Abstarct: In question answering systems, users ask their questions in a natural language and the system returns the one or more possible answers. However, in question answering systems there is no continued dialogue between the user and the system and the users cannot explain their meaning if it is misunderstood by the system. Hence, getting access to the needed information is not so easy for the user. Adding interaction to these systems enables users to ask related questions and to clarify their meaning and increases the accuracy of the answers returned by the system. The purpose of this research is to design an interactive question answering system which uses statistical techniques to extract the needed information from unstructured texts. In contrast to other systems which make use of linguistic techniques, this system does not use grammar and semantic techniques and so, it is independent of language. It can answer the questions asked in any language if the appropriate dataset in that language is given to it. In this research, the designed system makes use of a Persian dataset for training and testing. The efficiency of the designed system was assessed by both qualitative and quantitative measures. The obtained scores indicate a 22.3% increase in the accuracy of the answers given by the designed system in comparision with the answers given by the QA system. Moreover, the examination of the users' comments, clearly showed their satisfaction of their interaction with the designed system.