User-centered Study into Photo Sharing

By ​Ben Francis

Executive Summary

This report outlines a study into photo sharing.

The study started with a literature review of related work and then went on to define core "concepts" in information sharing in general through knowledge elicitation techniques.

Multiple task models were created based on the use of existing photo sharing systems and then combined into a single general task model. New tasks were defined which extended the task model to address issues which were identified in the knowledge elicitation process but are not supported by existing systems. The extended tasks addressed user's needs through the use of emerging trends and technologies identified in the literature review.

Related Work

Among the first to study the sharing of physical photographs was Richard Chalfen (1987). Chalfen discovered that physical photographs are typically shared in a home setting where family and friends share photographs of events such as birthdays and family holidays. This activity which Chalfen referred to as the "home mode" of the "Kodak Culture" would often involve telling stories about the photos being shared to add a narrative to the images which could in turn be built on through group discussion.

A recent study by Miller & Edwards (2007) into the users of the "Flickr" photo sharing web site discovered that the invention of digital cameras (including camera phones) and the Internet has given rise to new photography practices, socialisation styles and perspectives on privacy. The study revealed two distinct groups of users which were labelled as "Kodak Culture" and "Snaprs".

The Kodak Culture group demonstrated how traditional practices have been augmented by digital technology so that photo sharing still occurs within an existing social group, but the sharing occurs via the Internet. This group was very concerned with the privacy of their photographs and tended to use more established technology such as email to share their photos in a private and controlled way. The subject of new emerging privacy concerns in online and mobile photo sharing was also studied by Ahern et al. (2007).

The Snaprs group were a group of hobbyists who demonstrated completely new practices such as group "photo-strolling", public sharing of photos and the concept of folksonomy. This group was much less concerned with privacy and used photo sharing as a creative outlet, building new social bonds through messaging, tagging and commenting on photos.

The study also identified a distinction between photo sharing and mobile blogging or "moblogging". Photo sharing tends to more closely share the goals of traditional photography where as moblogging tends to tell stories with photographs rather than about photographs.

An ethnographic study by Crabtree, Mariani & Rodden (2004) explored ways in which people collaborate around and share collections of physical photographs in order to derive requirements for future digital photo sharing systems which support distributed collaboration. They identified that digital photo sharing is currently restricted to "passing on" photos by email, web pages or multi-media messaging on mobile phones and that there is great potential for systems which move users away from a single computer screen and use image-based communication practices. The study outlined several challenges in the design of such systems including creating collaborative interfaces which support tangible interaction techniques and gestures, coordination protocols and awareness mechanisms.

Various work has been done in trying to make the sharing of photos easier from camera phones. The Mobile Media Metadata 2 System (Ahem, S., King, S. & Davis, M. 2005) provides an automatic upload capability which utilises wireless networking and uses spatial, social and temporal metadata to suggest likely recipients of an image.

Wilhelm et al. (2004) describe a system which allows users of cameraphones to annotate their photographs at the time of capture, provides assistance by making guesses about the context of the photos and leverages networked collaborative metadata resources. Problems with network latency and reliability are identified, though thought to be less of an issue as networks improve. Other problems identified are text entry and navigation on mobile devices and the need to better integrate desktop and mobile application components.

The inclusion of GPS receivers in mobile phones has meant that new location based applications are possible. Naaman, Paepcke & Garcia-Molina (2003) describe a system called LOCALE in which the latitude and longitude of the location in which a photograph was taken is used to automatically assign a label based on other photos which were taken in the same area.

Snavely, Seitz & Szeliski (2006) created a system which allows 3D exploration of photo collections using a novel user interface. The system is able to take an un-ordered set of photos from personal collections or photo sharing web sites and automatically compute the viewpoint of each photograph. This allows the creation of 3D virtual photo tours of tourist areas and can even generate a non-photorealistic 3D model of objects based on the images. Whilst the system can use metadata from camera phones such as location, orientation and focal length to initialise algorithms, it does not rely on this data for geo-referencing an image. Instead, the system uses sophisticated computer vision techniques to automatically calculate location, orientation and geometry to determine the relative position of photographs. In order to view photos on a map, these relative positions can then be turned into geographical positions using a known reference point (e.g. from satellite imagery).

