►
Description
Presented by Anton McConville
From 'Space Oddity' to 'Blackstar', Bowie inspired us with his diverse personas. What can natural language processing reveal about those iconic characters? That's the mission of 'Ziggy' a little Python app that builds a database of lyrics, and digs into the style and tone of writing within them. This session explores the findings, and shows how it works. We'll consider real world use cases, and future iterations.
About CodeConf
CodeConf improves the software community by providing a forum for thought-provoking talks and forging social connections. The fourth installment of the CodeConf series took place in Hollywood in 2016. This year's event focused on systems engineering projects, practices, and programs in the open source community.
For more information on this year's CodeConf, go to:
https://codeconf.com/
A
So
damage
behind
me
spells
the
word
bully
using
components
of
a
black
star
and
black
star
with
the
title
of
David
Bowie's
last
album.
So
what's
sort
of
really
fascinating,
for
me
is
just
that's
really
cool
Stedman.
There's
a
lot
of
different
opinion
about
what
the
black
star
metaphorically
means
and
it
being
his
last
album.
It
was
a
statement
from
him.
It
was
an
expression.
A
A
His
expression
through
you
know,
visuals
iconic
images
he
made
and
was
the
songs
that
he
wrote,
but
the
sort
of
more
interesting
application
of
it
is
understanding
the
expression
from
our
clients,
because
we're
on
the
cusp
of
being
able
to
preset
our
clients,
our
users
in
a
way
on
a
grand
scale
individually
for
the
first
time
using
natural
language
analysis
using
their
text,
in
their
words
their
conversations
as
a
way
of
understanding
them
and
then
the
third.
So
development
of
expression
is
through
digital
creativity.
A
So
when
I
think
of
Bowie
I,
think
of
these
amazing
iconic
images
that
he
made
these
characters
that
he
gathers
throughout
his
career,
constantly
changing
so
Ziggy
Stardust
Aladdin
Sane,
the
sin
white
dick
in
the
metal
hero,
the
and
Lazarus
at
the
end,
just
examples
of
Bowie's
personas
when
he
died
or
when
I
heard
that
he
died
earlier
at
the
beginning
of
the
year.
I
was
I
was
really
surprised,
very
surprised
that
he
died.
But
it's
really
most
of
all
surprised
at
how
I
felt
about
it.
A
I
didn't
know
him
I'm,
not
the
world's
biggest
bully
fan,
but
I
felt
a
loss.
I
felt
something
can
something
missing
and
I
reflected
on
it.
Quite
a
lot
I
was
watching
concerts
on
YouTube
and
interviews
on
YouTube
and
I
realized
that
it
was
his
songs.
It
was
some
of
the
songs
that
he'd
written
I
have
really
clear,
strong
memories
of
listening
to
Annie
Nightingale
on
BBC
Radio
1.
When
I
was
a
teenager,
devising
for
exams
play
an
absolute
beginners.
A
I've
got
really
strong
memories
of
seeing
bully
playing
the
old
name
cards
of
stairs
oddity
had
an
open-air
concert
in
the
north
of
London
when
I
was
an
intern
and
I
really
warm
summer's
day
among
friends.
So
it's
the
music,
it's
the
songs
and
it's
the
connection
that
those
songs
had
that
made
me
think
of
people
made
me
think
of
times
and
maybe
think
of
loss
and
the
more
I
thought
about
it.
A
They
used
it
a
few
years
ago
to
win
Jeopardy
natural
language
analysis
so
where
I
can
take
written
words
and
make
sense
of
it,
while
a
computer
can
make
sense
of
it
either
the
emotional
tone
of
it.
It
can
pick
out
the
emotions-
or
in
this
case
it
can
give
a
psychological
profile
based
on
what's
written
down
there,
and
so
using
this
approach.
So
I
thought
I
could
get
almost
like
an
identity
for
each
of
these
personas,
based
on
their
the
lyrics
in
the
music.
A
So
what
some
personality
insights
uses?
This
theory
from
psychology
called
the
Big
Five
personality
traits.
So
openness
is
the
measure
of
creativity.
Conscientiousness
is
a
measure
of
organization.
Self-Discipline
extraversion
versus
introversion
is
how
much
you
project
I
would
leave
vices
in
middly
agreeableness.
Is
your
ability
to
cooperate
and
to
work
with
others
and
neuroticism
is
more
about
yourself
focused,
so
the
software
can
take
text
and
break
it
into
percentages.
