Enhancing Museum Narratives with the QRator Project: a Tasmanian devil, a Platypus and a Dead Man in a Box
Steven Gray, Claire Ross, Andrew Hudson-Smith, Melissa Terras, Claire Warwick; University College London, UK
Emergent mobile and web-based technologies offer museum professionals new ways of engaging visitors with their collections. Museums are powerful narrative learning environments and mobile technology can enable visitors to experience the narratives in museum objects and galleries and integrate them with their own personal narratives and interpretations. The QRator project explores how handheld mobile devices and new Internet-enabled interactive digital labels can create new models for public engagement, personal meaning-making and the construction of narrative opportunities inside museum spaces. This project is located within the emerging technical and cultural phenomenon known as “The Internet of Things.” The term refers to the cultural shift that is anticipated as society moves to a ubiquitous form of computing in which every device is ‘on’, and every device is connected in some way to the Internet. The project is based around the project “Tales of Things” (http://www.talesofthings.com) which has developed a “method for cataloguing physical objects online, which could make museums and galleries a more interactive experience” (Giles, 2010).
Keywords: Museum narratives, digital interactive labels, iPad, Crowd sourcing, public engagement
The following paper presents the technical development and data analysis of the QRator research project which aims to understand how digital technologies, such as interactive labels, tablets and smart phones, create new ways for users to engage with museum objects; investigate the value and constraints of digital sources and methods involving cultural content; and demonstrate how crowd-sourced digital interpretation may be utilised as a research source.
Museum exhibitions have been transformed by the addition of digital technology to enhance the visitor experience. Ubiquitous mobile technologies offer museum professionals new ways of personally engaging visitors with content, creating new relationships between museums and their users. Museums and other cultural institutions have made significant investments in developing and disseminating digital content in the physical museum space to reach and engage users, marking a shift in how museums communicate publicly their role as custodians of cultural content and their attitude towards cultural authority. Despite recent technical advances in collections access and interpretation, a number of key issues still remain. Does the rapidly changing technological environment provide a more engaging and participatory visitor experience?
The QRator project explores how mobile devices and interactive digital labels can create new models for public engagement, visitor meaning-making and the construction of multiple interpretations inside museum spaces. The Horizon Report (2011) indicates that Smart Objects are part of the future of digital museums. The QRator project highlights the ability of Smart Objects and is centrally located within the emerging technical and cultural phenomenon known as “The Internet of Things:” the technical and cultural shift that is anticipated as society moves to a ubiquitous form of computing in which every device is ‘on’ and connected in some way to the Internet. The project is based around technology developed at the Centre for Advanced Spatial Analysis, UCL and is an extension of the “Tales of Things” project (http://www.talesofthings.com) which has developed a “method for cataloguing physical objects online which could make museums and galleries a more interactive experience” (Giles, 2010) by means of QR codes, or Quick Response codes: a two dimensional matrix which encodes some data, in this case a uniform resource locator (URL) reference to an object. This project links a Quick Response code to a conversation about museum objects where museum curators can give insight into an object background, hence the name “QRator.” QRator provides the opportunity to move the discussion of objects from the museum label onto digital interfaces, allowing the creation of a sustainable, world leading model for two-way public interaction in museum spaces. The QRator project aims to stress the necessity of engaging visitors actively in the creation of their own interpretations of museum collections. This paper will focus on the integration of QR codes, iPhone, iPad, and android apps into UCL’s Grant Museum of Zoology, demonstrating the possibilities for visitor engagement. Although this paper will concentrate on mobile technology created for a Natural History Museum, issues of meaning-making and co-creation of content between the museum and its visitors through digital technology are applicable to any museum. This paper presents the technical development of the QRator project so far within The Grant Museum at UCL, highlights the design and development of the technical components, infrastructure, installation and user evaluation of the QRator application and stresses the opportunities and challenges in utilising digital technology to enhance visitor meaning making and narrative construction.
2. The QRator Application
The main component of QRator is a custom application that is built for Apple’s iOS platform running on ten iPads within the UCL Grant Museum. Each iPad asks a different question ranging from “Should we clone extinct animals” to “Is domestication ethical?”
The application is composed of four separate view states that automatically transition after a one-minute interval, although can be configured to transition after any length of time. The design of the application mirrors the current wooden museum labels that are displayed throughout the UCL Grant Museum. The question is framed in a virtual, interactive museum label (fig 1) that displays the question presented by the curators along with a short background of the issue. At various stages throughout the application, users are invited to interact with the device and contribute to the continuing conversation.
