Field association (FA) terms are a limited set of discriminating terms that offer humans the knowledge to identify fields exiting in the document (text). Field association knowledge base is composed of FA terms and their potential hierarchical relationship of the fields they belong to. The main purpose of this research is building Chinese FA knowledge base. After this, the new knowledge base is tested through a system which can imitate the process whereby humans recognise the fields by looking at a few special terms. In doing so, a novel approach makes use of the structured knowledge in Chinese Wikipedia. A totally new Chinese FA knowledge base is built including 115,696 FA terms. The resulting FA knowledge from this knowledge base is applied to text categorisation. The average accuracies, 97.7% and 89%, are both higher than values obtained by SVM.