A
A
This
is
a
timeline.
This
is
how
he
changed
over
time.
So
when
I
click
on
one
of
these
different
persona
characters,
it
will
take
me
to
a
sort
of
breakdown
of
the
personality
traits
of
that
individual
persona
and
the
way
that
I
present
it
is
in
the
form
of
a
mock
Twitter
page.
So
I
thought
I
wanted
it
to
be
colorful
and
interesting.
A
I
didn't
want
to
present
like
dry
data
in
the
form
of
like
pie,
charts
and
bar
charts,
and
what
better
way
to
emphasize
the
fact
that
their
individual
personas
done
with
a
social
media
account
for
each
of
them.
I
was
pretty
fearful
for
the
template
that
Twitter
for
the
lazy
eye.
You
know,
I
have
a
little
description.
A
The
location
is
where
the
album
was
recorded
at
the
time
is,
instead
of
when
the
joint
Twitter
is
the
time
frame
was
when
the
albums
were
recorded,
and
that
is
from
album
art
and
there
in
each
of
the
tweets.
It
has
one
of
those
big
five
personality
traits,
so
there's
an
openness,
tweet,
a
conscientiousness,
tweet
and
so
on
and
they're
detected
in
the
form
of
Sankey
diagrams.
A
A
If
you
end
up
with
maybe
20
or
30
different
data
points
of
you
want
to
analyze
the
psychology
of
that
written
text
over
here.
There's
a
histogram
plotting,
the
percentage
of
each
of
them
will
be
able
to
miss
conscientiousness
and
we're
all
in
the
spectrum.
For
whatever
way
we
present
ourselves
through
psychology,
you
can
navigate
to
the
other
different
characteristics
may
through
the
follow
area.
So
I
can
click
on
one
of
them
and
look
at
a
different
persona.
A
So
that's
pretty
interesting,
but
where
perhaps
gets
a
little
bit
more
interesting
is
comparing
them
the
Internet's
a
little
bit
slow
here
there
we
go
so
there
are
five
tiles
here,
one
for
each
personality
to
it
and
in
each
of
the
tiles
it's
given
a
measure
over
time.
As
long
as
that
personality
trait
is
changing.
A
So
what's
really
cool
about
that
is
it's
watching.
Hi
a
personality
is
changing
over
time.
So,
if
we
look
at
it
in
this
form,
it's
I
vividly.
Ryx
are
changing
over
time.
You
can
see
and
they
open
this
chart.
That
is
pretty
consistent
and
that
sort
of
areas-
90th
percentile,
there's
sort
of
duration
of
his
career,
but
there's
a
marked
dip
in
1975
and
that's
reflected
also
in
the
agreeableness
chart.
There's
a
March
spike
in
1975
he's
not
at
all
agreeable,
isn't
was
notoriously
not
an
agreeable
person.
A
There's
a
great
garden
article
had
things
Bowie
turned
on,
so
he
turned
out
as
a
knighthood.
For
example,
he
turned
on
collaboration
with
Chris,
Martenson,
colse
and
so
on.
But
it's
curious.
You
know
this.
Is
this
debt,
this
1975?
Why?
What
makes
that
so
special?
So
it's
a
little
toggling
to
let
you
compare
persona
by
persona
and
look
at
that
corresponds
to
this
in
white
Duke
era
and
that
after
doing
some
readings
and
research
is
when
Bowie
was
at
his
lowest
point.
He
had
major
drug
addictions
with
cocaine.
A
He
was
actually
living
in
here
in
Los
Angeles
and
it's
my
third
era.
He
was
recording
things
like
young
Americans
saying
and
he
got
clean
after
that,
so
he
went
to
Germany
to
the
Berlin,
trilogy
and
I
think
there's
only
bicep
until
2002
smoking
cigarettes.
So
it
was
pretty
curious.
That's
about
depth.
A
A
Reflection
makes
sense
that
it
was
extrovert.
So
that's
just
some
of
the
sort
of
dipping
into
some
of
the
different
analysis.
I
put
a
pecker
in
there
as
well,
so
that
you
can
start
to
look
at
some
of
the
different
and
fading
attributes
as
well.
So
again
you
can
see
in
the
conscientiousness
the
thin
white
duke
has
a
spikes
or
orderliness
compared
with
the
others.
