Abstract: A model of semantic based text formalization, the context framework model（CFM） is presented in this paper, which is three coordinate and describes the text as domain, situation and background Based on the context framework, a text character extracting algorithm is developed The algorithm includes domain extracting which uses 4 element array, situation extracting which is triggered by domain sentence category, and background extracting which focuses on the confusion of the commendatory and derogatory based on object semantic stand net As a result, the CFM is a very good model for text retrieval, and the algorithm can remarkably improve the efficiency of text retrieval