My name is Gian-Luca Savino and I am a human computer interaction (HCI) researcher from the University of Bremen, Germany. I studied computer science and recently attended my first computational social science (CSS) conference in Chicago. Both fields, CSS and HCI, are very interdisciplinary research fields and share a lot of practices and methods. In this blog post, I wanted to highlight a method, typically used in HCI, namely collecting unsupervised usage data from mobile applications.
Widely used apps like Facebook, Twitter or Google Maps count millions of users and are already deeply entrenched in our daily social life. However, while we know that mobile map applications are used quite often, we know very little about how they are used. For instance, what is the most prominent action users take within these applications? Are there reoccurring interaction patterns? How are they making use of the search functionality? Do different users groups show different kinds of behaviour? Usage data from commercial providers is not publicly available, and this situation is likely to persist owing to the extensive competitive and privacy concerns associated with releasing this type of information. Such data that is especially valuable to HCI researchers like me, but can also be used in a variety of fields such as CSS. The problem however, is that this data is locked away behind corporate walls of secrecy.
To do so, we take inspiration from Carrascal and Church who used an in-situ approach to investigate mobile search application behaviour from 18 participants by creating an application which acted as a “wrapper” for a popular search engine on Android. We developed such a “wrapper” application for Google Maps called MapRecorder which “wraps” the Google Maps mobile website. This allows researchers to capture rich behavioural logs while affording a very similar experience to the standard application. Such wrapper apps, can be build for Android and iOS devices and not limited to a particular OS. Already in 2011 we published a large-scale deployment-based research study that logged detailed application usage information from over 4,100 users of Android-powered mobile devices. Back then we used an application deeply integrated into the OS and not able to log in-app-usage.
We used our application called MapRecorder in two studies already to collect over 500 minutes of Google Maps usage data. This data has been analyzed to understand when and where people are using which functions of the application to get a better picture about how people use mobile maps on a daily basis. Thus far we collected usage data from 34 participants but when it comes to sample size really the sky (and your server capacity) is the limit. If you can get your application out to thousands of people you will be able to build large data sets with high scientific value.
The method of using a wrapper application bares some difficulties though when it comes to users privacy. Websites that are displayed within a wrapper application can be quite deceiving as they suggest that one is browsing a normal website. It is very easy to forget about the fact that all the data one is producing within such an application is tracked and stored. That is why it is very important to have a transparent way of dealing with user consent.
Especially in the emerging field of CSS we have to learn to harvest the huge amounts of data that are being produced around us on a daily basis. “Wrapper” applications as presented above are one way that can be easily scaled and adopted to a range of web-based applications and once properly implemented for one application it can be scaled to as many users as you can acquire for your study. Whether you might be interested in mobile map application usage or not this article should encourage everyone to find ways to collect the data they are interested in themselves. As more and more computational methods find their way into social science they should be used to their fullest potential.
Gian-Luca is a PhD Student at the HCI lab at the University of Bremen, Germany. His research focuses on the interaction between humans and mobile navigation devices. With a lot of past research on the interaction with Google Maps in everyday situations, he is now approaching the domain of navigation for cyclists, investigating novel navigation modalities.
The research interests of the workgroup lie at the intersection between HCI, geographic information science and ubiquitous interface technologies. In the HCI lab they investigate how people interact with digital spatial information and create new methods and novel interfaces to help people interact with spatial information