Understanding Serendipity to Inform Novel Processes, Methods and Technologies for the Researcher Mel Woods

Stephann Makri & Ann Blandford

Jamie Forth & Geraint A. Wiggins

Duncan of Jordanstone College of Art and Design University of Dundee

UCL Interaction Centre, University College London

Goldsmiths University of London

{s.makri,a.blandford}@ucl.ac.uk

{j.forth,g.wiggins}@gold.ac.uk

[email protected]

Robert Stewart, Diana Bental & Ruth Aylett Heriot Watt University.

{r.stewart,d.s.bental,r.s.aylett}@ hw.ac.uk

Xu Sun & Sarah Sharples

Scott Piao & Jon Whittle

School of Mechanical, Materials and Manufacturing Engineering University of Nottingham

School of Computing and Communications Lancaster University

{s.piao,j.n.whittle}@lancaster.ac.uk

{sarah.sharples. xu sun} @nottingham.ac.uk

1. INTRODUCTION Many scientific and artistic innovations have been attributed to serendipity, the faculty of making and recognising fortunate and unexpected discoveries by accident. However, while there is a widespread acknowledgement that serendipity is a major contributor to innovation, there are no cohesive theories or the design of technologies that may facilitate it, furthermore there is disagreement as to whether digital technologies promote or stifle serendipity. This poster reports on research to date from an interdisciplinary RCUK funded SerenA1 Project. It describes usercentered approaches to understanding serendipity and the design of systems, technologies and spaces to promote or support serendipity. The SerenA project focuses on: 1) building an understanding of serendipity through empirical studies, specifically within information discovery and research. 2) the development of a system to support and promote connections and information between people and ideas 3) implementing and evaluating technologies with novel approaches in digital and physical spaces.

3. The Serendipity Arena To achieve an initial understanding of the nature of serendipity, workshops were conducted with researchers. What is Serendipity? A Workshop held at SerenA Project Meeting 1, Dundee Contemporary Arts. Dundee. June, 2010. Aims: 3 groups of 4 people, each from different institutions and research interests. Method: the groups generated of examples and experiences considered to be serendipitous from either their work or personal life. A discussion of ‘Yes’ examples: ‘chance’ scientific discoveries, general library and Web-related examples such as coming across an interesting book when looking for another one. ‘Mediating Connections’2 A Workshop held at DE All Hands in Nottingham, October, 2010. Integrating arts, engineering and interaction design: Requirements Gathering and Novel Approaches. Aims: To explore potential user requirements with researchers. The workshop was in three parts 1) What is Serendipity?’ 2) Your Serendipitous Future 3) Designs on Serendipity. Examples created: The serendipity object creator; The ideas sharer; A graffiti pen; A behaviour monitor; The post1

EPSRC Project EP/H042741/1,“SerenA–Chance Encounters in the Space of Ideas”,http://www.serena.ac.uk

2

http://mediatingconnections.wordpress.com/

production lever; An online play-room; In-flight connectionmaker. A classification of the features and affordances of the examples have informed designs for technologies for scenarios for SerenA in a public context. The workshops were a useful first step for understanding the nature of serendipity and for providing the SerenA team with a reference tool for ‘looking back’ at our initial understanding.

3.1 Understanding Serendipity in Information Research The following studies apply a user-centered approach to understand and influence serendipity in an information research context. A Diary Study: Eleven participations over one week. Aims: to understand serendipity in information research using a mobile diary application developed on the Android mobile platform to allow participants to rapidly capture how serendipity happens and the context in which they experience serendipity. The resultant diary entries were discussed during post-study interviews. An Emergent Themes Analysis was conducted to understand our data. Serendipity Stories A study of semi-structured Critical Incident interviews with 28 researchers’ across different disciplines ranging from Architectural Design, to Computational Musicology, to Security and Crime Science. Aims: to gain a rich understanding of researchers’ perceptions of their broad experiences of serendipity. A Story Example: An Urban Design researcher was thinking about how to design a joint university faculty building for sustainability and architecture. The aim of the building design was to ‘try to bring together’ researchers from the two related disciplines. The interviewee wanted to design the movement within the building to provide many different paths between rooms so that, in her words, “you had to walk past a series of exhibitions and possible inspirations on your way to your office every single day.” Whilst sitting in the studio and after a long day of thinking about how to design the building, the interviewee was watching the US news-based comedy programme ‘The Daily Show.’ The interviewee frequently watched this particular television show - where the host makes fun of the news and talks to a guest towards the end of the show. In this particular show, the guest was a geologist who had found the wreckage of the RMS Titanic. When asked about his most significant (rather than best known) finding, the geologist described finding large ecosystems at the bottom of the Pacific Ocean where giant tubeworms had a symbiotic relationship with creatures that would capture sulphur that had escaped from the earth’s crust and live inside the

tubeworms. The interviewee found this particularly inspiring and translated the concept of the tubeworms’ symbiotic relationship with the creatures into a metaphor to guide her building design. She used this metaphor to design the building based on a symbiotic, sustainable concept where spaces within the building “live off each other” – where “if you had taken something and cut it off, you were cutting off some of the life source.” The results of these studies identified 1) key elements to support understanding of serendipity, 2) the influential role of context in serendipitous experiences, 3) a framework of understanding how serendipity happens and 4) the positive impacts of serendipity in people's information research. The framework for classifying serendipity should consider aspects associated with the activity, the value of the information, the source of the information and the interaction between the context and individual. In the context of our empirically based finding in research we define Serendipity as ‘a mix of chance and insight that leads to a valuable, unanticipated outcome’.

