA context. In order to do that as one of the case studies we will use logs from search engines activities to extract knowledge regarding users activity.

Source: Project report, S-CUBE, Deliverable #PO-JRA-2.3.7, 2011