This is an overview of the ‘Music Data Hack #1’, which took place over the weekend of 10th-11th February 2017, and was organised by Craig Hamilton from the Birmingham Centre for Media and Cultural Research in association with CHIME.
The aim of these events is to bring together BCMCR and CHIME researchers, BCU students, practitioners from the surrounding area, and festival directors to work collaboratively on the development of online data visualisation tools, product prototypes, and experimental analytical methods related to music. The first of these events explored possible uses and applications for data collected about music festivals.
Using data collected from a small group of volunteers at the 2016 Cheltenham Jazz Festival through a pilot version of a mobile application currently being developed for CHIME, the hack aimed to explore the ways in which the data could be visualised in ways that may be useful to researchers, festival organisers and/or music fans. Working in small teams, participants were asked to conceive, plan and build a working prototype within 24 hours.
Friday 10th – 2pm-7pm
As had been our hope and intention, the hack attracted attendees from a variety of backgrounds, including staff and students from different BCU faculties, and also external practitioners from large and small organisations. In all 18 participants arrived for the Friday session, which began with informal networking and team building.
At 3pm the overall aims of the event were presented to the group. This included an overview of the CHIME project provided by Loes Rusch, and an introduction from one of our project partners, Annemiek van der Meijden of ZomerFietsJazzTour. The processes of data collection used in the Cheltenham Jazz Festival pilot were then explained to the group, along with information about the next phases of the project, and how this hack could help inform that development.
The data provided to the group comprised of c15,000 lines collected over the weekend of 29th April – 2nd May, 2016, and was made up of the following:
- Text/images collected via the CHIME prototype mobile interface at 2016 Cheltenham Jazz Festival (n = 145)
- Tweets with the official Cheltenham Jazz Festival hashtag, #cheltfest (n = 1703)
- Tweets with the UNESCO #jazzday hashtag (n = 12953)
Alongside the ‘core’ data sets, the following, additional data sets were provided:
- Basic programme information from Cheltenham Jazz Festival 2016 (Stage times, artists, venues, capacities)
- Results of automated computational Textual Analysis on core data (Sentiment Analysis, Hashtag searches, Character/word counts, etc)
Armed with this data, the teams were then set the following brief:
Working in teams, create an interactive online visualisation/interface that displays the data provided in ways that may help explore issues of cultural heritage among festival audiences, organisers and performers. We are particularly interested in how participants experience festivals through space (both the physical locations of festivals and/or through online spaces) and in terms of time (their experience during a particular festival and/or their memories of festivals over time). Ideally visualisations would have potential applications for data collected at future festivals and should be useful and/or meaningful to one or more of the following groups:
• Fans of Music (general), Jazz (particular), or Music Festivals
• Festival Practitioners
• Academic Researchers looking at Music Festival, Social Media or Audience Data
• General Audience
From 4pm onwards teams began to plan their ideas, organise their teams according to specific roles, and devise work plans for the next 24 hours.
By 6.30pm and the end of the first session teams had started to form ideas for their prototype. It was pleasing that the ideas ranged in technical scope and ambition, and that teams had negotiated internally ways of working that enabled participants with less experience of code and coding to contribute through other important activities, such as design, additional research, and planning.
Overall this session provided important groundwork for the following day, when the building of prototypes would begin in earnest.
Saturday 11th 10am-4pm
Our first session on Saturday focussed on an open discussion between CHIME Researchers, hack participants, and festival practitioners, lead by Nick Gebhardt. Annemeik van der Meijden from ZomerJazzFietsTour was joined by Ian Francis of the Birmingham-based Flatpack Festival, and William Soovik from GMLSTN Jazz in Sweden, who kindly joined us via Skype, to provide the perspective of festival organisers.
The discussion raised interesting points regarding the potential use cases for mobile apps at festivals that took in to consideration the activities of festivalgoers, the aims of the CHIME project, and the needs of festival organisers. The core map-based approach taken in versions 1 and 2 of the mobile app was generally seen as a positive feature, with ideas for possible applications ranging from the capture of information about and engagement with more ‘casual’ visitors to festivals (as opposed to more committed ‘regulars’ with whom festival organisers have already developed good channels of communication) that could prove useful, particularly at fringe/free events in public spaces. Navigation and familiarity with festival sites and artistic programmes were also considered as useful features, providing users with the feeling of ‘belonging’ to a wider festival experience, particularly where festivals are dispersed across a wide geographic areas, such as cities, and also extended timeframes. Hack participants posed useful questions and observations to both researchers and festival practitioners, and took on board responses in terms of the day ahead and the development of their prototypes.