The proliferation of digital cameras and particularly cameraphones is also having an effect on news reporting. Some of the most striking and timely images of recent news events such as the 7th July London bombings, Mumbai terrorist attack and plane crash in the Hudson river in New York have come not from news correspondants but from members of the public armed with cameraphones. These examples of so called "Citizen Journalism" are becoming more frequent and the media is looking to leverage this rich new source of information (Thurman, 2008).

Critical Decision Method

The Critical Decision Method (Shadbolt, 2005) was used as a semi-structured interview and protocol analysis technique. Four participants were asked a set of probe questions to elicit specific information about cues, choices, action plans and experience in decision making during an incident from the participant's experience. Three of the participants were male engineering students and one was a female psychology student.

In table 1.2 below, the responses of the engineering and psychology students to the probe questions are compared and contrasted.

Probe	Engineering Student Response	Psychology Student Response
Cues	A socially significant event ocurred.	Friend was moving away from the area, wanted to give them a present which would remind them of good memories.
Knowledge	Put in a situation where knowledge meant could help family, friend or stranger.	Knew that friend was leaving.
Analogues	Previous experience of helping or sharing information with people.	Created a similar present before.
Scenarios	Helping people with less knowledge or equipment. Various useful training.	Creating a present which involves photos. Using basic IT skills from school.
Goals	Share information, satisfy friends or family or elicit feedback to solve a problem.	Remind friend of memories.
Options	Get someone else to carry out the task or use a less hi-tech solution.	Used a less hi-tech solution.
Choice	Easiest, cheapest, quickest, or longest lasting option.	Less risk in making mistakes.
Anticipation	Correctly anticipated emotional response from peers.	Correctly anticipated emotional response from relative.
Experience	IT skills, technical experience, using mobile phone, experience of relationships or social situations.	Basic IT skills used.
Decision Making	Time pressure in most cases, quick decisions made.	Some time pressure, but planned in advance.
Aiding	More technical knowledge could have helped in some cases.	Knowledge of more sophisticated software would have helped.
Situation Assessment	Sharing photos or information with friend or stranger to solve a problem or record significant events.	Made a photo collage for my cousin to share memories.
Errors	Made technical mistake, wasted money, mis-understood request or not shared enough information.	Could have used original physical photos and ruined them.
Hypotheticals	Might have used different technology, requested more information or not bothered at all.	Would have shared different photos for a different event.

Table 1.2 Critical Decision Method Participant Responses

Firstly, because there were three engineering student participants and only one psychology student participant the engineering responses tend to be more general as they are summarising a range of scenarios. However, some useful comparisons can be drawn.

The engineering participants tended to be motivated to share information in order to solve a problem or to help a friend or family member, where as the psychology participant had more sentimental reasons. Both sets of participants were acting within a well known scenario with sufficient knowledge to support their decisions.

Engineering students chose a solution because it would save time or money but the psychology student felt it reduced the risk involved if errors were made, both agreed that a more low-tech option was available but chose not to use it. Both groups carried out tasks under some level of time pressure, though the psychology participant demonstrated more forward planning. Both groups correctly anticipated a particular emotional response from their peers and identified errors that could have been made.

This analysis is based on a limited sample and future work could carry out interviews with more participants which would be likely to generate different results.

Repertory Grid

A second interview technique, repertory grid, was used to carry out factor analysis in order to derive a set of key characterising factors which describe people's perceptions of different methods of information sharing.

The method was used on two participants, one male engineering student and one female psychology student. The repertory grid method provides a way of describing people's perceptions without pre-judging terms of reference because it is the inteviewee who defines the constructs on which a topic is rated.

First, the participant was asked to make a list of different ways of sharing information. Groups of three items were then picked from this list at random and the participant was asked to state a way in which two of the items were the same and one was different - thereby arriving at a list of differentiating factors or "constructs".

A grid was then constructed to allow each of the different methods of information sharing to be rated based on each construct. Each method was rated as a 1 or a 0 to either be characterised by a construct or its contrast respectively. If a construct could not be applied in a sensible way to a particular method, that construct was discarded.

