Roles, Trust, and Reputation in Social Media Knowledge Markets: Theory and Methods by Elisa Bertino & Sorin Adam Matei

Author:Elisa Bertino & Sorin Adam Matei [Bertino, Elisa & Matei, Sorin Adam]
Language: eng
Format: epub, pdf
Tags: Science, System Theory, Computers, Data Science, Data Analytics, Mathematics, Applied, social science, Methodology, General, Physics, Mathematical & Computational, Research
ISBN: 9783319054674
Google: d8NpBAAAQBAJ
Publisher: Springer
Published: 2014-09-02T20:23:10.733954+00:00

Invisible Algorithms and Their Alternatives in Practice

Music

The digital revolution enabled a reconsideration of how music is discovered, distributed and organized. The distribution of music has always been bound up with socially constructed notions of taste as well as technical constraints (such as FM, cassette players, radio, size of an amphitheatre, etc.). However, when music became widely available digitally, it became possible to access heretofore unimaginable volumes of music with relative ease and convenience, prompting entirely new means for discovery and organization. Now that music can be integrated into databases that signify and calculate the relatedness of music, one does not need other media to discover new music. Virtually every streaming music service, whether it is Google Play, Spotify, iTunes or Pandora, has some means to impose a machine learning logic onto music. Thus, one can become acquainted with new music through the Internet without getting an explicit recommendation from a specific other person but from a representation of music as a data structure.

Representing music as data poses a number of significant challenges for either invisible algorithms or interactive affordances. This is because music exists within a highly tiered multilevel structure. A song is on an album that is by a band who are on a label. Each song might be more or less popular and include certain guest artists. At each level (song, band, etc.) one might also ascribe a genre (e.g. dance music), a subgenre (house music) or even a subsubgenre (electrohouse). While such features would seem to present an explosion of potential ways to explore new music, in most cases, these features are not represented structurally. Instead, music listening is reduced either to sorting by single rational signals (such as most downloaded, alphabetical, most recent) within any given single column, or via an invisible algorithm based on some amalgam of co-listening and other factors. One of the challenges of representing this relationally is not the paucity of possible relevant features for the graph, but the abundance.

In addition to the invisible algorithms of iTunes Genius, Amazon, Pandora and Google Play, interactive affordances are meant to provide an overview of music. These have yet to be adopted on any broad scale. I contend that one reason for this is the complexity of providing a Graph Search-like capability using database systems designed for sorting and ranking. Thus, we are left with user-curated networks of music, such as Iskur’s Guide to Electronic Music, 2 a flash-based app that shows subgenres networked and linked in larger genres, all subjectively assessed. Other approaches include eMusic’s “Infinite Explorations”, based on user co-downloading, and MusicPlasma, based on Amazon co-buying. In most cases, however, these systems do not provide a categorical system based on sets and filters. Instead, the edges linking different music (typically artists) are themselves based on invisible algorithms.

This leads to an interesting if somewhat underwhelming situation. The emergent maps do not tend to show macroscale features such as clusters. Instead, they merely project invisible algorithms into a different space. For instance, the music app Discovr operates as a music discovery service that draws in streaming from other services.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.