Figure 1: Interactive Museum Label displayed within QRator application.
The core interactive element of the application is a QR code (fig 2), which is prominently displayed in all views, which users can scan with a smart phone using the ‘Tales of Things’ application, available on both iOS and Android platforms, to record their response to the question asked by the device. The application allows users to engage with the curators on the iPad through the device’s virtual keyboard.
Figure 2: QR code used to allow users of smartphone app to contribute to discussion.
Each iPad is configured with a particular Twitter hashtag (e.g. #qrator), which allows the application to display a list of tweets that visitors inside museum can view and respond to from their own Twitter account via a smartphone. This social interaction allows users to carry on the discussion of the question at a later date.
QRator relies heavily on the “Tales of Things” or ToTeM, infrastructure to record users’ answers to the questions poised by Museum staff (fig 3). “Tales of Things” (http://www.talesofthings.com), a collaboration between Brunel University, Edinburgh University, University of Dundee, University of Salford and UCL funded by Research Councils UK Digital Economy Program, allows users to attach their memories and stories to any object and share them with other users (Barthel et al., 2011). The system uses read-write QR codes, which can be printed out and physically attached to the object, to link the objects to their owners’ stories. These codes can be scanned by any QR code application on smartphones to allow new owners of the tagged objects to see the history of the object through the eyes of the previous owners.
Figure 3: QRator System diagram with links to “Tales of Things” infrastructure.
The “Tales of Things” database is central to QRator as each question is stored as an object. In the same way that users tag their memories to physical objects, museum visitors tag their comments and interactions to the question in our application. The museum staff enter their question into the “Tales of Things” website, just as any user would an object. The QR code produced by the system contains an URL that links an object, via a unique 11 digit alphanumeric code to the entry in the database. It is this unique identifier that allows QRator to contact Tales of Things to retrieve the relevant data about the question via a private, internal API (Application Programming Interface).
The QRator Website (http://www.qrator.org) provides a portal for the public to engage with the system from outside the museum. The website, a customised version of the WordPress blogging platform (http://www.wordpress.org), displays all current and archived questions that have been asked by the curators. Users of the site can interact as if they were inside the museum and see a live updating feed of a question’s conversation, which is syncronised with the interactive labels within the museum. Each user contribution is inherently anonymous and is defined from being from one of 3 sources: ‘Museum iPad,’ ‘via QRator.org’ or ‘TheGrantMuseum’ curator. Users can clearly see where the comment has originated from or whether it is an authoritative post from a curator in the Museum.
When a web user adds their views to a question, they are presented with a ‘captcha’ from the Google service reCaptcha (http://www.google.com/recaptcha), an automatic challenge presented to prove a user is human, which reduces the levels of spam messages that could be generated by automated bots. The reCaptcha service displays two challenge words that have been extracted from books by OCR software. One of challenge words is known to the system and one unknown so by comparing the input from the user the system can calculate wither the input has been made by a human and therefore preventing automated systems from posting unwanted messages.
The QRator website aggregates the tweets for each question asked, mirroring the iPad application, which allows users to follow the conversation on both social media outlets and visitor contributions. The custom WordPress plugin takes advantage of a public API to pull the data from the “Tales of Things” database and display them to users. The API uses an Open Authentication (OAuth) token to verify that a website user is authorised to send a message. The site acts as a separate client to the ToTeM infrastructure and acts as a test case for a future, fully public API for Tales of Things.
Moderating User Generated Content
From the outset the museum required that all user-generated content submitted via the website and digital labels be moderated. The choices that were made at requirement capture were either a system that queued messages until approved by a member of museum staff or a system that was post moderated daily. It quickly became apparent, through user testing and discussion that that users expected their message to be displayed instantly over all mediums after input. It was decided that messages would be uploaded immediately with the caveat that a swear filter was applied to the message before being sent to the server.
QRator contains a pre-defined list of 48 swear words that is bundled with the application. Simultaneously, a list is maintained on the server, which is loaded at runtime. This remote list can be updated at any point if Museum staff find words within a message that are unacceptable. The system will revert back to the standard list if the list on the server cannot be retrieved.
