In English

Detecting usage patterns - A study on developmental benefits achieved through detecting usage patterna in applications by using a logging component

Jacob Larsson ; Thorvaldur Gautsson
Göteborg : Chalmers tekniska högskola, 2015. 96 s.
[Examensarbete på avancerad nivå]

Background: The inherent complexity of software development has in recent decades necessitated the advancement of engineering methods in order to facilitate the construction of software applications in an efficient manner. For this reason, many different methodologies and processes have been developed. Problems still remain however, for example in the field of usage pattern analysis. In many cases there is a gap between the assumed use of an application and its actual use. Understanding e.g. which features are used, when they are used, and whether they can cause crashes can enable developers to better focus their efforts towards productive development and thus contribute towards further engineering knowledge which can then be applied in software development. Aim: To explore how logs can be used to understand usage patterns in order to improve a software application. This is achieved by building a logging component which can with a minimum effort be integrated into an existing application. Method: The study presents and analyzes a logging component developed using the design science research methodology. The process consisted of five steps, in which the problem domain was analyzed, requirements were identified, a prototype was designed, built and evaluated, and finally conclusions drawn. Evaluation took place through quantitative measurements as well as a survey which was posed to employees at four different companies. After being completed, the logging component was then additionally assessed through a case study in which the logging component was integrated with two different applications and semi-structured interviews held with participants at two workshops. Results: The research found that a logging component that visually presents usage patterns is useful for developers in order to to aid further development of an application. Participants of the workshops as well as those who were surveyed considered the most useful information to be data on exceptions and crashes, but data on feature usage was also considered important. All developers interviewed and over 90% of the people surveyed knew one or more projects which would have benefited from an external logging component with the functionality presented in this research. Conclusion: The study confirms that a visual presentation of logs by an external logging component can aid developers in further developing an application through presenting information about exceptions and crashes as well as features used by users of the application. An external logging component of the type developed in this research can therefore lead to faster and more efficient development. Recommended future work includes extending the study to cover applications written in more programming languages, as well as testing the logging component through an experimental study.

Nyckelord: Logging, usage patterns, features, code injections, data analysis



Publikationen registrerades 2015-07-20. Den ändrades senast 2015-07-20

CPL ID: 219818

Detta är en tjänst från Chalmers bibliotek