Semantic Annotation for Knowledge Management

Do you remember your annotation home-works from literature courses? Researching the qualifications of an author, figuring out the topic, tone, rhetorical strategy, audience, and purpose of an essay, or thinking about the connections between what you just read and other work in the field ... Imagine everyone being able to put such facts and relations into a machine understandable form and having machines harvest those relations on our behalf. That is what semantic wikis enable.

Semantic annotation defines the domain concepts and relations between concepts. Formally, an annotation is a tuple consisting of annotation (subject, object, relation, and context) as defined in "Annotation and Navigation in Semantic Wikis" by Eyal Oren et al. Our Semantic Wiki for Complex Operations uses Semantic MediaWiki, that allows annotations to a wiki page. For instance the insurgency page has the following annotation:

* [[has characteristic::Popular Support]]

Here the subject of the annotation is the 'insurgency' concept represented by this wiki page, 'has characteristic' is the annotation relation, and 'popular support' is the object of the annotation. While Semantic MediaWiki only allows single level annotations of wiki pages and does not formally separate the page and the concept it represents, we think it still serves as the widely adopted standardized semantics syntax necessary for semantic wiki applications to take off.

Annotation clearly introduces an additional burden on the knowledge worker. So unless the return on investment on semantic annotation provides value to the community that the wiki serves, it would be hard to expect widespread adoption. Semantic MediaWiki extensions provide such value. Again referring back to our Semantic Wiki for Complex Operations, the wiki page for each social science data set (e.g. Minorities at Risk Project Dataset, CIRI Human Rights Data Project, etc.) is annotated by using the built-in 'category' attribute:

Category: Dataset

That is, each social science dataset in our Complex Operations wiki is annotated to be of category dataset. Clicking on the link Dataset above gives a table that lists currently avaiable social science datasets in our wiki:


In a traditional wiki, this table needs to be manually specified by:

wiki syntax.tiff

In contrast, semantic annotation enables us to generate this table dynamically using only one statement in Semantic MediaWiki:

{{ #ask: [[Category:Dataset]]

| ?title

| ?year }}

As this example illustrates, semantic annotation provides a significant ROI to the knowledge worker in knowledge organization. Moreover, when a researcher adds another dataset to our wiki, this table will automatically include the new dataset, thus improving knowledge maintenance.