5 A Systems Overview The development of the SerenA system is focused by the empirical research described above with its classification, definition and model of serendipity. At the core, SerenA is a Webdistributed multi-agent system; it is closely aligned with current developments surrounding the World Wide Web, namely the principles and technologies underlying the Semantic Web and Linked Data initiatives. SerenA users each have a number of agents acting on their behalf. Some agents continually try to gather new information about their user, for example, by analysing publications, notes, emails, or social networking media, according to individual user preferences. Information gathered is analysed in order to produce rich models of users' research interests, which in turn directs the behaviour of other agents responsible for searching for new information accessible over the Internet, or from within internal semantic knowledge bases that is highly relevant but currently unknown to the user. As well as assisting users to discover new information, SerenA will facilitate serendipitous encounters between people, capitalising on the prevalence of mobile computing to transform unrealised colocation opportunities into valuable real-world interactions. Mapping Connections in Real Time. A Study at DE Summer School, Lancaster University July 20113. Aims: to explore the potential of a social network technology as a demonstrator of the types of information that might be displayed on a large screen, as a probe to form the basis of follow-up interviews with people in a conference context. Method: Three data visualisations were employed, Two driven by twitter API search and one by a prototype system TAUCONS (Twitter Analysis and User CONnection System). User Feedback and evaluation conducted with 13 people in groups of 3 discussed attitudes to data sharing and privacy, the use of technologies such as twitter to promote serendipity, opportunities for data visualization and attitudes towards the concept of technologyenabled serendipity. Interviews used the visualisations as a probe but were not focused solely on twitter technology. Conclusions: technologies, and the public display of real time data through social networking systems offer potential for encouraging serendipity however the usable interactions and value needs to be designed and managed appropriately if such a system is to have a novel and positive impact on research ideas, thoughts and outcomes.

3

http://www.docfest.net/?p=460

5.1 Linked Open Data A primary objective of the SerenA backend infrastructure is to manage and discover pathways through rich networks of knowledge - combining domain specific concepts, user goals and interests, and real-time context-sensitive information. Therefore, a large body of richly structured knowledge is an essential ingredient for SerenA to adequately fulfill its functional requirements. Some of this information will be generated internally, particularly information regarding individual users gathered from the analysis of free text. However, SerenA has been designed for the Semantic Web, thus enabling access to the rapidly growing body of knowledge that is Linked Open Data. The Linked Open Data graph is an extensive and extensible set of interconnected, machine-readable sources of information, including geographically relevant information, research publication meta-data, and Open Government Data. Linked Open Data is a collection of millions of concepts, described on the distributed Semantic Web. As such, the Semantic Web is a crucial enabling technology, well suited to the ambitions of the project.

5.2 Semantic User Profile Base The design and implementation of the SerenA user profile knowledge base is heavily influenced by the motivations, principles, and technologies underlying the Semantic Web. RDF is used as the data model for SerenA user profiles, which represent information about research interests, goals, social relationships and day-to-day on-line and real-world activities. This knowledge is expressed using several commonly used OWL ontologies, with additional domain-specific specialisations. URIs are used to refer unambiguously to SerenA users, domain-specific concepts and properties, as well as for meta-data concepts necessary for the representation of provenance, data privacy and system control.

5.3 Agent Framework As a Web-distributed agent system, SerenA comprises a number of different kinds of agents, each providing modularized affordances to other agents or users directly. Some agents perform simple tasks, such as querying linked user data, whilst others have goals and long-term motivations, and perform actions coordinated by a BDI runtime system. Additionally, the agent framework is responsible for managing user permissions, and adhering to data privacy requirements. It is anticipated that as the agent community grows, and the level of interaction between agents and with external Linked Data increase, the potential for emergent behaviour, and possibly even serendipity, may increase.

5.4 Agent Communication Language The knowledge passed from Linked Open Data and the user profile database, to the agent framework, is described in RDF, a fundamental data representation in the Semantic Web. As such, the SerenA system has been designed to allow this information to be transparently passed between agents. The message envelopes used for agent communication conform to the common FIPA standard, and the SerenA framework extends this by allowing RDF representation in the content of the message. This brings the agent framework closer to the SerenA knowledge representation, eliminating the need to translate agent dialogue for long term episodic memory.

6. Acknowledgement Thanks to all those who participated in workshops and interviews. This work is funded and supported by SerenA EPSRC (EP/H042741/1) and Horizon Digital Economy Research (RCUK grant EP/G065802/1).