Extracting information from semi structured documents is difficult task. It is more crucial as there is a huge amount of digital information on the Internet is growing rapidly. Sometimes, documents are often so large that the data set returned as answer to a query may be large to even convey interpretable knowledge. This paper describes an approach which takes RSS feeds as input for which Tree-based Association Rules (TARs): mined rules are used. It provides more approximate and intentional information on both the structure and the contents of eXtensible Markup Language (XML) documents which can then be stored in XML format as well.