Information Technology Reference
In-Depth Information
associated with these events, we found that “wake up” had also occurred
between 18:00 and 21:00 and “lunch” had also occurred between 16:00
and 19:00. Analyzing the detailed information about these events, we were
able to find some reasons for such outliers. These suggested that waking
up late and eating lunch late were caused by jobs and studying.
We compared dataset #1 with dataset #2. We found that daily events
such as “wake up,” “lunch,” and “get home” were placed at almost the
same positions in both sets of data. However, the “wake up” event has a
notable feature. In dataset #1, this event occurred later on holidays than on
weekdays, but in dataset #2, the event occurred later on weekdays than on
holidays. Thus, the rhythm of people's lives might have changed. We also
found that some events occurred frequently only during one period. These
events are found in one dataset but not the other. We found the event
“cobb” at the left-centre of the view in dataset #1, and inferred from its
position that the event had occurred between 12:00 and 21:00. When we
clicked the circle, radial lines were seen to be concentrated from 12:00 to
15:00 and from 17:00 to 21:00. We may conclude that this event occurred
at lunchtime or dinnertime for many people. From the day view, we found
that the event had occurred on Thursday, Friday, and Saturday. By
studying the detailed view of the “cobb” event, we came to understand that
a hamburger chain had used Twitter to promote their new product.
Discussion
We were able to understand some rough features of events using
ChronoView. Examples of these features are how often events occurred
and whether an event was dependent on a specific time and when an event
had occurred. Even if two or more events have almost the same frequency,
we can find some differences between them, such as their dependence on a
specific time and periodicity. We were able to understand the occurrence
time and the distribution of times regarding an event, by adding radial
lines from the circle corresponding to the event. The radial lines overcome
some of ChronoView's ambiguity.
Exploiting three types of view, as shown in the case study, we can
observe differences in event occurrence times for each day of the week
and for weekdays and holidays. We can understand details of the events by
seeing the original tweet text. In this way, we can analyse the different
features of daily events to give an overview and breakdown of the details
from various views in the ChronoView interface. It is possible to
understand distinct features of daily events in different periods by
switching between the three view types and using the radial line
Search WWH ::
Custom Search