Semantic Wikis and Disaster Relief Operations
by Soenke Ziesche | Pages: 1, 2

4.0 Scenario for a Disaster

The information management of a complex humanitarian disaster can be handled by taking advantage of these new services. Immediately after the disaster occurs, a wiki would be created (ideally facilitated by the UN), powered by Semantic MediaWiki, an extension of MediaWiki providing semantic technology and thus turning it into a semantic wiki. Let's look at the South Asia earthquake in Pakistan on 8 October 2005 as an example.

One requirement is that the minimum standards of the Sphere handbook, introduced earlier in this article, are already available in this wiki while it is otherwise blank.

A field of application for the typed links would be the administrative structure of locations, since exact localizations are crucial in disasters. For example, in Pakistan there are the provinces on the top level, which consist of districts, while districts themselves consist of tehsils, etc. If the wiki article for the Pakistani district Mansehra contains a linked phrase like "One of the tehsils most affected by the earthquake is Bala Kot," the machine doesn't understand what the relation between "Mansehra" and "Bala Kot" is. However, if the link gets the annotation "is part of" or "is tehsil of," the relationship could be determined and would look as follows, using the syntax of [2]:

Source of a clipping of a wiki article about Mansehra (including a typed link):

One of the most affected tehsils by the earthquake is [[has tehsil::Bala Kot]].

Attribute-value pairs can be adopted to formalize one of the essential questions at the onset of a disaster, "Who, i.e., which organization, is doing what, where?" For example, the entry for the UN refugee agency UNHCR would provide the information that the agency currently works in the district of Mansehra and has 3,000 16m2 tents available.

Source of a clipping of a wiki article about UNHCR (including a typed link and an attribute-value pair):

UNHCR arrived on the ground after the earthquake in [[district of organization::Mansehra]] and has [[number of 16m2 tents:=3,000]] 16m2 tents available.

This assumes that a needs assessment for the tehsil of Bala Kot reveals that 20,000 people still need shelter. Moreover, the indicator from the Sphere handbook is formalized as follows:

Source of a clipping of a wiki article about shelter and settlement (including an attribute-value pair):

The initial covered floor area per person is at least [[required space per person:=3.5m2]].

While conventional wikis offer only a full-text search of their content as well as a categorization of articles, a semantic wiki would provide a query opportunity based on an RDF query language such as SPARQL. However, for relief workers under extreme time pressure, a convenient interface is necessary, such as that provided by [2] in their implementation. Queries regarding typed links can be reduced to three tuples (subject article, typed link, object article), where one or two fields are left empty while attribute-value pairs are two tuples. The results are given by two- or three-column tables, respectively.

Now, given that there are further articles containing this information, as well as typed links and attribute-value pairs, you can now find, with only a few queries, that there are 5,300 tents provided by the organizations UNHCR and CHF available in Bala Kot's district.

UNHCR	3000
IOM	2500
CHF	2300
Medair	800

Figure 1. Query: (Attribute: number of 16m2 tents, Value: ?)

UNHCR	district of organization	Mansehra
IOM	district of organization	Muzaffarabad
CHF	district of organization	Mansehra
Medair	district of organization	Batagram

Figure 2. Query: (Subject article: ?, Typed link: district of organization, Object article: ?)

Mansehra

has tehsil

Bala Kot

Figure 3. Query: (Subject article: ?, Typed link: has tehsil, Object article: Bala Kot)

5.0 Conclusion

I have proposed here that an extensive use of semantic wikis would provide enormous benefits for the exchange of information in humanitarian emergencies, as they provide, without redundancy, a combination of a repository of articles and a database that can be queried.

---