Knowledge vs. Information

The more people I meet, the more definitions of knowledge vs. information are collected.
Here are three statements:

1. Today I measured the tree in my garden, it is 2.20 metres tall.
2. Trees grow.
3. In 1994, a scientist discovered a 3000 year old tree in the Brazilian jungle.

Having passed these statements on to you, which contain information and which knowledge?
If you were a computer capable of reading English, which RDF triples (or any other knowledge representation formalism) would be asserted in your knowledge/information base?

Comments

[Rogier Brussee]

# This is RDF (and some OWL) in the N3 format
# It assumes the existence of domain ontologies

# the statement we want to formalise is
# today I measured the tree in my garden. It is
# 2.20 metre tall.

:me msmnt:measured [ a msmnt:Measurement;
msmnt:ofwhat [ a msmnt:Height ; msmnt:ofwhat :mytree];
msmnt:value [rdf:value "2.20"; msmnt:unit si:metre];
msmnt:date :today].

:me a people:Person;
rdf:seealso < http://anjoanjewierden/bloggers.com >.

:mytree a plants:Tree;
spatial:locatedIn :mygarden.

:mygarden a housing:Garden.
:mygarden property:usedBy :me
:today a xsd:dateTime.
< http://anjo.blogs.com/metis/2004/03/knowledge_vs_in.html > dc:date :today.

#my tree is the only tree in my garden
# warning, not sure this works

:theTreesInMyGarden owl:intersectionOf(plants:Tree, [a owl:Restriction; owl:onProperty spatial:locatedIn; owl:toValue :mygarden]).

:theTreesInMyGarden oneOf (:mytree).

[Rogier Brussee]

Hai Anjo,

I agree that the difference between information and knowledge is largely overemphasized. This is particularly annoying when people start to suggest that knowledge is really a much nicer warmer and socially acceptable thing than that bland cold thing called information.

As a workable definition I think knowledge resides in the head of people, and information is something externally visible. Both knowledge and information and knowledge are, I believe, a symbolic representation of the real world (we should be inclusive in our interpretation of the real world. Social behaviour for example is certainly part of the real world). However implicitly or explicitly we as people keep track of that relation to the real world and this shapes our context, our expectation of how these "symbols" should behave (like the real thing that is). Unless we really do our best (like in the rdf example in the comment above) or we decided that the symbols are part of the real world (like in the case of money or computer programming) we rely on people to make this link when we exchange information.

corollary: people can be imaginative: they create a virtual reality by just using the symolic representation and the relations they have come to expect. In some sense this virtual reality can be more real than the real thing because it can leave out all sort of annoying complicating details. Thus the success of stories, TV, movies, literature etc.

corollary: if I exchange information with you, I want you to build up a symbolic representation of the reality I have made a symbolic representation of. I donot want you to build up my symbolic representation. It helps if we build roughly the same mental representations if we are on the samewavelength so to say, although it can also be very refreshing if you have a different point of view, because that pointof view may be revealing. But eventually what matters is if we talk about the same things (or people or behaviour or even enmotions). Needless to say however that getting my point accross is a lossy process.It is especially lossy if our implicitly assumed contexts turn out to be different.

corollary: The whole process is more or less recursive. Even though our representations are eventually based in reality, once internalised enough, representations of our abstract concepts become part of our foundations. We may even become stongly emotionally attached to them.

corollary: some mental representations may be extremely hard to serialise : skills, emotions, intuitions, deep complex knowledge, just lots of stupid little facts.

corollary: communication and information are very closely related.

corollary: it helps to make the context explicit.

Both knowledge and information need some substrate to be represented lets call that data. Thus for a computer data is just a blob of bits and bytes in RAM or harddisk, for a newspaper it is the stream of characters on the pages, and for a human it is ze little grey zells (or more precisely the connections between them). Actually is probably useful to not be too pecky and think of every bit of information with very little context as data. So for example text files are data if you run find on them but information if you read them. The effectiveness of data is directly proportional to the amount of effort it takes to represent information/knowledge. So what do I mean by reprensent here. Moddesty is certainly required here, but it should be something like maintaining the relationships that map on the relationships that exist in the real world. You are a programmer, so you know that this is non trivial and the effort strongly depends on the datamodel and the infrastructure you have chosen.

Corollary: not all information has to be represented by the same data.