A
A
The
last
sort
of
piece
of
it
is
I
can
answer
a
Twitter
ID,
so
code
counts
and
whatever
will
do
is
it'll
go,
read,
tweets
from
that
Twitter
account
and
run
those
tweets
through
personality
insights
and
then
map
them
to
the
closest
matching
boli
persona.
So
this
shows
you
that
the
best
on
sort
of
psychological
profiles
and
natural
language
analysis.
A
Not
only
can
you
start
to
analyze
the
psychology
behind
text,
but
you
can
also
start
to
target
it
and
make
decisions
on
axis
touches
on
that
idea
of
being
able
to
automatically
have
responses
based
on
a
person's
personality
type.
So
here
COCOMs
is
most
matching
the
bully
of
modern
love,
so
Bowie's
added
booyah
the
brightest
and
most
obvious
that
is
sort
of
appropriate,
say
orange
taunt.
The
orange
lines
in
the
right
are
chart
represent
code.
You
can
see
that
COCOMs
content
conscientiousness
is
stronger
than
the
bully
characters.
Yet
the
emotional
rant
is
lacking.
A
So,
like
I
was
saying,
this
is
a
Python
app.
What
some
personality
insights
is
something
that
you
can
access
through:
bloomix
bloomix
Azure,
a
lot
like
Amazon,
Web,
Services
or
Google
client
accepted
that
also
has
Elijah
run
yourselfer,
but
also
has
access
to
lots
of
services
like
the
cognitive
services
and
Watson
cognitive
computing
sides
like
big
deal,
but
it's
actually
only
one
line
of
code.
It's
a
one-line
call
to
get
those
personality,
insights
bypass
them
in
a
bunch
of
text.
A
For
me,
I've
got
a
background
and
design
and
development
on
one
of
the
more
troubled
areas
I
have
whenever
I'm
working
with
something
is
picking
colors
but
twice
a
year
pencil
and
give
us
this
amazing
guest
and
spring
and
fall.
They
offer
colors
of
the
air,
and
so
this
is
my
my
tip
to
if
you
don't
know
about
it
and
you're
interested
in
picking
colors.
A
All
of
the
bully
characters
were
specially
colored
from
this
palette,
so
what
I
did
was
and
all
of
the
photographs
that
I
took
and
pinned
to
the
Pinterest
board,
I
matched
them
to
those
colors
and
the
work
seamlessly
together.
The
color
experts
have
said
it
out
away
she
to
not
make
mistakes
when
combining
them.
A
I'm
really
really
strongly
opinionated
about
the
use
of
libraries
in
web
development.
So
all
of
the
app
was
built
using
just
flax
boxes.
There's
no
bootstrap,
there's
no
other
web
spam
workers
as
peers
and
LR
JavaScript
I
love
working
with
flax
boxes.
It's
a
revelation
for
me
to
actually
be
able
to
layout
webpages
properly
for
the
first
time,
it's
a
joy
to
be
able
to
use
if,
if
you're
not
familiar
with
them
again,
that's
my
my
tip
to
is
to
learn
them
because
it
makes
development
so
much
more
fun.
A
I
also
think
there's
a
real
tight-knit
miss
the
data
driving
the
app
the
content
behind
it
and
the
Flex
boxes
are
presented.
There's
no
sort
of
really
simple
information
model
behind
this
app
as
well.
So
there's
a
database
of
personas,
a
database
of
albums
and
a
database
of
songs.
Every
persona
has
one
to
em
album
an
album
is
really
selected.
Songs
and
the
songs
are
in
a
dilemma.
X
of
the
songs
are
in
a
database.
A
This
is
a
sort
of
JSON
model
for
a
persona.
It's
not
really
rocket
science,
but
it
sort
of
is
that
the
life
and
soul
of
the
whole
app
in
that
you
know
the
sketches
that
I
did.
This
go
back
to
graphics,
sketches
that
are
marked
and
they're
the
start
and
end
point
sort
of
each
of
the
personas
a
little
description,
and
that,
of
course,
the
album's
and
the
Wikipedia
links
in
there.
A
A
So
it's
a
really
perfect
quote
and
all
of
this
stuff
that
I
showed
you
is
open-source.
You
can
get
to
it
from
this
repo.
This
is
the
IBM
bluemix
team's
repo
as
well,
so
there's
lots
of
other
stuff
in
there.
Although
I
use
Python
for
personality
and
slight
software
runs,
and
there
are
no
de
P
eyes
and
swift
IP
eyes
and
Java
API
and
restful
ones,
so
it's
pretty
easy
to
use
for
other
applications.