This proved tricky to develop an effective solution, as discovered in early versions of the software. Designing an effective swear filter is not a trivial problem. Many false positive hits can be generated, for example “Classic” will be converted to “Cl***ic”, which is an incorrect replacement. We fixed this by including spaces before and after each word to mitigate this problem, but users quickly found ways around this. Ultimately museum staff moderate content within the system regularly by deleting inappropriate content when detected. An effective plan to deal with this limitation will be addressed in future versions of the application.
Installation of the iPads in the Museum
One of the major hurdles in this project was mounting the iPads in such a way that they were useable by the general public without distracting from the exhibits in the museum. The UCL Grant Museum is contained within a listed building so designing a mount that would not damage any part of building was a challenge.
Apple markets the iPad as a device which a single user experiences content on the device at any one time. The user may share the device but there are few applications where multiple people use an iPad at the same time. The iPad is designed to be rotated from portrait to landscape and interactively changes to suit it orientation. The on-screen keyboard is designed to be used either one handed or sitting on a users lap. Mounting the iPad flush with the wall, at 90 degrees from the floor, would create an unnatural typing posture when users type a message on the device (fig 4).
Figure 4: iPad mounted on wall creates uncomfortable typing position.
The iPads had to be secure so that visitors did not walk out of the museum with expensive hardware. In Apple retail stores, the iPad is secured to the table by a locking plate, which is attached, by a steel cable, to a desk. In a museum this would not be suitable as the iPads remain out of sight and so may be damaged by users holding and dropping the device. Care had to be taken to make sure that users could not exit the QRator application by pressing the home button and using the device to access the device browser giving the user an open gateway to the University network.
A mount was designed to address all these issues. A custom, lockable desk mount pitched at a 75-degree angle was built and bolted to the surfaces around the museum (fig 5). The power button on the top of the device is still accessible to museum staff but the device is secured by two hinges padlocked together to prevent removal. The iPad is also secured with a 4 digit code which is activated after the power button is pressed so that users are prevented from further attempting to tamper with the device.
Figure 5: QRator iPad mounted in UCL Grant Museum
The mount completely encloses the iPad around all sides, physically preventing access to home button. Surrounding the device with a metal frame to prevent the home button being pressed had some unexpected issues.
The iPad relies on a Wi-Fi connection to connect to the network and the antenna is situated on the topside, portrait orientation, of the device. Enclosing the iPad in a metal frame created a Faraday cage preventing the radio waves of the Wi-Fi signal getting to one of the iPads mounted in a weak signal area. The museum has only one access point mounted high on the ceiling in one corner blocked, in line of sight, by stone structural pillars and wooden cases. Fortunately, this iPad was mounted on a moveable case, which can be easily resituated.
For the UCL Petrie Museum a Perspex case was trailed but access to the device was limited due to the 2 screws that fix the case around the iPad. Although, the Perspex case solved the problem with Wi-Fi reception, access to the device took significantly more time to unlock and reconfigure than the Grant Museum mounts. Aesthetically, the matt black metal stand blends into the Grant Museum’s surroundings than the Perspex mount would and, to date, the network problems have only been seen on one device in one location. Moving the access point to another location or adding additional Wi-Fi hardware to the Museum space would provide full coverage for the museum floor and ultimately solve this issue.
Maintenance and Distribution of the Application
The QRator project undertook an agile approach to development of the iPad software. Since initial launch there have been over 15 updates to the application, mainly requests for new features from Museum staff. Updating the software of 10 iPads is a non-trivial problem since each device is locked in separate enclosures and the Museum is a fair distance from the development office. It would be unrealistic to remove all ten devices and carry them to the development office for updating every time a new version was available. The QRator development team proposes the following process as a solution to this problem.
iPad applications can be distributed one of two ways. An application can be distributed via the Apple App Store, a dedicated market place of applications that any user in the world can download once the application process has been approved. Internally, we felt that since users had no understanding of the short codes needed to set up the device with the relevant question and that we wanted to constrain the experience to only the Grant Museum, that this solution would not work. Fortunately, applications can be distributed internally via Enterprise distribution account to provide an in-house solution.
When the application is ready for deployment, it is archived and deployed for Enterprise distribution. The resulting binary is then uploaded to an internal server, not visible to the outside world, along with a provisioning profile, a file that dictates which devices can run the application, and a basic webpage to allow easy installation. The Museum curators are notified of the update when the application is ready to be installed and staff will reboot the devices to gain access to the iPads web browser, Safari. The curators navigate to the software update page on the internal network and click a single web link that instructs the device to download the new updated software. This process allows us to update the software and deploy a new version in under 20 minutes from compilation to running the new version within the museum.
