Psychonomics 2016: Consistency in Group Categorisations
November 19, 2016
Psychonomics is a broad church, ranging from highly theoretical to highly applied work. This paper is very much on the applied end of the spectrum. It provides a thorough description of people’s behaviour in a real-world scenario, categorising personal photos, which can then be linked back to pertinent underlying fundamental problems. (Apparently, such links back to theory can be submitted to CRPI once completed.)
The photos were taken in somewhat artificial situations. Participants were asked to spend an hour on Edinburgh‘s Royal Mile to watch the Street Festival in August 2013 (Study 1) and to spend a day enjoying the Edinburgh Festival Fringe in August 2014 (Study 2). We then asked them many questions related to how they would manage their own photos. This paper summarises the findings from one of those questions – how they would categorise photos into groups.
This has immediate practical relevance: If people are consistent in the way they categorise their photos, this means that a machine learning algorithm can learn these categories. We also wanted to look at how consistent people are in categorising photos into events. This is a popular way of organising sets of photos which is provided automatically by Google, Facebook, and Apple.
We found that people are indeed internally consistent, keeping to a similar number of categories, with a clear match between initial sorts of photos repeated after a day, week, or month – even though the only instruction we gave to people was to sort them into meaningful categories.
However, this consistency disappeared when people were asked to sort their photos after a year – unless they had sorted their photos both on the day they took them and the very next day. Since we were only able to recall a small percentage (1/3) of our participants for logistical reasons, we need to follow up this finding with a longer-term study.
Our results were fed back into the prototype for Personal Information Preservation that was developed in the EU project ForgetIT. Our work also complements the literature on working with digital photos that exists within Human-Computer Interaction, as these studies often focus either on snapshots of how people use digital artefacts, including photos, at a given time, or on how people use a given photo organisation app over time.
If you would like a copy of the poster, please email me at maria dot wolters AT ed dot ac dot uk – or check my Edinburgh Research Explorer page, where this poster will be listed in December 2016.
Kirk, D; Sellen, A.; Rother, C.; & Wood, K. (2006). Understanding photowork. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI ’06), Rebecca Grinter, Thomas Rodden, Paul Aoki, Ed Cutrell, Robin Jeffries, and Gary Olson (Eds.). ACM, New York, NY, USA, 761-770. DOI=http://dx.doi.org/10.1145/1124772.1124885
Logie et al. (2016). Foundations of Forgetting and Remembering: Final Report. Deliverable 2.4, ForgetIT project
Maus et al. (2016). Personal Preservation Report. Deliverable 9.5, ForgetIT project.
Niederée, C., Kanhabua, N., Gallo, F., & Logie, R. H. (2015). Forgetful Digital Memory: Towards Brain-Inspired Long-Term Data and Information Management. S I G M O D Record, 44(2), 41-46. DOI: 10.1145/2814710.2814718
Tullis, T & Albert, B. (2013): Measuring the User Experience. Burlington, MA: Morgan Kaufmann. Second Edition