IEEE Consumer Electronics Magazine - March/April 2020 - 58

Mobile Applications

IN-APP USER'S ACTIVITY
DETECTION OVERVIEW
& WITH THE raise of the use in mobile devices
and apps,1 the study of the in-app user activity is
necessary since it provides essential information
to identify trends and understand behaviors.
Nevertheless, the central challenge in this scenario has always been focused on the analysis of
the obtained data,2 leaving aside the way they
are collected. This is usually considered as a
programming issue since it depends on how to
integrate third party libraries, also known as
trackers, with the app.
In the current scenario, the obtained process
is clear: apps owners analyze which are the characteristics points to be detected inside the app
and then the developers include marks in the
code to track users' actions. The inclusion of
these marks involves code changes that lead to
new releases of the application that should be
updated in the client side to take effect. Moreover, this procedure has to be repeated every
time a new point has to be tracked, increasing
the total number of app updates.
This process implies high costs in terms of
time and money, making dynamic track of the
user behavior inside mobile apps only affordable
by those stakeholders who can face this extra
effort. It can also imply a loss of data if the user
skips any update, so the total track in real time
becomes an almost impossible aim.
In this article, we propose the use of aspectoriented programming (AOP) as a new approach
for improving the user mobile application path
detection by easing the integration of the measured points. This will imply a costs reduction in
terms of time to market and efforts and an
increase of the points detection control and
management.

RELATED WORK
Mobile App Analytics: Initial Considerations
Mobile apps analytics usually refer to data
collected while the app is being used. Through
different analysis, these data provide lots of
insights into user behavior together with information about overall app performance. In this
regard, from generic usage statistics to particular information, such as the queries users enter

58

into the e-shop search bar, can be provided if
the measured points are correctly set.
Several trackers are typically used when talking about mobile analytics, which can be classified
among three main groups:3 first, the advertising
trackers, such as Google AdWords and Inmobi,
which represent almost the 65% of the trackers
used. They are mainly serve within in-app advertisements and they are used to collect personal
data from the users. Second, the analytics trackers
(24% of use), such as Google Analytics (GA) and
AppsFlyer (AF), which mainly track users' actions
inside and across apps for client attribution and
other marketing purposes. Finally, the utility trackers (11% of use), as Bugsense and Crashlytics,
focused on tracking bugs.
According to this, our article is focused on
the integration of analytics trackers which
obtain different inputs for improving the application overall performance and increase the user's
experience by providing essential feedback to
different schema recommendation.4
There are several maintenance activities that
must be done to maintain the high quality and stable performance, and most of them are needed to
assure continuous improvement and new features
to be recognized by current and potential users.
Nevertheless, and due to the high associated cost
of these modifications (representing about 40%-
70% of the total cost of software life cycle,6 having
the corrective maintenance a 20% of this effort),
an efficient management and planning of them is
needed.
Considering software maintenance as "the modification of a software product after delivery to correct faults, to improve performance or other
attributes, or to adapt the product to a modified
environment",7 current solutions for user path
analysis within mobile applications can be seen as
one of these tasks. In fact, these activities can be
classified as corrective tasks since they comprise
code changes for detecting different user's actions
but they do not really correct any malfunction of
the application. For this reason, finding another
way to include the associated measure points that
does not imply a continuous and iterative code
modification process could be an effective way to
avoid overloading the maintenance scenario, allowing an efficient effort allocation for the appropriate
application adjustment.

IEEE Consumer Electronics Magazine
IEEE Consumer Electronics Magazine - March/April 2020

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - March/April 2020
