TED2011
Deb Roy: The birth of a word
Deb Roy: Rođenje riječi
Filmed:
Readability: 4.2
2,809,941 views
Deb Roy, istraživač na MIT-u, želio je shvatiti kako je njegov novorođeni sin naučio jezik. Zato je po svojoj kući postavio videokamere kako bi zabilježio svaki trenutak (s nekim iznimkama) života svog sina. Zatim je parsirao 90.000 sati kućnih videosnimki da bi vidio kako se "gaaaa" polako pretvorilo u "voda". Zadivljujuće istraživanje, s mnoštvom podataka i s dubokim uvidom u proces učenja.
Deb Roy - Cognitive scientist
Deb Roy studies how children learn language, and designs machines that learn to communicate in human-like ways. On sabbatical from MIT Media Lab, he's working with the AI company Bluefin Labs. Full bio
Deb Roy studies how children learn language, and designs machines that learn to communicate in human-like ways. On sabbatical from MIT Media Lab, he's working with the AI company Bluefin Labs. Full bio
Double-click the English transcript below to play the video.
00:15
Imagine if you could record your life --
0
0
4000
Zamislite da možete snimiti svoj život --
00:19
everything you said, everything you did,
1
4000
3000
sve što ste rekli, sve što ste učinili,
00:22
available in a perfect memory store at your fingertips,
2
7000
3000
dostupno savršeno pohranjeno, na dohvat ruke,
00:25
so you could go back
3
10000
2000
tako da se možete vratiti,
00:27
and find memorable moments and relive them,
4
12000
3000
pronaći značajne trenutke i ponovno ih proživjeti,
00:30
or sift through traces of time
5
15000
3000
ili preletjeti kroz tragove vremena
00:33
and discover patterns in your own life
6
18000
2000
i otkriti uzorke u vlastitom životu
00:35
that previously had gone undiscovered.
7
20000
3000
koje ranije niste ni primijetili.
00:38
Well that's exactly the journey
8
23000
2000
Točno je na takvo putovanje
00:40
that my family began
9
25000
2000
krenula moja obitelj
00:42
five and a half years ago.
10
27000
2000
prije pet i pol godina.
00:44
This is my wife and collaborator, Rupal.
11
29000
3000
Ovo je moja supruga i suradnica, Rupal.
00:47
And on this day, at this moment,
12
32000
2000
Ovog smo dana i u ovom trenutku
00:49
we walked into the house with our first child,
13
34000
2000
ušli u kuću s našim prvim djetetom,
00:51
our beautiful baby boy.
14
36000
2000
našim prekrasnim sinom.
00:53
And we walked into a house
15
38000
3000
Ušli smo u kuću
00:56
with a very special home video recording system.
16
41000
4000
s vrlo posebnim sustavom za kućno snimanje.
01:07
(Video) Man: Okay.
17
52000
2000
(Video) Muškarac: U redu.
01:10
Deb Roy: This moment
18
55000
1000
Deb Roy: Taj trenutak
01:11
and thousands of other moments special for us
19
56000
3000
i tisuće drugih posebnih trenutaka
01:14
were captured in our home
20
59000
2000
zabilježeni su u našem domu
01:16
because in every room in the house,
21
61000
2000
jer u svakoj sobi u našoj kući,
01:18
if you looked up, you'd see a camera and a microphone,
22
63000
3000
ako biste podigli pogled, vidjeli biste kameru i mikrofon,
01:21
and if you looked down,
23
66000
2000
a ako biste pogledali prema dolje,
01:23
you'd get this bird's-eye view of the room.
24
68000
2000
dobili biste ptičju perspektivu sobe.
01:25
Here's our living room,
25
70000
3000
Ovo je naša dnevna soba,
01:28
the baby bedroom,
26
73000
3000
dječja soba,
01:31
kitchen, dining room
27
76000
2000
kuhinja, blagovaonica,
01:33
and the rest of the house.
28
78000
2000
i ostatak kuće.
01:35
And all of these fed into a disc array
29
80000
3000
Sve se to spremalo na niz diskova
01:38
that was designed for a continuous capture.
30
83000
3000
izrađenih za trajno snimanje.
