Databases Reference
In-Depth Information
or common ontology, the NisB system identifies and exploits bits and pieces of past experiences and
practices which will be re-composed and orchestrated to solve new problems.
One of the main design principles of NisB is tolerance towards incomplete, erroneous, or
evolving information, such as uncertain or erroneous micro-mappings and occasional changes in
business schemata. Uncertainty is modeled along the lines outlined in Chapter
3
, requiring no
reference ontology and allowing decentralized control. Reasoning techniques (see Section
3.3
),
algorithmic solutions for ensembles (Chapter
4
), and top-
K
matchings (Chapter
5
) enable a pay-
as-you-go approach to schema matching, where micro-mappings accumulate according to business
needs without high upfront investment from individual network members while supporting quick
setup for new members.
The research challenges of NisB cover four main areas. First, schema matchings are quantifiable
using uncertainty analysis, so that users can be informed about the usefulness of reusing a matching.
This approach promotes the sharing and reusing of fine-grain interoperability information. Second,
schema matchings should make their way through the network, giving it enough context information
to make them (re)usable. Third, the capabilities of domain experts should be analyzed to offer new
methods for utilizing indirect feedback from users. This will empower users and reduce the time,
effort, and expertise needed to establish interoperability. Finally, the use of top-
K
schema matchings
should be extended. Together, resolving these challenges will lead to techniques and protocols for
collaborative establishment of shared knowledge and understanding among communities of business
users.
6.2
DISASTER DATA MANAGEMENT
Disaster management is part of the discipline of emergency management, which is aimed at both
dealing with and avoiding emergency situations. By disaster management, we refer to the activities
that take place after a disaster has occurred, including rescue and recovery. The term encompasses
natural disasters (e.g., the 2005 hurricane Katrina), disease outbreaks (e.g., the 2003 SARS outbreak),
major accidents (e.g., the 2010 Chilean mine collapse), and terrorism.
While disasters by definition affect the lives and the welfare of many individuals, they differ in
their geographical impact and time scale. All disasters require immediate response, but in some cases,
there is more time to establish an appropriate IT infrastructure than in others. Given this fact, a
position paper by
Naumann and Raschid
[
2006
] identified the shortcomings of current IT solutions
for disaster data management, highlighting the importance of reliable information integration and
information sharing among the various bodies that must cooperate in disaster response: government
agencies, NGOs, individuals, communities, and autonomous industry organizations. The discussion
in this section follows closely the discussion in
Naumann and Raschid
[
2006
].
Disaster data management involves many different types of information, including data on
victims and relief personnel; reports of damage to buildings, infrastructure and goods; weather
forecasts; geographical data on roads and other landmarks; logistics of vehicles and delivery times;
communications; details of aid and donations; and blog data. In disaster management, the difficulty of
