Colloquium at Stanford
The Unfinished Revolution

Memorandum

Date: Mon, 24 Jan 2000 17:00:00 -0800

From:   Eric Armstrong
eric.armstrong@eng.sun.com

To:     unrev-II@onelist.com
Subject:   Augment/NLS

Quoting from Jeff Miller Tue, 25 Jan 2000 10:55:43 +1100...

I too would like to see a system design discussed. specific examples would be good.

As a start, how about taking a specific document and comparing possible methods on encoding, storage, searching, etc ? these are generalised concepts. We could then move onto referring to specific points (paragraphs, lines, sentences, words), annotating the document, versioning, summarisation, etc?

Anyone have a list of features we can test against to see if we meet all the requirements?

At the moment I'm leaning towards an xml encoding ...

While writing this I went searching for some possible examples and came across the following;

Annotated XML Specification

http://www.xml.com/xml/pub/axml/axmlintro.html

  1. My initial start in that direction was "Towards a DKR". Feedback on that elicited several interesting requirements. I paln to write up a preliminary version tonight. Doubtless several more iterations will be required.

  2. XML does indeed seem like a good starting point. Apparently early versions of the XML spec were intended for this purpose, but the end result was constrained. We'd sure love to know what the early spec looked like.

  3. Unfortunately, the Annotated XML Spec is not a spec for capturing annotations in XML, but rather an annotated version of the XML spec. Interestingly, the annotations were captured in an XML document, using links to the spec. A Java program then broke up that document into a multitude of annonations, and inserted links to them into the XML spec.


Sincerely,


Eric Armstrong
eric.armstrong@eng.sun.com