artist popularity: do web and social music services agree?
DESCRIPTION
Poster on ICWSM '13TRANSCRIPT
Artist Popularity: Do Web and Social Music Services Agree? Alejandro Bellogín, Arjen P. de Vries, Jiyin He
Centrum Wiskunde & Informatica, The Netherlands
Universidad Autónoma de Madrid, Spain
{a.bellogin, arjen.de.vries, j.he}@cwi.nl
7th International AAI Conference on Weblogs and Social Media (ICWSM 2013)
Boston, USA, 8-10 July 2013
Practical implications
No two indices are completely equivalent: we can
use them to promote diversified recommendations
Service-based indices change at a much lower speed
The Web index is highly dynamic
Limitations
No constraint on regions, but concerts and culture are
local events, important within a specific community
Positive vs negative popularity: exploit the context
(where an artist is mentioned) in the Web index
Open questions
Are users more interested in currently popular artists
or they prefer those that are always popular?
Should all the reasons why an artist could be popular
considered equally important?
Tag clouds
rockalternative
pop
alternative rock
Hip−Hop
rnbclassic rock
indie
dance
Nu Metal
metal rap
female vocalists
indie rockpunk rock
alte rnati ve metal
hard rock
hip hop
punkbritish
electronic
Pro
gre
ssiv
e r
ock
americanGrunge
80s
60s
pop punk
glam rock
blues
female vocalist
Akon
Rihanna
Nu−metal
00s
male vocalists
r&b
70s
Lady Gagarapcore
LinkinPark
britpop
Queen
soul
black eyed peas
90s
green day
The Killers
electropop
funk
Muse
post−g runge
se
xy
System of a D own
USA
blues rock
emo
favorites
foo fighters
po
p r
ock
reggae
female
heavy metal
hiphop
political
Rock and Roll
urban
cross over
dance−pop
legend
singer−songw riterthe rolling stones
da
nce
ha
ll
electronica
english
experimental
gaga
glam
Progressi ve
rhythm and blues
UK
rockalternative
alternative rock
indie
indie rock
punk rockpunk
Nu
Me
tal
metal
pop
pop punk
alternative metal
britishdance
electronic
Progressi ve rock
hard rock
Grunge
femalevocalists
american
britpop
female vocalist
Nu−metal
emo
Lady Gaga
rapcore
00
s
Linkin Park
Pop−punk
90s
green day
Th
e K
ille
rs
Mu
se
post−grunge
System of a Down
ele
ctr
op
op
male vocalists
favorites
foo fighters
pop rock
USA
arctic mon keys
Blink 182
political
cro
ss
ove
rh
ea
vy m
eta
l
singer−songw riter
UKAwesome
blin
k−
18
2
brit rock
experimental
gaga
Garage Rock
indie poppost−punk
Progressi ve
sexy
skate punk
Ca
lifo
rnia
div
a
electronica
english
glam
hardcore
new wavePost−punk r evival
so
ad
Hip−Hopdance
Housejazz
rapelectronic
pop
rnb
female vocalists
swing
hip hop
club old
ies
vocal jazz
west coast
techno
blues
Dirty South
jazz
vocal
easy listening
americanRihanna
electronica
G−funk
malevocalists
swedish
electro
Flo Ridasoul
french
r&b
Snoop DoggDavid Guetta
femaletrance
dj
hiphop
50s
frank sinatra
reggae
Big
Ba
nd
gangsta
sexy
00s
crooner
dancehall
USA
vocal
Avicii
barbadian
barbados
California
deep house
dubstep
ella
fitzg
era
ld
floridalegend
lounge
pa
rty
rock
tech
ho
use
EchoNest
Last.fm
Spotify
playcounts familiarity popularity
Artist’s
name
clicks
webpages
1
2
cleaned
webpages
3
4
Popularity indices
Service-dependent Web-based
C
A B
Temporal dynamics
A. Removed from stage at Obama’s reelection. B. Announcement of new album’s title. C. 3 Grammy awards received
Stability of popularity rankings
Ranking-based comparison
Correlations
Kendall
Spearman
Check the source code and dataset at:
http://www.cwi.nl/~media/papers/ICWSM13
Dataset
1,312 artists
Timestamped popularity indices (~ every 3 hours)
Jan 21 – Mar 21
including the Grammy awards and The Oscars
Change in Jaccard similarity
when more artists are
considered
Top-5 popular
artists throughout
the timespan of
our dataset
@abellogin
varyatanil