This weekend I finally got the picture: you have the quantities user and knowledge (as being relevant data) and you have the variables trust and relevance. And they all interact with each other. My guess is that all my previous post fit in these categories. User would be: “Personal portal, how to keep track of the second me” and “I am online“, knowledge would be: “Open source knowledge“, trust is also mentioned in “I am online” and relevance comes along in “Plaatsen vs Matchen” and “Chaos Control“.

Today I would like to dig a little deeper into “knowledge”. Eric Miller’s Thesis “Towards personalized metadata and knowledge systems” (2003) gave me some useful clues.

His paper is mainly about the use of metadata (which could be guessed from the title) But the focus of this blogpost will mainly be on the route from data to knowledge (if metadata is your thing, and you understand Dutch, please read this post “Folksonomies en de mythe van het semantische web“) Knowledge is the information that is useful to a user. And information is data that has meaning to the user. Using Brown and Duguid’s The Social life of Information Eric creates a scheme that made the differences clear to me.

Data
What we can access
Any accessible node of content on the network

Information
What we can understand
Nodes of content that fulfill certain requirements for usability. May qualify by language, culture, presentation, accessibility, timeliness, quality, reputation, distribution, technical compatibility, or other attributes that facilitate comprehension.

Knowledge
What we find useful
Nodes of content that are specifically relevant and informative for a specific user in their specific context. Differentiated from information because it adds value to a user’s personal ontology.

Given these scheme I find it also easier to understand why relevance and trust are important factors of what make data knowledge. Eric puts it in one line: A given node(st: data-object) may be accessible (data) and comprehensible (information) but what makes it valuable is the fact that it contributes to an individual’s pattern of understanding (knowledge).

So knowledge isn’t something that is definable as being there, it is only knowledge if it makes relevance to you. Should this conclude that the title of my previous post could better be “Open Source Information” One argument against it could be that we are looking from the personal point of knowledge, if you would look at it from a top view than the information in weblogs could be called knowledge too. The internet although not being a semantic web does learn from its users. (I don’t completely understand what I’m saying here)

In my vision most of the web2.0 applications also fit in the 4 categories. (most fit in more categories, but I like simplicity even when its a lie)

User
Flickr/svirsk (personal photos)
svirsk/blog (personal thoughts)

Data:
Flickr (photos)
blogger (weblogs)

Relevance (as being relevant at this moment)
del.icio.us (bookmarks)
Flickr/tagcloud (tags of today)
digg (interesting articles)
Del.ico.us/populair (what’s hot today)

Relevance (as being relevant to me)
del.icio.us/svirsk
Flickr/svirsk/tags
blogline weblogs

Trust
Ebay reputation (for sellers)
google pagerank (for pages and websites)
Microsoft passport (for people)

As you can see especially trust can use some extra research from my side, Raphaël Mazoyer mentioned in his paper that trust has something to do with transparency and Eric points that trust systems based on metadata are very hard to design and refers to the url of Prof. Suhl from the Lehrstuhl für Wirtschaftsinformatik.

Concluding my post for today I think there is a need for an application that measures the trustworthy of publishers and their expertness in a certain field. Which allows the user in understanding the relevance of information.