User-centered design of a social game to tag music

Abstract

We present "Herd It", a competitive, online, multi-player game that has the implicit benefit of collecting tags for music. We describe Herd It's user-centered design process and demonstrate that the game can collect both musical and social data. This data can be used to build machine learning models that automatically associate music with tags. Herd It differs from previous "games with a purpose" in that it is designed to be social: the game runs on the Facebook online social network and scoring is based on consensus between a large group of listeners -"the Herd". By presenting music in a social context, Herd It adds demographic context to the semantic music descriptions that it collects. KeywordsAuditory I/O and sound in the UI, Distributed knowledge acquisition, Computer audition, Web-based games with a purpose

An Affordance-Based Framework for Human Computation and Human-Computer Collaboration

“…1c). We have also surveyed examples utilizing human audio linguistic ability for audio annotation [5,44,49], transcription [16], and even crowdsourced word processing [8].…”

“…Another important affordance of human collaborators is creativity 5 . As noted by Fitts [27], Dekker [22] and many others, humans are capable of incredible creativity, generating spontaneous arrhythmic approaches to problems that may be difficult or impossible to simulate.…”

“…For example, systems like Verbosity [75] and others [43,45] aggregate and store information generated by human users to create commonsense knowledge repositories. It is also used in the collection of behavioral scripts for autism treatment [11] and human-robot interaction [17], as well as collecting tags for music and image annotation [5,12,21,49,59,62,73,74]. In a world that is growing ever more big data-centric, storage capacity and efficient retrieval are critical advantages afforded by machine collaborators.…”

An analytical model for generalized ESP games

“…By taking advantage of people's desire to be entertained and exploiting "human cycles" in computation, GWAP not only attract people to play voluntarily, but also produce useful metadata as a by-product. The paradigm has shown promise in solving a variety of problems, such as image annotation [3,4,5], audio annotation [6,7,8,9], and commonsense reasoning [10,11], which computer programs have been unable to resolve completely thus far.…”

“…Unlike the ESP game that collects independent guesses from randomly paired participants, [20,21] use a set of candidate labels known by all participants and promote coordination among participants to complete game tasks. In addition, the Herd It [6,8], Major Miner [9], and TagATune [7] systems, which provide annotation for sounds and music, can improve audio searches. The Verbosity system [11] and the Common Consensus system [10] collect "common-sense" knowledge that is valuable for commonsense reasoning and enhancing the design of interactive user interfaces.…”