We will now look at a concept using typed links for indexing of
data. The purpose of indexing by links is to group the data by subjects. What a
subject is will be determined through links. To be able to use
concrete examples we will specialize on terms used in
news-publication.

The Objects we will talk about are simply documents related to any
topic. We will not look closer to the content of these documents, we
only want to look at link-relations between these documents describing
relationships between them.

The relations used for linking will be typed. According their type
the relations may be symmetric or transitive, we may even have
multi-relations (BETTER NAME???) pointing from one document to many documents.

While looking on such a Network of linked Documents there may be
isolated sub-networks linked together in any way we are interested in:
These clusters will be called subjects.

Further relations are related to the publication-date of the document
and will be called implicit , because they will not have to be
providet by the author.

There should only be a small amount of well chosen types for the
links. There are two reasons:

- The more types we have, the smaller the clusters may get and the
  more complicated search may get.

- If types are free to choose, the same problems as found in
  keyword-indexing may occur; selected terms may be ambigous and
  therefore the search for a sertain topic may produce too much noise,
  On the other hand different terms with similar semantics can be used
  and it would nessesary to know all of them or make use of thesauri
  to cover the whole subject.

[pending: what happens, if using a thesaurus for the terms used for
link-types to pseudo-extend the link-type-set?]


Some suggests for typing
========================
We will now look at a simple but powerful set of links-types and then
examin how powerful it is as basis for describing topics:

- The "enforces" link-type: <DEPRECATED>
  This gives us the feature to express that a document may enforces
  the content of another document. It is possible to use 1 to many
  links for this type. It seems to be a very basic type that  can be
  extendet by other types.

- The "disclains" link-type: <DEPRECATED>
  This type is meant to be the opposite of "enforce". It is not
  implied, if "enforce" does not match and may therefore be modelled.

- The "relatedTo" link-type:
  It is possible that both previous types can be described by "related
  to" - a very general type - and therefore these types are deprecated
  right now. I will define "relatedTo" as symmetric relation.
	
- The "cites" link-type(extends "relatedTo"):
  This is a relation which is often used in the domain of publication.
  The symmetric nature of "relatedTo" is overridden here to be symmetric.
  The relation is not transitive.

- The "partOf" link-type(extends "relatedTo"):
  This relation does not specify that a document is a part of the other document,
  it states that one document covers a part of the subject covered by the other 
  document. This relation is antisymmetric and transitive. There is also a
  inverse relation to this on; it is called

- The "wholeOf" link-type(extends "relatedTo"):
  It is also antisymmetric and transitive.

- The "equals" link-type(extends "relatedTo"):
  One documents "equals" another, if they cover exactly the same
  subject. This relation is symmetric, transitive and reflexive.
  Maybe someone would want to point out the opposite. Therefore it may
  be interesting to add a "notEqual" relation.

- The "narrower" link-type(extends "partOf"):
  This link is similar to "partOf", but the content of the document
  being narrower may be more precise than in "partOf"-relations.
  "narrower" is transitive and symmetric. The opposite relation to
  "narrower" is

- The "wider" link-type(extends "wholeOf"):
  see "wholeOf". It even seems that "wider" is a synonym for "wholeOf".







APPENDIX:
========

german translation of the relations:

- zitiert (inv(?).: wird zitiert von) 		[cites/citedBy]
(- ergnzt (inv.: dementiert)			[enforces/dements] ??? GEHT DAS ZUSEHR AUF DEN INHALT EIN ???)
- teil von (inv.: das ganze von)		[partOf/wholeOf]
- ist gleich (inv.: ist ungleich)		[equals/notEqual]
- ist spezieller (inv.: ist allgemeiner)	[narrower/broader]