3. Usage within UCL Museums and Collections
UCL holds a range of collections cover a wide variety of disciplines, reflecting the breadth of the university's academic work. Three collections; the Petrie Museum of Egyptian Archaeology, the Grant Museum of Zoology and the Art Museum are open to the public. Other collections are primarily for teaching and research but can be seen and studied by appointment. The QRator Project is currently installed in UCL's Grant Museum of Zoology, which houses one of the country's oldest and most important natural history collections. It is the only remaining university zoological museum in London and houses around 67,000 specimens. It has a strong history as a teaching collection but also functions as a key gateway for the public to engage with academic issues in innovative ways.
Data from the ten QRator iPads was collected by archiving contributions from March to November 2011. Collection of the data has been made simple within the QRator system. Each thread of the conversation for each question is stored as a tale within ‘ToTeM’. These comments are then aggregated together based on the current questions originally asked by the museum. A custom module was build for WordPress to collect the data from the public API and display the output as a CSV (comma separated values) file, which can be imported to statistical analysis packages for further analysis. This resulted in a corpus of 2784 visitor contributions, totaling 29,842 words and 4,496 unique tokens, providing a rich dataset for the analysis of visitor experience (fig 6).
Figure 6: Word cloud visualisation showing top 200 words of all visitors’ contributions
Visitor contributions were categorized qualitatively using open coded content analysis where each comment was read and categorized. Contributions were divided into three basic categories; about the current question or topic, about the museum, or noise. Despite the apparently simplistic categorisation, it is possible to discover patterns of use and begin to understand how visitors are relating to and interpreting the exhibitions, and making meaning from their experience.
For the purpose of this study, various quantitative measures were used such as analysing the frequency of comments according to date and time, comparing comment rate between the ten iPad’s and suitable text analysis tools were used to interrogate the corpus. In addition, the corpus was analysed using a Sentiment analysis tool, SentiStrength (http://sentistrength.wlv.ac.uk) developed by Thelwall et al, (forthcoming), in order to automatically measure emotion in the visitor comments, which provides an indication of a positive or negative museum experience.
4. Analysis of Data
The majority of the comments in the corpus fell into two main categories (fig 7): about the museum (42%) and the category of on topic (41%); triggered predominately by the QRator interface and questions posed by the museum curators, suggesting that visitors are inspired to share their own experiences, thus co-constructing a public multiple interpretation of museum objects. This is a mark of the project’s success since this was exactly what the museum professionals had hoped might happen. Interestingly, many of the visitor comments focused on opinions of the museum as a whole (42%). This raises the question of whether a digital technology used in this way promotes of an opportunity for visitors to make meaning from their whole experience, rather than engage with the exhibit specific content and interpret the exhibitions themselves.
Fig 7: Percentage of Visitor Contribution by Category. The majority of the comments in the corpus fell into Comments about the museum and comments on topic.
The lack of spam and inappropriate commenting is surprising (19%). Many museums have been hesitant to open up communication to greater participation by visitors. There is an ingrained fear in the museum profession that visitors will leave inappropriate comments when there is no moderation or intervention by the museum (Russo et al., 2008) despite research showing that museum visitors want to engage with complex, controversial topics by making comments or talking to staff and other visitors (Kelly, 2006). The QRator project and the Grant Museum have, however, adopted the concept of “radical trust” in the visitor community:
Radical trust is about trusting the community. We know that abuse can happen, but we trust (radically) that the community and participation will work. In the real world, we know that vandalism happens but we still put art and sculpture up in our parks. As an online community we come up with safeguards or mechanisms that help keep open contribution and participation working. (Fichter, 2006)
Inherent in the term is the suggestion of a previous lack of trust shown by museums towards visitors, but also the admission that such trust is regarded as new and perhaps dangerous. Nevertheless, the QRator data suggests that ‘radical trust’ in visitors does indeed work: spamming and inappropriate commenting does not appear to have happened to a significant extent in the Grant Museum. When focusing on the individual tablet interfaces and current questions it is possible to see that some current question prompts produce higher levels of on topic comments then others (fig 8). The Real or Fake current question received the most contributions by visitors which focused on the topic raised by the museum (170 comments); followed by Pet vs. Wildlife (154 comments) and Humans vs. Animals (146 comments). This is likely to be because the current questions posed were more direct, easier to directly associate with visitors’ previous experience and own perspectives, provoking a higher frequency of posts. In comparison with Bulldogs and Brown Hares which asks, “What makes and animal British” received a lower number of on topic posts (87 comments) but a high number of comments about the Museum (136 comments). This may be due to the question prompting visitors to think more scientifically about their response.
