Communications of the ACM

PTT—to red kite chicks immediately prior to fledging, using a backpack harness designed for minimal hindrance. The tags were solar-powered and programmed to record up to six location fixes per day. Although this maximum could indeed be achieved during the summer months, a lack of sunlight in Scotland meant fewer fixes (a maximum of four per day) were obtained in spring and autumn and only the occasional fix during winter. To further preserve battery power, data was transmitted from the tag to the satellite only once per week. We thus configured Blogging Birds to produce a blog every week, or each time data was received from a bird.

Figure 1 outlines the overall architecture of the Blogging Birds system. We next describe the main components; see also Ponnamperuma et al. 12

Data augmentation. The system processes an email messages with GPS fixes from the tags fixed to the red kites and enriches that data from readily available online sources about the local weather ( https://www.metoffice.gov. uk/datapoint), habitat (such as different types of grassland and forests, https://eip.ceh.ac.uk/lcm), and geographic features (such as rivers, lochs, roads, and location names, https:// www.ordnancesurvey.co.uk/). Table 1 presents a sample of the enriched data used by Blogging Birds.

Data analysis. The system then ap-
plies data-analysis procedures for
identifying home ranges and patterns
of movement with respect to these
temporary settlement areas. Home
ranges are identified as polygons using
the Adehabitat package for R2 by clus-
tering the previous locations of an indi-
vidual using 90% kernels. As described
by van der Wal et al., 24 we modeled lo-
cal movement patterns as angular and
radial velocity vectors to identify excur-
sions, characterized by travel in rela-
tively straight lines at higher speeds.
This data analysis allows the document
planner (described next) to detect the
three prototypical patterns of move-
ment in Figure 2, whereby the kite re-
mains within a home range, explores
an area outside its home ranges, or
moves from one home range to anoth-
er. Figure 3 shows the calculated home
ranges for a bird (gray polygons), as
well as the fixes classified as excursions
home ranges, or moves from one home
range to another. An ecological-do-
main model further defines different
travel, foraging, and social behaviors
as rules that can apply under specific
environmental and geographic condi-
tions; for instance, following heavy
rain, a kite observed on any of the
grassland habitats might feed on earth-
worms or a kite observed near a wood-
land habitat late in the afternoon is
likely to be preparing to roost. These
rules are implemented as JBoss Drools
( http://www.jboss.org/drools), a busi-
ness-logic-integration platform that al-
lows us to instantiate messages when
(black crosses) and non-excursions
(amber crosses).
Document planner. The document
planner in Figure 3 identifies pat-
terns in the data that signal different
red kite behaviors and creates “mes-
sages” (implemented as Java classes)
that encode these behaviors for use
by the “micro planner” and “sentence
realiser,” which then generate sen-
tences in English.
The data analysis allows us to detect
the three prototypical patterns of
movement outlined in Figure 2, where-
by the kite remains within a home
range, explores an area outside its
Figure 2. Prototypical red kite movement patterns: C1 is small and constricted movements
within an area of intense usage (home range); C2 is exploratory movement from a home
range (round trip); and C3 is direct movements between separate home ranges.

C1 C2 C3

Figure 3. Calculated home ranges (gray polygons) and classification of fixes as excursions (black crosses) or non-excursions (amber crosses) for a particular bird.