Page 349 -

12.3 Activity-logging software 339

A final limitation of this approach involves the challenge of extracting data.
Converting these computational artifacts from their native form to a representation
suitable for analysis can be challenging. You may need to write special-purpose soft-
ware to extract data from these tools. In some cases, this may require interpreting (or
reverse engineering) nonstandard file formats.
Although log files and implicitly stored data may prove useful for the analysis
of many important tasks and activities, these approaches have some very real limita-
tions. These tools are often limited in the granularity of the data that are collected.
A web log that provides detailed information about the paths followed by various
users in the course of completing some tasks does not contain any information about
the users' activities while they were on a given site. Similarly, an email client may
provide information regarding the structure of nested mailboxes, but information
about intermediate states—such as the names of mailboxes that were created and
later deleted—may not be captured.
Numerous studies have looked at email use from a variety of perspectives, in-
cluding understanding how users “refind” old emails (Whittaker et al., 2011), using
content to personalize search results (Teevan et al., 2005), and understanding how
batching and work practices influence productivity and stress associated with email
(Mark et al., 2016).

12.3 ACTIVITY-LOGGING SOFTWARE

Software tools for logging and recording user activity can provide rich data for
usability studies. Tools that capture mouse actions (movements and clicks), key-
strokes, and other interactions can help us identify common sequences, understand
actions used to complete tasks, and often to gather information about transitions
between different tools. Unlike the web server logs described earlier, these data col-
lection tools can be applied to many different applications, providing the possibility
of insight into the use of email, office productivity tools, and core operating system
features.
These tools generally fall into two main categories: proxies and interaction
recording tools. Proxies intercept user actions and record appropriate data points
before passing the actions on to the original software (Figure 12.6). Data returned
from the application can also be intercepted and modified before being returned to
the user. Both user interaction data and application response data can be stored in
a log file.
Interaction recording tools generally capture screen video and potentially micro-
phone audio, providing a record of what happened and when. The resulting video and
audio streams provide context and details not possible with simple proxies, allowing
us to know not just that the user was working with a word processor, for example,
but what she was typing and often why. Some usability tools use a combination of
recording tools and proxies to capture both raw events and video, providing a rich
mix that puts recorded actions in context.