Abstract: To improve the ability of the analysis of knowledge base, an ontology framework that introduces functional analysis techniques was proposed. In this paper, a functional analysis method was proposed based on the problem model with the material-field model of TRIZ innovation theory. The proposed method can effectively help users to analyze and solve existing problems. A CRF's named entity recognition method was also proposed based on patent text of rice breeding to achieve automatic extraction and addition of instance, which effectively improve the construction quality and efficiency of the domain knowledge base. The experimental results show that the knowledge base of rice breeding can effectively analyze user's problems and give reasonable suggestions.