As XML (eXtensible Markup Language) is becoming a standard widely applied in data representation and data exchange in web service. In this paper, we propose a novel approach to type-ahead keyword search in XML data, call TakeXIR. The IR-style approach basically utilizes the statistics of underlying XML data to address the following challenges in XML IR system: (1) Identify the user search intention, i.e. identify the keywords to express user interests and identify nodes user wants to search for and search via. (2) Resolve keyword ambiguity problems: synonyms and polysemy exist in natural language, and a keyword can appear as the text values or tag value of different XML node and carry different meanings.