Through this rating procedure and a series of sums and template construction, the method finally outputs a short list of key factors which represent how the participant perceives different methods of information sharing.

The key factors derived from the interview with the engineering participant can be described as below. Each factor has been assigned a label which approximately describes it.

1. Phone - frequent, social, one recipient, one sender, low max bandwidth, few contributors, trusted
2. Blog - Text, one way, electronic, public
3. Instant Messaging - synchronous, discreet, instant

This list of factors shows that the engineering participant mainly distinguishes between information sharing methods by technical characteristics. The first factor represents one-to-one trusted communication which is used for frequent social interaction. The second factor represents publishing of text-based information to the public and the third describes instant messaging services. It is interesting to note that all appear to be electronic methods of communication.

The psychology participant arrived at a different set of factors which are listed below:

1. Electronic - Free, external storage, sent via Internet, login, address needed, stored online
2. Physical - Physical action, one recipient
3. Private - Control over privacy

This participant distinguishes between information sharing methods based on whether they are electronic or physical and views privacy as a significant factor, re-enforcing previous work (Ahern et al.) which indicates increasing concerns over privacy. As might be expected, this participant's perceptions are less based on technology and more based on human factors.

These two participants could be linked back to the two groups of photo sharers identified by Miller & Edwards. The psychology participant could be said to relate to the "Kodak Culture" group which views privacy as a major concern and are slow to uptake new technology, compared with the more tech-savvy "Snapr" group who are less concerned with privacy.

Affinity Diagram

An affinity diagram (Beyer & Holtzblatt, 1998) was used to collate all of the knowledge elicited from the critical decision and repertory grid methods and attempt to reveal common issues and themes. The diagram is built from the bottom up, allowing common structure and themes to emerge from individiual notes captured during interviews.

Fig. 1.1 Affinity Diagram

Some significant and interesting themes emerge from the affinity diagram. Firstly, it reveals that as human beings we tend to share information more in socially significant times. That might be at a special social occasion or when something remarkable or unusual happens. Humans are also very good at predicting the emotional responses of others and judging what is and isn't a significant event.

Another theme that emerged is that we like to store memories and give stored memories as gifts or use them to remind ourselves of events at a later date.

It appears that we like to use technology to help other people and consider it useful to develop our technical skills in order to be better at doing so. We consider low-tech solutions to be clumsy or a last resort, but they have the advantage of greater privacy than high-tech solutions. High-tech solutions on the other hand are fast, inexpensive, less prone to errors and tend to result in more public sharing of information with less control over privacy.

Key Concepts

A review of literature of related work suggests that technological developments present new opportunities for applications which:

• Allow sharing of photos of significant events or places
• Take advantage of the proliferation of digital cameras and camera phones
• Provide opportunities for new social activities for tech-savvy users
• Take into account the privacy concerns of less technical users
• Allow communication through commenting and tagging to build social bonds
• Tell stories with photos or about photos
• Support distributed collaboration using image-based communication
• Accommodate coordination protocols and awareness mechanisms
• Utilise spatial, social and temporal metadata with images
• Attempt to make intelligent guesses based on metadata
• Utilise wireless networking
• Support text entry and navigation on a mobile device
• Allow the 3D exploration of photo collections
• Support citizen journalism

Analysis of the knowledge elicited from interviews leads us to believe that there are certain issues that seem to be of a high significance to the potential customers interviewed. A successful user-centered system might:

• Leverage human ability to judge social situations and the potential responses of others
• Facilitate the storing of memories
• Facilitate helping other people through technology
• Take advantage of the benefits of high-tech solutions
• Take into account the current advantages of low-tech solutions such as privacy

Task Analysis

A task analysis was carried out on existing photo sharing systems using a hierarchical task analysis (Newman & Lamming, 1995) technique. Two participants, one male engineering student and one female psychology student, were interviewed about an incident when they have shared photos.

As they described their goals and how they acted upon them, goals and sub-goals were written on post-it notes on a whiteboard. Once the activity interview was complete, the post-its were arranged into a hierarchy of goals and sub-goals and added to by asking "why" the user carried out a particular action (to identify parent nodes) and "how" they carried out their goals (to identify children nodes) to arrive at a more complete hierarchy graph.

