In the past few years, with many organisations providing web services for business and communication purposes, large volumes of XML transactions take place on a daily basis. In many cases, organisations maintain these transactions in their native XML format due to its flexibility for xchanging data between heterogeneous systems. This XML data provides an important resource for decision support systems. As a consequence, XML technology has slowly been included within decision support systems of data warehouse systems. The problem encountered is that existing native XML database systems suffer from poor performance in terms of managing data volume and response time for complex analytical queries. Although materialised XML views can be used to improve the performance for XML data warehouses, update problems then become the bottleneck of using materialised views. Specifically, synchronising materialised views in the face of changing view definitions, remains a significant issue. In this dissertation, we provide a method for XML-based data warehouses to manage updates caused by the change of view definitions (view redefinitions), which is referred to as the view adaptation problem. In our approach, views are defined using XPath and then modelled using a set of novel algebraic operators and fragments. XPath views are integrated into a single view graph called the XML Fragment Materialisation (XFM) View Graph, where common parts between different views are shared and appear only once in the graph. Fragments within the view graph can be selected for materialisation to facilitate the view adaptation process. While changes are applied, our view adaptation algorithms can quickly determine what part of the XFM view graph is affected. The adaptation algorithms then perform a structural adaptation to update the view graph, followed by data adaptation to update materialised fragments.