Semantic Wikis and Disaster Relief Operations
Access to timely information is critical for relief operations in emergency situations. Over the last years social-networking web systems, such as wikis, have become more and more sophisticated and can also be applied fruitfully in humanitarian information management. However, a major drawback of the Web currently is that its content is not machine-readable, a shortcoming that is addressed by the Semantic Web approach.
In this article, I'll first propose using wikis to share information faster and more easily during emergencies, and secondly, I'll introduce a way to enhance them semantically. It is particularly promising to create a Semantic Web extension for wikis, i.e., to provide them with an underlying model of the knowledge described in their entries.
There are a number of websites where information on humanitarian emergencies and disasters can be found, e.g., ReliefWeb and AlertNet. Moreover, there are sites dedicated to specific disasters, e.g., the Humanitarian Information Centres sites. A wiki related to the South Asia earthquake from October 2005 can be found here.
However, most of those sites rarely use the social-networking concept of fast and massive user participation. This becomes apparent when listing the information products: situation reports, press releases, contact lists, databases of assessments, who-does-what-where, etc. Certainly, those products are produced by or based on the input of the concerned community. But with wikis, information can be provided much quicker and more directly, which is critical in humanitarian disasters -- particularly in the early stages.
Situation reports are a common format for reporting in emergencies. They are usually produced daily at the onset of a disaster, later less frequently. Typically, a reporting officer, who collects the information from various colleagues (and other sources), writes the reports. By contrast, wikis enable the persons providing the information to publish it directly, thus skipping the reporting officer. They can publish immediately and, with the adequate device, from anywhere and up to several times per day.
To classify content, the cluster and the location concerned are essential pieces of information. Cluster is the term used to refer to the nine areas of response -- including shelter, water, food, and health -- during an emergency. For specifying the location, the P-code system, which is similar to a common zip code system, is recommended and is usually introduced by Humanitarian Information Centres in disaster areas; see here for Pakistan.
An essential and widely acknowledged tool during emergency operations is the Sphere handbook  issued by the Sphere project. It declares minimum standards for the level of disaster assistance to which all people have a right, regardless of political, ethnic, or geographical specificity. For example, for the cluster "shelter and settlement," the third standard for covered living space states:
People have sufficient covered space to provide dignified accommodation. Essential household activities can be satisfactorily undertaken, and livelihood support activities can be pursued as required.
One indicator that this standard is fulfilled is: "The initial covered floor area per person is at least 3.5m2."
The Semantic Web approach addresses the shortcoming that the machine itself does not understand the content of the Web, nor of wikis. For example, it is not easy to get the answer to the following question from a wiki about the South Asia earthquake, although it might exist:
Which relief organizations can provide how many tents for the earthquake-affected region Bala Kot in Pakistan?
The information required to find this answer cannot be readily retrieved because the information is hidden in the text of wiki articles; automatic retrieval of the required data is just impossible. So far, the only way to deal with such an issue is to provide the required data sets manually, as is done in Wikipedia with articles that basically consist of a listing only. However, this method is prone to errors and needs much maintenance.
In , an approach to enhance wikis semantically is introduced by adding "typed links" and "attributes" as new features. While links within a wiki — and also anywhere in the Web — do not normally state the way in which linked entities are related, typed links between articles are classified according to their meaning. The other addition is attribute-value pairs, a fundamental data representation in many computing systems. These pairs declare a relation between an article and a data value.
It is essential, especially in the field of disaster management, that these enhancements don't require much additional or even technical work, as time is crucial. Yet, the approach in  is based on an easy-to-learn syntax and leaves the user freedom to pick arbitrary names according to the folksonomy idea. The following scenario illustrates that by applying this method, the problem above can be solved with simple queries.
Pages: 1, 2