01:41
So here we are flying through a day in our home
31
86000
3000
Sada ubrzano gledamo jedan dan u našem domu.
01:44
as we move from sunlit morning
32
89000
3000
Krećemo od sunčanog jutra
01:47
through incandescent evening
33
92000
2000
preko osvijetljene večeri
01:49
and, finally, lights out for the day.
34
94000
3000
i na kraju gasimo svjetla.
01:53
Over the course of three years,
35
98000
3000
Tijekom tri godine
01:56
we recorded eight to 10 hours a day,
36
101000
2000
snimali smo 8 do 10 sati na dan,
01:58
amassing roughly a quarter-million hours
37
103000
3000
što na kraju čini oko četvrt milijuna sati
02:01
of multi-track audio and video.
38
106000
3000
paralelno snimanog audio i video materijala.
02:04
So you're looking at a piece of what is by far
39
109000
2000
Dakle, gledate dio daleko najveće
02:06
the largest home video collection ever made.
40
111000
2000
kolekcije kućnih videa ikad napravljene.
02:08
(Laughter)
41
113000
3000
(Smijeh)
02:11
And what this data represents
42
116000
2000
Ono što ovi podaci predstavljaju
02:13
for our family at a personal level,
43
118000
4000
našoj obitelji na osobnoj razini,
02:17
the impact has already been immense,
44
122000
2000
utjecaj je već do sada ogroman,
02:19
and we're still learning its value.
45
124000
3000
i još uvijek otkrivamo njihovu vrijednost.
02:22
Countless moments
46
127000
2000
Bezbrojni trenuci
02:24
of unsolicited natural moments, not posed moments,
47
129000
3000
neglumljenih prirodnih trenutaka, nenamještenih,
02:27
are captured there,
48
132000
2000
ovdje su zabilježeni,
02:29
and we're starting to learn how to discover them and find them.
49
134000
3000
i polako ih učimo otkrivati i tražiti.
02:32
But there's also a scientific reason that drove this project,
50
137000
3000
Ali postoji i znanstveni razlog za ovaj projekt,
02:35
which was to use this natural longitudinal data
51
140000
4000
a to je da upotrijebimo ove prirodne linearne podatke
02:39
to understand the process
52
144000
2000
kako bismo shvatili proces
02:41
of how a child learns language --
53
146000
2000
kojim dijete uči jezik --
02:43
that child being my son.
54
148000
2000
a to dijete je moj sin.
02:45
And so with many privacy provisions put in place
55
150000
4000
I tako, s mnogim ograničenjima zbog privatnosti,
02:49
to protect everyone who was recorded in the data,
56
154000
3000
kako bismo zaštitili sve koji su snimljeni,
02:52
we made elements of the data available
57
157000
3000
učinili smo dijelove podataka dostupnima
02:55
to my trusted research team at MIT
58
160000
3000
mojem istraživačkom timu na MIT-u
02:58
so we could start teasing apart patterns
59
163000
3000
kako bismo mogli početi izdvajati uzorke
03:01
in this massive data set,
60
166000
3000
u tom ogromnom skupu podataka,
03:04
trying to understand the influence of social environments
61
169000
3000
pokušavajući shvatiti utjecaj društvene sredine
03:07
on language acquisition.
62
172000
2000
na usvajanje jezika.
03:09
So we're looking here
63
174000
2000
Sada gledamo
03:11
at one of the first things we started to do.
64
176000
2000
jednu od prvih stvari koje smo počeli raditi.
03:13
This is my wife and I cooking breakfast in the kitchen,
65
178000
4000
Ovdje moja supruga i ja spremamo doručak u kuhinji.
03:17
and as we move through space and through time,
66
182000
3000
Kako se krećemo kroz prostor i vrijeme,
03:20
a very everyday pattern of life in the kitchen.
67
185000
3000
svakodnevni uzorak života u kuhinji.
03:23
In order to convert
68
188000
2000
Kako bismo pretvorili
03:25
this opaque, 90,000 hours of video
69
190000
3000
ovih mutnih 90.000 sati snimaka
03:28
into something that we could start to see,
70
193000
2000
u nešto što možemo jasno vidjeti,
03:30
we use motion analysis to pull out,
71
195000
2000
koristimo analizu pokreta kako bismo izvukli,
03:32
as we move through space and through time,
72
197000
2000
dok se krećemo kroz prostor i vrijeme,
03:34
what we call space-time worms.
