Biology Reference
In-Depth Information
of an extensive location resource for English as is the Yahoo! Geocode API
(Application Programming Interface). To the best of our knowledge though,
no open resource exists that encodes population frequencies by province;
this is the minimum level necessary for the determination of at-risk popula-
tions. The BioCaster group is now working toward expanding the system's
location ontology to include this information.
At the user level, it is important for BioCaster developers to more closely
engage the public health community in countries with both strong and weak
health systems to explain the benefits of news media sources. This two-way
dialogue is particularly important for BioCaster as, currently, the group
lacks the human resources to carry out a full-time manual check on system
output. In the future, the BioCaster group envisages that it will be crucial to
engage user understanding of the technology, and involve users in extend-
ing the reporting and checking of outbreak reports.
In the future, it should be possible to solicit even more timely news sources,
such as blogs and microblogs (Twitter feeds), from social networks, and also
monitor social semantic metadata such as Hashtags.
8
The time to publication
for these reports is extremely short; anyone with SMS capabilities can post a
microblog to Twitter almost instantaneously. However, one limitation with
this method is a high signal-to-noise ratio due to lack of editorial control
and the limited context each message provides. Other challenges include the
highly informal language used on blogs, as well as their often subjective
nature. Making practical use of these resources is a current research area
and will require much stronger filtering than we currently employ, includ-
ing perhaps novel techniques to model trust
in a report.
15.6 Conclusions
BioCaster is one of a range of systems for international disease monitoring.
The benefits, limitations, and successes of the Web as a near real-time means
for collecting and distributing data about public health events have been high-
lighted in this chapter. Based on advanced text mining technology, we have
been providing a freely available service to the global public health commu-
nity since 2006. The BioCaster group is continually working to improve the
system by (a) increasing accessibility and better integration with public health
users by conducting a usability survey, (b) developing qualitative metrics to
measure and improve detection rates, (c) extending coverage to new languages
and health threats, (d) collaborating with other system developers and user
groups to enhance data sharing and interchange standards, and (e) continu-
ously improving the quality of linguistic signals that indicate unusual events.
Access to BioCaster can be requested by public health organizations from
the first author.
