TR: [newsml-g2] subject codes and NewsML-G2
FYI I forward (with her permission) a post from Lena T. about the subject codes shown in one of the samples I uploaded yesterday.
De : Laurent LE MEUR
Envoyé : mardi 17 juillet 2007 17:33
À : 'Lena Tenenboim'
Objet : RE: [newsml-g2] Digest Number 175
Dear Lena ,
They are in fact two facets in your question:
- One related to the syntax of the current IPTC Subject codes taxonomy,
- One related to the generic way of representing the subject of a news item.
It is true that the current IPTC Subject codes taxonomy has been implemented with a coding system that represents the hierarchy inside the coding itself.
Thus 01026002 is more specific than 01026000 which is more specific than 01000000.
- This is not a generic rule for the construction of taxonomies (some are based on English tokens, others are a pure list of incremental codes). Therefore, depending the taxonomy, it is possible that you get no information about the ancestors of a term from its code; in such a case, including the chain of terms in a newsItem can be helpful for the recipient. For sure, if the recipient has knowledge of the taxonomy tree, he can infer the hierarchy by itself (this is called auto-posting in English I think).
- This will certainly change in the future for the IPTC Subject codes taxonomy itself (there is a newscode-G2 coming). The reason why this will change is not part of the discussion around NewsML-G2: please move to the http://tech.groups.yahoo.com/group/newscodes/ Yahoo! List if you want to discuss this matter.
Maybe I should remove the comment for the example: this is not really a NewsML-G2 issue. I put it to show that including a term and its ancestors was possible.
Can I forward your post to the list, so others can benefit from the answer?
Laurent Le Meur
De : Lena T.
Envoyé : lundi 16 juillet 2007 14:20
À : Laurent LE MEUR
Objet : Re: [newsml-g2] Digest Number 175
Dear Laurent LE MEUR,
I've checked the posted examples of NewsML-G2.
My question is regarding the IPTC Subject codes taxonomy.
You've written in the comments that "one should not assume that the hierarchy is derived from the code structure"