Music-related collective influences, trends and behaviours

From MIReS

Jump to: navigation, search

Music is a social phenomenon, thus its understanding and modelling requires the inclusion of this dimension. Social interaction is a driving force of music listening, categorisation, preference, purchasing behaviour, etc. Additionally, teams or crowds are usually able to achieve feats that go beyond what individuals accomplish, and this is especially relevant for annotation and other collaborative scenarios. Finally, scattered in different virtual places, formats and time-scales, there is much data available that contains implicit information about music-related social factors, which could make possible the understanding and prediction of trends and other collective behaviours related to music. To carry out this research, which would complement the other, more traditional, approaches to music description, we need to involve people working in Social Computing, Sociologists and experts in Dynamic Systems and Complex Networks. Social computing will help to gather massive annotations and obtain knowledge on the key actors and factors of collective-mediated processes in musical choice, interaction and conceptualisation. Human dynamics will make possible massive-scale predictions about trends, ways and moments to listen to music, and provide pointers to the best locations and conditions for commercial and marketing activities (e.g., Buzzdeck). The main obstacle to promising advances is the scarcity of open data and the privacy issues associated with access to data. Contrastingly, there are also issues to be solved when managing and analysing extremely large amounts of data of this kind.


Back to → Roadmap:Socio-cultural perspective


Contents

State of the art

Even though most of the XXth Century technologies have made possible different modes of experiencing music individually, if we consider all the cultures in the world, music is still mostly experienced and valued in a social context. Even in the Western culture individual listening becomes a social activity as the experience is frequently, afterwards or simultaneously, interpreted and shared with other people. Hence, the value of music as social mediator and the social dynamics it makes possible have yet to be properly addressed by researchers. In addition to a traditional view corresponding to the social psychology of music/sociology of music (see section Knowledge-driven methodologies) we consider two research perspectives on social aspects of music: human dynamics and social computing.

How has MIR addressed, supported or capitalised on the social aspects of music? What is still to be done? As an orientation, the word "social" can be found in more or less 100 papers presented in the past 12 ISMIR editions but in most of the cases it is just a passing word, or part of a somewhat shallow expression like "social tags" or "social networks". In the bunch of papers that really deal with social aspects, social psychology and social computing are dominant perspectives, whereas human dynamics has been, up to now, absent.

Most basic research on social aspects of music has focused on individuals with relation to significant groups (i.e., peers, family, gang, nation), as we have summarised above. Alternatively, social behaviour can be considered globally, nearly getting rid of the individual (we cannot avoid the link to Asimov’s "psychohistory", like researchers on social animals (especially insects) usually do. A global understanding of the flow patterns of spread, influence and consumption/enjoyment of a specific musical agent or content calls for new techniques such as complex network analysis or human dynamics [Barabási, 2005]. Our knowledge of the interplay between individual activity and social network is limited, partly due to the difficulty in collecting large-scale data that record, simultaneously, dynamical traces of individual behaviours, their contexts and social interactions. This situation is changing rapidly, however, thanks to the pervasive use of mobile phones and portable computing devices. Indeed, the records of mobile communications collected by telecommunication carriers provide extensive proxy of individual symbolic and physical behaviours and social relationships. The high penetration of mobile phones implies that such data captures a large fraction of the population of an entire country. The availability of these massive CDRs (Call Detail Records) has made possible, for instance, the empirical validation in a large-scale setting of traditional social network hypotheses [Wang et al., 2011]. Taking advantage of them for music-related purposes is still pending because massive geo-temporally tagged data is still one of the bottlenecks for MIR researchers. We are still lacking knowledge about listening patterns and how they are modulated by interaction with peers, by sharing of musical information with peers, or by geographical and environmental conditions (e.g., weather, time of the day) [Herrera et al., 2010]. In order to study massive concurrent behaviour patterns we only have available a large dataset of last.fm scrobblings harvested by Òscar Celma. It is interesting to note that the most recent "Million Song Dataset" does not include any geo-temporal information. Telecommunication service companies should then be targeted by researchers and research project managers in order to make some progress along this line.

The social computing view, on the other hand, addresses either the creation of social conventions and context by means of technology (i.e., wikis, bookmarking, networking services, blogs), or the creation of data, information and knowledge in a collective and collaborative way (e.g., by means of collaborative filtering, reputation assignment systems, tagging [Lamere, 2008], game playing [Ahn, 2006], collaborative music creation tools, etc.). It is usually assumed that social computation, sometimes also called social information processing, will be more effective and efficient than individual or disconnected efforts [Surowiecki, 2004]. When information is created socially, it is not independent of people, but rather is significant precisely because it links to people, who are in turn associated with other people [Erickson, 2011]. Games with a purpose (GWAP) are a paradigmatic example of social computation for annotation of different knowledge domains. Major Miner, The Listen Game, TagATune, MagnaTagATune [Law et al., 2009], Moodswings [Kim et al., 2008], Mooso, HerdIt [Barrington et al., 2009], etc., have been successfully used for gathering massive ground-truth "annotations" of music excerpts or for generating data about music preference or relatedness (see above section Collecting music related data). A further step in generating knowledge consists in building ontologies from tagging and writing behaviour inside a delimited social network [Levy and Sandler, 2007]; [Pan et al., 2009]. A unified model of social networks and semantics where social tagging systems can be modelled as a tripartite graph with actors, concepts and instances (e.g., songs or files) makes possible, by analysing the relations between concepts both on the basis of co-occurrence in instances and common usage by actors (users), the emergence of lightweight ontologies from online communities [Mika, 2007]. A completely different approach to community knowledge extraction for the design of ontologies is the implementation of Web portals with collaborative ontology management capabilities [Zhdanova, 2006]. It has been recently reported on these strategies related to the Freesound community [Font et al., 2012]. In addition to games and tag-related activity, collective musical knowledge can be generated by means of musical activity itself (and not just by tags or texts). Collective generation of playlists has been studied under different perspectives [e.g. Sprague et a., 2008; Stumpf and Muscroft, 2011]. Precisely in this category Turntable.fm (unavailable in many European countries) is one of the recent successful musical apps for the iPhone (but see also Patent US7603352, or just the collective playlist creation function as available in Spotify). Mashups [Sinnreich, 2010] are another contemporary type of music content that benefits from music audio and context analysis technologies [Griffin et al., 2010] although it is still pending to study how collective knowledge emerges inside communities that are focused on them. To conclude, a proper multidisciplinary forum to discuss music social computation would be the "International Conference on Social Computing, Behavioral Modeling and Prediction" (held since 2008).


References


Challenges



Back to → Roadmap:Socio-cultural perspective

Retrieved from "http://www.mires.cc/wiki/index.php/Music-related_collective_influences,_trends_and_behaviours"
Personal tools
Namespaces
Variants
Actions
Navigation
Documentation Hub
MIReS Docs
Toolbox