transforming unique nodes thru XSLT

I have an XSL Transformation question. I hope I can raise XSLT issues here. I am new to XSL and want to write a template
to transform only unique nodes from a well formed XML document to HTML.

Below is a small snippet of this well formed document which I want to transform
<xyz>
    ...
    ...
    <a name="school">
        <a name="grade5">
            <b item="pencil"/>
            <b item="paper"/>
        </a>
        <a name="grade10">
            <b item="pen"/>
            <b item="paper"/>
            <b item="backpack"/>
        </a>
    </a>
    <a name="office">
        <b item="cuibicle"/>
        <b item="paper"/>
        <a name="engg">
            <b item="pc"/>
            <b item="backpack"/>
        </a>
    </a>
    ....
    ...
</xyz>

Now, from this XML document I want to create a HTML document which would be a list of all the 'b' element's attributes
(i.e. item). As you see there are a few items which are repeated in the XML document. If I want only the unique elements
depending on b's attribute item then how should go about writing the XSLT template. Is this possible? Maybe I have to parse
the entire document more than once to see if a particular node is unique, but how to do this? Any help!!!

PS: I am using  XT (Ver 19990822) by James Clark for the transformation.

Thank you,
Anand