The remaining sessions, either side of lunch, were a race against the clock for teams attempting to get their prototype visualisations and applications ready for the 3pm deadline, and included a ‘Paper Demo’ session as the final two hours approached. This provided teams with the opportunity to present their ideas in simple, graphic form. Feedback was intended to drive and focus the activities of teams in the remaining hours of the build.
At 3.15pm teams convened to hear 5 minute presentations of the completed hacks.
BCMCR Music Data Hack #1 – Completed Prototypes
The following ideas/prototypes were presented by the teams in the final session of the hack. Where possible, links are provided to the finished prototypes.
A Role/Sentiment Map of Cheltenham 2016
This visualisation was built using the Carto service and was based on the 145 data points collected via the mobile app at Cheltenham 2016, and subsequent ‘Sentiment Analysis’ of text. The map enables users to isolate data points in terms of sentiment scores and/or the roles of participants in the pilot. You can view this map here.
This particular idea was inspired by a test of the Carto software explored by another team during the Friday session, which displayed the same information using a Time and Location as the basis for its visualisation. Click here to see that visualisation, created by Paul Bradshaw.
Virtual Reality Data Exploration
Perhaps the most ambitions of the hack products, and taking also the Cheltenham data as it’s starting point, this idea sought to render the data collected from pilot participants into 3D virtual space. Users would be able to explore Cheltenham at street level across a condensed timeframe and encounter reflections (including text and images) at the very locations where they were originally posted. The team also described how further development could enable users to (re)experience an entire festival by ‘visiting’ stages and hearing recordings from artists pulled in via the APIs of music streaming services.
Festival Management Back-End
Based on designs and ideas for new features in the mobile application, this idea focussed primarily on the needs identified by festival practitioners during the morning session and sought to generate real-time analysis related to the experience and mood of festival visitors. The user interface would include several ‘gamification’ elements that the team hoped would make using the app more attractive, with activity linked to prizes such as free tickets or drinks discounts. Data collected via the app, along with real-time analytics, would be displayed to festival staff via a back-end dashboard, which the team demonstrated based on analysis and visualisations created in the R Package. The team felt this would enable more effective, proactive management of the festival space whilst at the same time forging a sense of community amongst festival visitors.
Exploring Festival Emotions
Based on components of automated text analysis this team built two visualisations using the Tableau and Carto services that displayed text according to colour-coded scales and separate emotional categories such as Anger, Joy, Sadness and Anticipation, derived from Sentiment Analysis. You can see both via the following links: Tableau Visualisation and Carto Visualisation
Recommendations & Discussion
On the whole we feel that the event was a worthwhile exercise and certainly one worth repeating at some stage. Participants informally canvassed at the end of the event indicated they would also consider returning for a further session. The main advantage of an event of this kind was the possibility of exploring different ideas and perspectives in a very short amount of time, with the technical aspect allowing some ideas to be developed beyond conceptual stage.
The next phase of this strand of the CHIME project will collect data at two further European Jazz Festivals, at GMLSTDN (Sweden) in April 2017 and during ZommerFeitsFest (Netherlands) later in the summer. This will produce two further datasets that can be analysed alongside the existing data from Cheltenham 2016, and could form the basis of a further hack event.
The sessions featuring Research Staff (on Friday) and Festival Practitioners (on Saturday) were useful in terms of providing context around the project, the data collected, and the aims of the hack. Additional sessions at future hacks could include those from practitioners involved in App/Visualisation/Data projects, either in terms of academic research and/or commercial projects. This would help provide context and guidance around the more ‘nuts and bolts’ elements of building data-derived prototypes.
Thanks to all participants for giving their time, efforts and expertise over the weekend. Thanks also to BCMCR and CHIME for their support in making this event possible.
A Storify of Tweets from the weekend can be found here
If you are interested in more information about the event please contact Craig Hamilton (firstname.lastname@example.org).