Popis: |
The proposed hyper knowledge discovery system is designed to facilitate domain users to interactively discover knowledge from complex data sets. At the first stage, this system integrates data preprocess, lexicon establishment and database construction together. Then it provides domain users with flexibility and fuzzy mechanism to retrieve information and mine knowledge interactively. The system introduces an approach to obtain joint probability distributions of an arbitrary number of variables of multi valued states. This is fundamental for mining correlation and causal relations from the complex data sources. To attain these objectives, the system creatively exploits the functionalities of the relational database and SQL engine to reduce programming while achieving good performance. All the functionalities are online and hence facilitate resource and knowledge sharing. |