Biology Reference
In-Depth Information
As noted above, the multilingual ontology is made available for browsing
on the Web portal and also for download as a Web Ontology Language (OWL)
file. Users can lookup unfamiliar terms, identify causal agents, find syn-
onyms in different languages, and link to external resources such as Unified
Medical Language Service (UMLS), SNOMED CT, and ICD-10 (Lindberg et
al. 1993, Stearns et al. 2001, World Health Organization 2001). A recent inno-
vation has been the use of micro-blogging to disseminate BioCaster reports.
During the influenza A H1N1 outbreak in May 2009, human analysts sub-
mitted more than 400 micro-blogs to Twitter.
5
This helped to reach out to a
new community of users and also reduced the user load on our server dur-
ing the surge period.
15.4 Results
The BioCaster Web portal currently receives on average between 2000
and 3000 unique user accesses per month from 30 to 50 countries with the
majority coming from United States, South Korea, Japan, China, the United
Kingdom, Germany, and Thailand. Partnership networks are also crucial for
disseminating data directly to key user groups. For example, since the start of
the influenza A (H1N1) epidemic, BioCaster has collaborated with InSTEDD
(Innovative Support to Emergencies Diseases and Disasters) Evolve to make
geo-located reports available via RSS feed (Kass-Hout and Tada 2008). It fol-
lows that Evolve amalgamates BioCaster as one of its data sources, and adds
a new layer of collaborative filtering. The second version of the open access
multilingual ontology is widely accessed online and has been downloaded
by more than 73 academic, industrial, and public health groups worldwide,
including the WHO and other groups located in North America (25 groups),
Asia (32 groups), Europe (14 groups), and Oceania (1 group).
In 2008, an average of 68 positive reports were collected each day for
English-language news covering 150 countries and 88 infectious diseases in
humans and animals. Normalized frequencies are shown for the top 31 most
frequent countries and diseases in Figures 15.4 and 15.5. Although, there is
a clear bias toward English-speaking countries, the graph indicates the key
importance of the Asia Pacific region with significant numbers of reports
from China, Indonesia, Japan, Philippines, Thailand, and Vietnam.
Returning to Figure 15.1, several characteristics of how the recent influenza
A (H1N1) epidemic was reported and queried on the Web can be clearly seen.
First, it is clear from the BioCaster data that the number of news reports is not
related in a simple way to the real-world frequencies of cases or deaths from
the outbreak. Editorial control, country bias, user interest, media distribution
capacity, and so on all play a part in determining how news is reported. Second,
we see a small but significant lead in the reporting and querying of news about
