The main KA problems concerns decisions about what to include and how to encode it given the numerous domain sources (multilingual texts, pictures, etc.). At present, we are actively evaluating our approach and the KB by comparisons to other existing approaches. We are also studying how to encode more elaborate semantic relationships by using more conceptual relations and their influence on the inheritance mechanism.
Another interesting question is the appropriate explanation of semantic similarity and difference in a more elaborate manner, i.e., not relying solely on the distance between the two concept types in the type hierarchy. Another source of measuring the similarity of two concepts are the domain facts encoded in contexts and conceptual graphs. For instance, two concepts are comparable based on the thematic roles of a process where they participate. Such explanations, however, require deeper investigation of the necessary content of the domain knowledge and its transformation into semantics of the future explanation.