[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

Re: "Maximize the ratio of content to markup" What's the under

• From: "bryan rasmussen" <rasmussen.bryan@g...>
• To: "Costello, Roger L." <costello@m...>
• Date: Wed, 5 Mar 2008 13:40:54 +0100

>
>  For example, this is not good design:
>
>  <div>
>     <div id="Main">
>         <p>Hello World</p>
>     </div>
>  </div>

Sure, but the technique of having a wrapping outer div is pretty
traditional given the need to make content viewable across multiple
browsers with various levels of support for various standards.

Second of all there are different levels of semantic markup. Search
engines sure don't look for high levels of semantic meaning (mainly
because there isn't such a widespread level of it) by which I mean
they don't look for XML.


>  The outer div is providing no benefit.  It can be more simply expressed
>  as:
>
>  <div id="Main">
>     <p>Hello World</p>
>  </div>
>
>  The later version provides a higher ratio of content to code (tags).
>  And from the quote above, search engines rank higher documents with a
>  higher ratio of content to code.

SVG. The content is the code.




>  What is the underlying principle?  Why do search engines prefer
>  documents with a higher ratio of content to markup?
Because they are free text search engines in an untrustable
environment, where they must use complicated techniques to find out
what the meaning of things are buy the text content of html pages.

Furthermore modern search engines seem to prefer a weighting of
linking relative to content and markup. So the original statement does
not apply.
>  Can the principle be applied to XML data design?
>
Not really.
>  For example,
>
>  This is not good design:
>
>  <Author>
>     <Name>Paul McCartney</Name>
>  </Author>
>
>  The Name element is providing no benefit.  It can be more simply
>  expressed as:
>
>  <Author>Paul McCartney</Author>
>
>  The later version provides a higher ratio of content to code (tags).
>

I have to say no.

I would argue this is not a good design:
<Book>
<Author>
<AuthorName>Paul McCartney</AuthorName>

</Author>
<BookName>Sir Paul wrote a book!?</BookName>
</Author>

but it is a design one sees  a lot which I think is actually based on
the needs of maintaining large XML Schema libraries.(personal opinion,
no one agrees)

The requirements for what makes good semantic markup, whether for
semantic markup aware search and what makes good free text search that
tries to use some semantic rules to figure out the level of importance
of particular documents are pretty different from each other.

>  What do you think?  Is there a principle of data design being
>  illustrated here?
No, there is are several principles of datamining in a large untrusted
hypermedia environment (where you can not trust the quality of the
data at all) being illustrated
>  Can you articulate the principle?

Probably not without being fired given I have a large non-functioning
rendering server to deliver in two days and I should be working :(

Cheers,
Bryan Rasmussen

• Follow-Ups:
  • Re: "Maximize the ratio of content to markup" What's the underlying principle?
    • From: "bryan rasmussen" <rasmussen.bryan@g...>

• References:
  • "Maximize the ratio of content to markup" What's the underlying principle?
    • From: "Costello, Roger L." <costello@m...>