Rule of thumb specifications Polus Framework Modeling of Information

Generation of Knowledge base

In Polus, the details of the action object are generated using a combination of rule-of-thumb specifications and learning. Figure 6 shows how the prefetch object is internally maintained by Polus. The details of the action object may not all be available when the system starts off. A part of the information is generated using user-defined specifications (rule-of-thumb) and rest is generated using learning functions (as will be explained in this subsection). There is a ``confidence'' value associated with every piece of generated information. The confidence value is based on the error value of the learning algorithm that signifies the difference between the predicted value and the actual observed value.

Figure 6: Prefetch action object internally maintained by Polus