Fig 8: Category breakdowns from each of the Ten QRator iPads.
Text analysis tools were also used to interrogate the corpus of visitor contributions. It was assumed that frequent terms from QRator would reflect the topics and themes being discussed in the physical museum space. The QRator data was run through a commonly used text analysis tool Voyeur (http://voyeurtools.org), to highlight the commonly used words in the visitor contribution, and to enable a Sentiment Analysis to take place. The most frequent words in the corpus seem to highlight positive visitor contributions as well as the key topics discussed the Natural History specimens, the museum and the action QRator is encouraging visitors to undertake: animals (288), like (218), museum (186), think (159), love (148). The length of comment may also be used as an indicator of engagement, if we assume that those who are interested in an issue or topic may wish to write at greater length. Indeed the average length of comment increased significantly between categories. The noise category had an average of 4.071429 words, comments on the museum had 7.431599 words and visitor contributions on topic had an average of 15.36672 words. This is pleasing, since it suggests that visitors were inspired by the questions to engage with topics in a relatively complex fashion. Additionally, when compared to the SentiStrength results, which classifies for positive and negative sentiment on a scale of 1 (no sentiment) to 5 (very strong positive/negative sentiment), the comments on the museum were in average more positive in sentiment (2.04 positive), whereas the comments on topic had an equal positive to negative response (1.52 positive; 1.55 negative). This result suggests more engaged texts often contain a mix of positive and negative sentiment, in contrast to less engagement which is more likely to produce a single sentiment result.
5. Future Work
This paper has aimed to introduce the concept of digital technology in museums as a way to co-create visitor meaning-making and the construction of multiple interpretations inside museum spaces; however, museum digital technology cannot be used in isolation. These tools are important, but of equal importance is their relationship to other forms of museum interpretation, and of course, the visitors themselves. It is vital to incorporate the views and previous experience of visitors when undertaking collaborative content and digital technology development in museums. This research reinforces how complex museum experience is, as well as the difficulties of design for narrative construction in a museum setting. This paper offers insights into why and in what ways digital media, specifically QR codes and digital collaboration interpretation technology, have the potential to enhance the personal meaning making of museum experiences. Nevertheless, it is not until a strong research base has been developed that we will begin to truly understand the use of digital technology as tools for meaning making and creating a dialogue between visitors and museum staff, and to fully validate their value to museums and their visitors.
The Authors of this paper would like to acknowledge the creators of ‘Tales of Things’ platform, Martin de Jode and Ralph Barthell from UCL CASA for their continuing support of the project, groundwork leading up to the project and help with the public and private API that QRator relies heavily on. Thanks to UCL Museums and Collections, in particular Jask Ashby and Mark Carnall of UCL Grant Museum, Tonya Nelson (UCL Petrie Museum), Sussanah Chan (UCL Exhibition Coordinator) and Sally MacDonald, Head of UCL Museums and Collection. This work was funded through Beacons for Public Engagement Innovation Seed funding and we would like to thank Hilary Jackson (UCL Public Engagement) for her support and advice throughout the project.
Barthel, R., Leder Mackley, K., Hudson-Smith, A. et al (2011). “An internet of old things as an augmented memory system”. Personal and Ubiquitous Computing,. Accessed 8st January 2012. Available at http://www.springerlink.com/content/v8405w81p2j35451/
Fichter, D. (2006). “Web 2.0, library 2.0 and radical trust: A first take.” April 2. Accessed 27th October 2011. Available at http://library2.usask.ca/~fichter/blog_on_the_side/2006/04/web-2.html.
Giles, J. (2010). “Barcodes help objects tell their stories”. New Scientist, 17th April, 2010
Kelly, L. (2006). “Museums as sources of information and learning: The decision-making process”. Open Museum Journal 8. Accessed at http://archive.amol.org.au/omj/volume8/volume8_index.asp.
Russo, A., Watkins, J., Kelly, L., & Chan, S. (2008). “Participatory communication with social media”. Curator, 51 (1), 21–31.
Thelwall, M., Buckley, K., Paltoglou, G., Cai, D., & Kappas, A. (forthcoming). “Sentiment strength detection in short informal text”. Journal of the American Society for Information Science and Technology.