In this current world, we come across digital information available on the web like e-business transaction, e-shopping, e-learning etc., These XML documents are outsized and so the datasets returned as a answer query is also large, which in turn is a complicated process for the retrieval of interpretable knowledge. The query answering system has achieved its potential in creating queries and accessing XML documents by reducing the tremendous size of data. In this paper, we use an optimized Tree based Association Rule (TAR) to retrieve the results for the queries posed by the users on the XML document. TAR files provide intensional information on the structure and content of XML document. In addition we propose a method to dynamically update the TAR files when the dataset changes.