Taking Artbot Forward
By Josh Cowls on October 6, 2015
It’s great to be getting underway here at MIT, as a new graduate student in CMS and an RA in HyperStudio. One of my initial tasks for my Hyperstudio research has been to get to grips with the exciting Artbot project, developed by recent alumni Desi Gonzalez, Liam Andrew, and other HyperStudio members, and think about how we might take it forward.
The genesis of Artbot was the realisation that, though the Boston area is awash with a remarkable array of cultural offerings, residents lacked a comprehensive, responsive tool bringing together all of these experiences in an engaging way. This is the gap that Artbot sought to fill. A recent conference paper introducing the project outlined the three primary aims of Artbot:
- To encourage a meaningful and sustained relationship to art
- To do so by getting users physically in front of works of art
- To reveal the rich connections among holdings and activities at cultural institutions in Boston
With these aims in mind, the team built a highly sophisticated platform to serve up local art experiences in two ways: through a recommendation system responsive to a user’s expressed interests, and through a discovery system drawing on meaningful associations between different offerings. Both these processes were designed to be automated, building on a network of scrapers and parsers which allow the app to automatically categorize, classify, and create connections between different entities. The whole project was built using open-source software, and can be accessed via artbotapp.com in mobile web browsers.
I’ve spent some time getting first-hand experience with Artbot as a user, and several things stick out. First, and most importantly: it works! The app is instantly immersive, drawing the user in through its related and recommended events feeds. Experiencing art is typically a subjective and self-directed process, and the app succeeds in mimicking this by nudging rather than pushing the user through an exploration of the local art scene.
Second, it is interesting to note how the app handles the complexity of cultural events and our varied interest in them. On one level, events are by definition fixed to a given time and place (even when they span a wide timespan or multiple venues.) Yet on another level, a complex package of social, cultural and practical cues usually governs the decision over whether or not we want to actually attend any particular event. This is where the app’s relation and recommendation systems really become useful, drawing meaningful links between events to highlight those that users are more likely to be genuinely interested in but may not have searched for or otherwise come across.
Finally, the successful implementation of the app for Boston’s art scene led us to think about the different directions we might take it going forward. In principle, although the app currently only scrapes museum and gallery websites for event data, the underlying architecture for categorization and classification is culturally agnostic, suggesting the possibility for a wider range of local events to be included.
The value of such a tool could be immense. It’s exciting to imagine a single platform offering information about every music concert, sporting event and civic meeting in a given locality, enabling residents to make informed choices about how they might engage with their community. But this is crucially dependent on a second new component: allowing users to enter information themselves, thus providing another stream of information about local events. As such, we’re proposing both a diversification of the cultural coverage of the app, but also a democratisation of the means by which events can be discovered and promoted. We’ve also given it a new, more widely applicable name: Knowtice.
This move towards diversification and democratisation chimes with the broader principles of the platform. ‘Parserbot’ – the core component of Artbot which performs natural language processing and entity extraction of relevant data – is open source, and therefore could in future allow communities other than our launch locality Boston to adopt and implement it independently, shaping it to their own needs. At root, all events require some basic information: a time and date, a location, a purpose, and people to attend. This data is standardisable, making it possible to collect together information about a wide array of events in a similar format. Yet despite these structural similarities, in substantive terms no two events are ever the same, which is why we are committed to providing a platform which facilitates distinctiveness, letting communities to express themselves through their events.
We recently entered the Knight Foundation’s News Challenge with a view to taking the app forward in these new directions. You can view our submitted application (and up-vote it!) here. As we state in our proposal, we think that there’s tremendous potential for a tool that helps to unlock the cultural and social value of local activities in a way that engages and enthuses the whole community. We plan to build on the firm foundations of Artbot to create a social, sustainable, open-source platform to accomplish this broad and bold ambition. Keep checking this blog to find out how we get on!
