I'm trying to represent a linked relationship among between nodes as
follows:

<!-- Top level nodes -->
<Cities>
    <City>Los Angeles</City>
    <City>San Francisco</City>
    <City>Seattle</City>
    <City>Washington DC</City>
</Cities>
...
<Dimensions>
    <Weight>
        <Metric>
            <Tonnes>
                2.5
            </Tonnes>
        </Metric>
    </Weight>
    <Height>
        <Metric>
            <Metres>
                6.4
            </Metres>
        </Metric>
    </Height>
</Dimensions>
...

<!-- User level nodes -->
<Address>
    <Line1>...</Line1>
    <Line2>...</Line2>
    <City>/Cities/City[2]</City> <!-- Links to top level nodes -->
</Address>

<Truck>
    <Capacity>
        <Weight>/Dimensions/Weight/Metric/Tonnes</Weight>  <!-- Links
to top level nodes -->
        <Height>/Dimensions/Height/Metric/Metres</Height>  <!-- Links
to top level nodes -->
    </Capacity>
</Truck>


Is it possible to represent this relationship both at design-time (in
the schema/XSD) and at run-time (in the XML document)?

Is there a way to do this in XLink/XPointers such that in Java code I
could, for example, transparently use the linked objects as follows:

System.out.println(address.city);
System.out.println(truck.capacity.weight);
System.out.println(truck.capacity.height);

(Unless I absolutely have to, I don't want to write additional code to
"understand" the semantics of the links to the city, height and weight.
That is, in a Java program, the links should "automatically" refer to
and retrieve the top-level data as "objects" rather than just plain
strings.

Thanks in advance,
Kunal