A corpus and mutual information based ontology learning method
Xiang Li · Microcomputer & its Applications · 2010
A computer aided domain ontology building method from free Chinese domain text is developed.The key steps are exemplified by a group of experimental data.The domain word elements are identified semi-automatically through comparing the appearing probability of them in sample text with which in the corpus.Then the domain words composition ability among the identified word elements been calculated using mutual information.Domain ontology words and the relationship among them can be generated through using the data produced from the steps.A directed weighted word graph can also been generated,which could be used as a good visual ontology draft for the ontology builders works on.The result could be used to support large scale content based knowledge management.