Position Paper on Web Query Languages: 

XML Query Languages of, by and for the People

lars berg, inktomi corporation
lars@inktomi.com



1.      Table of Contents

        Introduction
        Position
        Issues
        Conclusion


2.      Introduction

As the web grows a more intelligent and robust system is necessary to 
allow metadata to be both represented and queried as represented within 
web pages (and even when not represented within the pages, as in the 
case of pages that consist entirely of metadata about other pages).  
Utilizing XML to improve the manner in which page metadata is treated 
will allow users to specify any range of relevant vertical 
field-specific information, whether it be focused on improved treatment 
of chemical structures or a more in depth way to refer to oceanographic 
information.  Presently, no standard exists that allows web site design 
teams to encode their pages (or parts of pages) with universally 
available metadata.  Herein we represent one voice in the search and 
indexing field; our perspective hence is shielded by our own standpoint.



3.      Position

In brief, a human readable (and writable) canon of XML objects designed 
with the intent of allowing the growth of a free and web-supported 
community of metadata structuralists needs to be supported.  Different 
groups have differing needs; hence, the structure needs to be adjustable 
enough to take into account the various degrees of knowledge and 
specialization inherent in vertical fields of expertise.  The structures 
need to foster the growth of knowledge communities that are free to 
create their own super (and sub) structures that are nonetheless based 
upon a standard XML implementation.  Growth of such communities is 
somewhat inevitable and relevant architectural plans need to be put in 
place otherwise much of the relevant and humanly important data on the 
web will be lost in a sea of skantily clad models.  As the web grows we 
must prepare ourselves to embrace the amazingly robust but chaotic 
nature of human beings; that is, we must provide a living online 
metadata structure that allows the creation and explosion of knowledge 
communities. 



4.      Issues

Metadata, alas, is not impervious to the natural degeneration of 
superstructures.  Meta-spam will have to be treated.  Objects that refer 
to objects, each with independent metadata (for example Dublin Core 
data) will need to have scripted relationships otherwise we could simply 
unleash another metadata flood on top of the current internet data 
flood.  Standards will need to be picked and adjudicated fairly in a 
court of human knowledge (real or not) with strong vision for the 
future, avoiding partisan and parochial battles that will surely doom 
the effort.


5.      Conclusion

A new meta-architecture needs to be proposed by a consortium of 
academics, industry figures and political figures.  Something akin to 
the Warwick Framework, perhaps a metadata package / container metaphor 
will be sufficient to drive those who create online knowledge in the 
future to filter their requests through a fine metadata strainer -- 
hoping to both remove chaff and add a refined, more searchable (and 
hence retrievable) aura to the creation of mind objects in the coming 
decades.  Nobody likes a library without a good card catalog.