You can use XPath.

/TaskManager/Project/Job[@id='4']

This asks the parser to start with the root "TaskManager" element, which also should be the first element in the document because we used the "/" preceding the element name. The TaskManager element should have a child element "Project", whose identity is insignificant... the only criteria is that TaskManager have a child element Project. Similarly, Project must have a child element "Job". The criteria you are specifying (kind of like a WHERE clause in SQL) is that the Job element must have an "id" attribute with value 4. If all of these criteria in the path are met, then a pointer to the "Job" element node is returned. From there, you can access its child elements or any descendant, or even ask it for its parent or ancestors. XPath is a very powerful language, but requires breaking out of the SQL mindset of returning tabular resultsets. XPath returns pointers to nodes.