Alternative task sequences were identified from the task hierarchy and drawn onto a whiteboard, showing different possible paths through the hierarchy and conditions which apply.

The next stage was to combine the two individual task models into a single generalised task model and the resulting hierarchy graph is shown in Fig. 1.2.

Fig. 1.2 Task Hierarchy

The task model supports the findings of the knowledge elicitation techniques used earlier which indicated that humans like to store memories and share those memories using a high-tech solution and they are good at judging social situations. It does not reveal any privacy concerns which were suggested by these techniques. The task model also supports previous work which indicates the new opportunities being created by the proliferation of digital cameras, communication & commenting on the Internet, telling stories with and about photos, distributed collaboration and the use of metadata in the form of tagging.

The task model also reveals additional social goals like re-affirming social identity by soliciting dialogue in relation to photos and correcting incorrect tags on photos to maintain social structure. It is clear from the hierarchy and plans that there are currently a huge number of sub-tasks required to fulfill the goal of sharing photos of an interesting event. Many of these tasks do not significantly contribute to the overriding goal which suggests this process could be streamlined.

The task model shows that whilst some of the primary issues of concern for the potential customers interviewed are fulfilled by the current systems evaluated, there are gaps. For example, using technology to help other people is not facilitated. The technologies evaluated also did not take advantage of many of the identified opportunities created by new technology. These include taking advantage of the wireless networking capabilities of cameraphones, providing opportunities for new social activities, coordination protocols, using spatial and temporal metadata of photos, making intelligent guesses based on metadata, supporting the use of mobile devices, allowing 3D exploration of photo collections and supporting citizen journalism.

References

Ahem, S., King, S. & Davis, M. (2005) MM2: Mobile Media Metadata for Photo Sharing, Proceedings of the 14th annual ACM interational conference on Multimedia, Hilton, Singapore.

Ahern, S., Eckles, D., Good, N., King, S., Naaman, M. & Nair, Rahul (2007) Over-exposed?: privacy patterns and considerations in online and mobile photo sharing, Proceedings of the 2007 SIGCHI Conference on Human Factors in Computing Systems, San Jose, California, USA.

Beyer, H. & Holtzblatt, K. (1998) Contextual Design: Defining Customer-Centered Systems, Morgan Kaufmann Publishers, San Francisco, CA, USA.

Chalfen, R. (1987) Snapshot Versions of Life, Bowling Green State University Popular Press

Crabtree, A., Rodden, T. & Mariani, J. (2004) Collaborating Around Collections: Informing the Continued Development of Photoware, Proceedings of the 2004 ACM confrerence on Computer supported cooperative work, pp. 396-405, Chicago, Illinois, USA.

Miller, A. D. & Edwards, W. Keith (2007) Give and take: a study of consumer photo-sharing culture and practice, Proceedings of the SIGCHI conference on Human factors in computing systems 2007, pp. 347-356, San Jose, California, USA.

Naaman, M., Paepcke, A. & Garcia-Molina, H. (2003) From Where to What: Metadata Sharing for Digital Photographs with Geographic Coordinates, On The Move to Meaningful Internet Systems 2003, vol. 2999/2003 pp. 196-217

Newman, W. M. & Lamming, M. G. (1995) Interactive System Design, Addison-Wesley Publishers, Harlow, England.

Shadbolt, N. (2005) Eliciting Expertise, Evaluation of Human Work, 3rd Edition, pp. 193-194, CRC Press, Boca Raton, FL

Snavely, N., Seitz, S. & Szeliski, R. (2006) Photo Tourism: Exploring Photo Collections in 3D, Proceedings of the 2005 International Conference on Computer Graphics and Interactive Techniques, pp. 835-846, New York, USA.

Thurman, Neil J. (2008) Forums for Citizen Journalists? Adoption of User Generated Content Initiatives by Online News Media, New Media & Society, Vol. 10, No. 1, pp. 139-157

Wilhelm, A., Takhteyev, Y, Sarvas, R. House, N. & Davis, M. (2004) Photo Annotation on a Camera Phone, Proceedings of the 2004 Conference on Human Factors in Computing Systems, pp. 1403-1406, Vienna, Austria