Upload
anna-jordanous
View
1.470
Download
7
Tags:
Embed Size (px)
Citation preview
Measuring cultural value using social network analysis
A case study on valuing electronic musicians
Anna Jordanous (University of Kent, CC/music informatics), Daniel Allington (University of the West of England, SNA/digital
culture), Byron Dueck (Open University, music/ethnomusicology)
Valuing Electronic Music http://valuingelectronicmusic.org
Types of value
Valuing Electronic Music http://valuingelectronicmusic.org
• Monetary value £ $ € – (what a thing is worth)
• “Good” things – e.g. “good” opinions – (Thanks Mike Cook)
• etc …
• Cultural value– Based on what we are culturally led to like– Subjective– Value shared between members of a community/culture
Introduction: Aim of this work
Can we measure the cultural value of a creative entity in a quantitative / algorithmic way?
In other words… Is there a way to judge cultural value, that can be computerised and stuck into the creative
process of a computational creativity system?NB Evaluation is an important part of the
creative process… but this is another talk…
Valuing Electronic Music http://valuingelectronicmusic.org
Evaluating value (as part of evaluating creativity)
We (CC) would like to have metrics for value that we can implement
computationally
Q How do we measure/assess cultural value?…using social network analysis?
Valuing Electronic Music http://valuingelectronicmusic.org
Some premises before we start
1. Evaluating value is an important part of evaluating how creative something is
2. If you want to know how good a creative entity is, ask someone (something) that knows what it means to be good in that way
3. This is a case study in judging value of electronic musicians, but can be useful for other creative domains (with a little work)
Valuing Electronic Music http://valuingelectronicmusic.org
Premise 1. Evaluating value is an important part of evaluating how creative something is
Creativity = novelty + value + …??
Valuing Electronic Music http://valuingelectronicmusic.org
Premise 2. To know how valuable a creative entity is, ask someone/something(s) that knows that creative area
Valuing Electronic Music http://valuingelectronicmusic.org
https://www.youtube.com/watch?v=FZu097wb8wU
e.g. Would you trust Dan Ventura on what art galleries are good in Park City?
Valuing Electronic Music http://valuingelectronicmusic.org
Premise 3: Our approach to measuring value in electronic music can be useful more widely
• ‘One size does not fit all’
• But [I argue] our approach is useful, if you want to analyse cultural value in domains other than electronic music
• (with a little work)
Valuing Electronic Music http://valuingelectronicmusic.org
Premises set out… let’s begin
We (CC) would like to have metrics for value
Q How do we measure/assess cultural value?…using social network analysis?
Case study: electronic music[ians]
– Who are the main ‘players’ (no pun intended..?)
– Who values who? How do they show this?
Valuing Electronic Music http://valuingelectronicmusic.org
Qualitative and Quantitative work informing each other
Interviews SoundCloud data analysis
Valuing Electronic Music http://valuingelectronicmusic.org
e.g. focus on valuing (as a verb), not value (as a noun)
e.g. it is possible to create SC groups - is this useful?
Distinct types of electronic music
Valuing Electronic Music http://valuingelectronicmusic.org
http://valuingelectronicmusic.org/2014/12/19/glitch-lich-live-performance/
http://valuingelectronicmusic.org/2015/01/02/winterlight-live-set/
http://valuingelectronicmusic.org/2015/02/01/slackk-live-set/
http://valuingelectronicmusic.org/2014/11/25/slackk-winterlight-glitch-lich-electronic-music-producers-panel/
(Where we left this - work-in-progress paper ICCC’14)
Valuing Electronic Music http://valuingelectronicmusic.org
Little interest in the ‘top x users’ across the whole groupMore important: Relationships / personal networks
smal ler networks of usersinfl uenced by genresgeographica l ly infl uenced
Who is on SoundCloud?
From analysing ‘follow’ relationships:–By countries, top 3 are US, UK, Germany–By cities, top 3 are London, New York, Berlin–People tend to follow people within same city–http://valuingelectronicmusic.org/2014/09/08/geography-soundcloud-following/
• And people in same genre
–By genre, top 3 are house, hip hop, techno–http://valuingelectronicmusic.org/2014/06/04/exploring-genre-on-soundcloud-part-i/–NB This is not filtered for electronic music genres - but electronic music dominates
Valuing Electronic Music http://valuingelectronicmusic.org
Comments: typically very positive
Valuing Electronic Music http://valuingelectronicmusic.org
Comments highlighted by interviewees
as an important valuing activity
Identifying sub-networks: place
• Ego-network of one of our interviewees and their immediate followers/followees
• Colours = places– London = Green– Bristol = Purple– LA = Red
(Tony what should we call these colours?)
Valuing Electronic Music http://valuingelectronicmusic.org
Identifying sub-networks: genres
• Three distinct macro-genres identified: – EDM– Urban– ‘other’
• (Corresponds with music research groupings)
Valuing Electronic Music http://valuingelectronicmusic.org
Distinct vocabularies per genres
# Techno Dubstep House Hiphop
1 set sick nice dope
2 great tune house shit
3 tracks nice super beat
4 loved big production leave
5 fantastic mix support song
Valuing Electronic Music http://valuingelectronicmusic.org
Top words appearing in comments per genre
Bringing this back to computational creativity - how can CC use this work?
As a proxy for value:– look for [meaningful]
interactions between people linked together in subnetworks / ego networks / cliques= evidence of interpersonal relationships
– On SoundCloud, interactions tend to be (or spam)
Valuing Electronic Music http://valuingelectronicmusic.org
For a ‘CC electronic musician’:– Set up London-based SC
account + upload tracks– Develop the system to
interact with other music-makers in similar genres: • Comment on other tracks and
respond to comments using that genre’s key vocab• Follow users in those genres
– Follow users in key cities (e.g London, NY, Berlin, LA)
Remember: comments usually positive
Valuing Electronic Music http://valuingelectronicmusic.org
Comments highlighted by interviewees
as an important valuing activity
Hmm.. Can commenting be used as a simple proxy for value judgments?
• Using commenting as a valuing activity• # chars track comments = proxy value metric?
EXPERIMENTING: Trying this on Scottish users in our data1. calvinharris (electro-pop, house)2. the-nibelheim-incident (Daryl Constance) (dubstep)3. === middleschoolfrown (John Kevin, electronic pop),
davecruickshank (techno), jasegallacher (techno)
• But differences in commenting behaviour by genre (not just based on the different vocabulary used)
• dubstep producers most prolific (2569 comments)
• then techno (2254), hiphop (2081) and house (1725)Valuing Electronic Music http://valuingelectronicmusic.org
Early days … work in
progress
Conclusions
• You can ‘proxy’ cultural value computationally by studying social network activity– But – look at smaller subnetworks, not the whole– In Soundcloud: networks around place and genre
• How do people show they value others’ work? – Look for relationships/links between people– Work out what linking activities indicate more value
• In Electronic Music it is commenting and collaborating
• We could take advantage of this to measure value computationally, as part of evaluating creativity
Valuing Electronic Music http://valuingelectronicmusic.org