| TITLE: | DISPOSITION OF COMMENTS ON SC 18 N 5468 - ISO/IEC CD 13250: Information Processing - SGML Applications - Topic Navigation Maps |
| SOURCE: | HyTime RG |
| PROJECT: | JTC1.18. |
| PROJECT EDITOR: | Martin Bryan |
| STATUS: | Approved for distribution to NBs |
| ACTION: | For information |
| DATE: | 15 November 1996 |
| DISTRIBUTION: | WG8 and Liaisons |
| REFER TO: | |
| REPLY TO: | Dr. James David Mason
(ISO/IEC JTC1/SC18/WG8 Convenor) Lockheed Martin Energy Systems Information Management Services 1060 Commerce Park, M.S. 6480 Oak Ridge, TN 37831-6480 U.S.A. Telephone: +1 423 574-6973 Facsimile: +1 423 574-0004 Network: masonjd@ornl.gov http://www.ornl.gov/sgml/wg8/wg8home.htm ftp://ftp.ornl.gov/pub/sgml/wg8/ |
Australia votes with APPROVAL on this document, however, would like the following comments to be taken into account
Topic Navigation Maps must be interoperable with Text Encoding Initiative elements (TEI DTD) like "taxonomy" and "category". This compatability should be demonstrated with examples using TEI DTD elements, perhaps in an non-normative annex: TEI is very widely-used, and a useful testbed for proof-of-concept. CD 13250 should follow TEI terminology, if there is any duplication.
Response:
It is currently unclear how easy it will be to introduce the HyTime-based facilities in CD13250 into TEI in a way that will be compatible with TEI as it stands. It will be possible to link a HyTime Topic Map to a TEI elements such as "taxonomy" and "category". An annex showing how HyTime Topic Maps and TEI classifications can be interconnected will be considered.
The assignment of a link to a topic seems unneedfully unary: a link is either present or not.
It is common practise that, at least initially, large text corpora are marked up or indexed automatically based on pattern-matching. For English and other languages with a large number of homographs (words that are spelled the same) there needs to be some indication that a link may not be relevent to the topic. This ambiguity is an important and continual fact of life in this area and must be allowed for.
Furthermore, it is an attribute of thesauruses that some topics are more or less relevant. Also, it is established practise in text indexing and search engines to allow "fuzzy searches" or "intuitive searches" that use the rankings of related terms as weightings or in scores to determine which links to return as "hits" from a search, and in what order.
So some further attribute should be provided to allow topics to be weighted. The nature of the score should of course be DTD dependent. Australia suggests the following (though perhaps the quantum needs to be defined better with a HyTime FCS)
<!attlist CAph.semanticAssignment
...
weight -- indicates relevence to topic or certainty
of information.
0 = means probably irrelevent.
100 = means definitely relevent.
If a weight is more than 100, it can indicate extreme
or compelling relevence, though weights greater
than 100 may be deemed to be 100 by any application.
--
NUMBER 100
>
It also might be useful for the editor to enquire of industry vendors like OpenText and Fulcrum what weight mechanism is suitable for ready importation of Topic Navigation Map documents into their off-the-shelf systems.
Alternatively, it may be desirable to have two types weights: one for relevence, one for reliability or certainty. This is used in some criminal and military intelligence classification systems.
On a theoretical level, this weight properly belongs to the Semantic Assignment, since it does not define a different type of relationship, merely an attribute of the particular location with respect to that relationship. So it is not correct to mimic the weighting by providing other relationships each with an intrinsic weight instead (e.g. the relationships "definitely defined by" and "probably defined by") and to define a realionship between these weighted relationships (e.g. "definitely defined by" and "probably defined by" are related as both being "defined by" relationships). Such topic schemas may be useful, but they do not provide the ease or open-endedness of the NUMBER system.
Response:
It will be made clear that Topic Map links are not unary by clearly defining each anchor as a list (which could contain only one member). The ability to use queries to locate references will also be clearly specified.
A mechanism for specifying the weighting of semantic assignments according to relevance, reliability and any other user-specified weighting scheme will be added to the specification. Identifying the domain of a searchable set of document elements/entities is provided as a standard function of HyTime. Use of this functionality to control the relevance of topic maps will be investigated.
7th paragraph beginning " The first anchor, called the topic anchor....
The paragraph is made difficult to read due to the excessive use of the words "itself", "it" and "its". The intent of the paragraph would be made much clearer if it was rewritten.
Response: Accepted; clause will be rewritten.
The National Body of Japan approves on ISO/IEC JTC1/SC18 N5397 (ISO/IEC CD 13250) with the following comments.
"Related Standards" should be "Normative References".
Response: Accepted
"CApH" should be replaced with "TNM".
Response: Accepted
This standard should includes an informative Annex which shows some examples of navigation maps.
Response: Accepted, though examples will be simplified/partial ones that are easy to follow.
This standard should includes an clause which specifies the architectural base declaration according to the HyTime TC Annex C.
Response: Accepted
P.6 default attribute of agtrav cannot be correct.
Response: This attribute is no longer relevant as it will be replaced by the new linktrav attribute for HyLinks.
The UK APPROVES the draft with the following comment.
The UK anticipates that the effect on the current CD of changes expected to be proposed in the HyTime TC will be considerable. Consequently, the UK considers that it would be inappropriate to comment in detail at the present tine.
Response: The CD will be rewritten to define topic maps as an example of a HyLink rather than an HyTime independent link (ilink).