@article{Murray-Rust_Papapanagiotou_Robertson_2015, title={Softening electronic institutions to support natural interaction}, volume={2}, url={https://hcjournal.org/index.php/jhc/article/view/47}, DOI={10.15346/hc.v2i2.3}, abstractNote={<p>A necessary feature of social networks is a model of interaction which is followed on the network---some structure which coordinates activity between the participants. These interaction models are typically implicit, making it a challenge to both design and communicate the protocols for interaction and coordination. Electronic institution systems are one of the principal ways in which multi-agent systems engineers address this issue of coordination in complex interactions between groups of agents. In electronic institutions, interaction models can be concisely specified as protocols which encode the norms which computational agents follow. However, the formality and the up-front costs of discovering and choosing to engage with these systems has limited their applicability to human interaction. The vast majority of human (and, increasingly, automated) social interaction is now taking place in social media systems where social norms are softer concepts regulated essentially by the people involved. Being able to leverage the power of electronic institutions in these systems would ease the application of computational intelligence in support of social tasks.</p><p>We describe a method by which electronic institutions can act in synergy with these sorts of social media streams and, in doing so, we define a ``softer’’ style of system that, nevertheless, retains connection to precise specifications of coordination. In addition, we question the tacit assumption that participating agents deliberately join appropriate institutions. Although our method is independent of choice of social media stream (given a few standard characteristics of these) we describe an implementation of the method using Twitter as a target media stream.</p><p>We illustrate the utility of our approach with an example which benefits from computational coordination, but where the use of a traditional EI would have prohibitive up-front costs. As well as a trace of a synthetic version, we demonstrate the functioning of a complete implementation which can run the example, and discuss how minimal the end-user configuration to setup complex examples can be.</p><p> </p><p> </p>}, number={2}, journal={Human Computation}, author={Murray-Rust, Dave and Papapanagiotou, Petros and Robertson, Dave}, year={2015}, month={Dec.} }