Abstract
Text documents are one of the means to store information. These documents can be found on personal desktop computers, intranets and in the Web. Thus the valuable knowledge is embedded in an unstructured form. Having an automated system that can extract information from the texts is very desirable. However, the major challenging issue in developing such an automated system is a natural language is not free from ambiguity and uncertainty problems. Thus semantic extraction remains a challenging task to researchers in this area. In this paper, a new framework to extract semantics for information extraction is proposed, where possibility theory, fuzzy sets, and knowledge about the subject and preceding sentence have been used as the key in resolving the ambiguity and uncertainty problems.