|
计算机科学 2004
Representation, Acquisition and Analysis of Psychological Commonsense Concepts
|
Abstract:
Psychological commonsense is needed in knowledge-intensive applications, such as natural language processing, machine intelligence and real-world reasoning. In this paper, centred on psychological commonsense knowledge, we introduce a framework of acquiring and analyzing psychological commonsense concepts, since such concepts are actually the foundation on which we finally capture psychological commonsense knowledge. In our framework, we have developed a method for acquiring elementary concepts and their semantic relationships based on psychological categories. We introduce five semantical classes of psychological attributes for psychological commonsense concepts, and discuss how to acquire psychological attributes and their relationships based on these semantical classes. To auto-matically generate compound psychological commonsense concepts, we introduce a rule-based method that generates compound concepts from elementary ones and their pre-defined psychological attributes. We also present an algorithm for checking the inconsistency and redundancy of the concepts.