Understanding an input in inference-based multistrategy learning

The inference based learning system uses a multistrategy approach that integrates different learning strategies for concept learning and knowledge refinement. During the understanding process, the system takes an input example and builds an explanation tree to connect the input to the system´s knowledge. The understanding process may face a situation where several rules are available for an explanation of a given concept or an incomplete knowledge base which may lead to a failure of building an explanation tree. Thus, the resulting explanation tree can be several or none without a proper control strategy. The paper contains a description of the control strategy which copes with the above problems and allows one to build the most useful explanation tree in terms of generality and plausibility