73
199000
3000
ono što zovemo prostorno-vremenskim crvima.
03:37
And this has become part of our toolkit
74
202000
3000
I to je postao dio našeg alata
03:40
for being able to look and see
75
205000
3000
koji smo koristili kako bismo vidjeli
03:43
where the activities are in the data,
76
208000
2000
gdje se aktivnosti nalaze u podacima,
03:45
and with it, trace the pattern of, in particular,
77
210000
3000
i njima pratimo uzorke, konkretno,
03:48
where my son moved throughout the home,
78
213000
2000
kuda se moj sin kretao kroz kuću,
03:50
so that we could focus our transcription efforts,
79
215000
3000
kako bismo mogli rad na transkriptima
03:53
all of the speech environment around my son --
80
218000
3000
i govornu okolinu mogli usredotočiti na mojeg sina --
03:56
all of the words that he heard from myself, my wife, our nanny,
81
221000
3000
sve riječi koje je čuo od mene, moje supruge, naše dadilje,
03:59
and over time, the words he began to produce.
82
224000
3000
i, s vremenom, na riječi koje je počeo izgovarati.
04:02
So with that technology and that data
83
227000
3000
S tom tehnologijom i tim podacima
04:05
and the ability to, with machine assistance,
84
230000
2000
i s mogućnošću da
04:07
transcribe speech,
85
232000
2000
uz pomoć strojeva transkribiramo govor,
04:09
we've now transcribed
86
234000
2000
do sada smo transkribirali
04:11
well over seven million words of our home transcripts.
87
236000
3000
više od sedam milijuna riječi s naših kućnih transkripata.
04:14
And with that, let me take you now
88
239000
2000
S tim na umu, sad ću vam predstaviti
04:16
for a first tour into the data.
89
241000
3000
prvi pregled podataka.
04:19
So you've all, I'm sure,
90
244000
2000
Svi ste, siguran sam,
04:21
seen time-lapse videos
91
246000
2000
vidjeli ubrzane videosnimke,
04:23
where a flower will blossom as you accelerate time.
92
248000
3000
na kojima je ubrzano prikazan cvat cvijeta.
04:26
I'd like you to now experience
93
251000
2000
Sada bih vam volio dočarati
04:28
the blossoming of a speech form.
94
253000
2000
cvat govornog oblika.
04:30
My son, soon after his first birthday,
95
255000
2000
Moj sin je, nedugo nakon prvog rođendana,
04:32
would say "gaga" to mean water.
96
257000
3000
govorio "gaga", što je značilo "voda".
04:35
And over the course of the next half-year,
97
260000
3000
Tijekom narednih pola godine
04:38
he slowly learned to approximate
98
263000
2000
polako se približavao
04:40
the proper adult form, "water."
99
265000
3000
ispravnom obliku koji koriste odrasli, "voda".
04:43
So we're going to cruise through half a year
100
268000
2000
Sad ćemo proletjeti kroz pola godine
04:45
in about 40 seconds.
101
270000
2000
u otprilike 40 sekundi.
04:47
No video here,
102
272000
2000
Ovdje nema videozapisa,
04:49
so you can focus on the sound, the acoustics,
103
274000
3000
tako da se možete usredotočiti na zvuk, na akustiku
04:52
of a new kind of trajectory:
104
277000
2000
nove vrste putanje:
04:54
gaga to water.
105
279000
2000
od "gaga" do vode.
04:56
(Audio) Baby: Gagagagagaga
106
281000
12000
(Zvuk) Dijete: Gagagagagaga
05:08
Gaga gaga gaga
107
293000
4000
Gaga gaga gaga
05:12
guga guga guga
108
297000
5000
guga guga guga
05:17
wada gaga gaga guga gaga
109
302000
5000
wada gaga gaga guga gaga
05:22
wader guga guga
110
307000
4000
wada guga guga
05:26
water water water
111
311000
3000
voda voda voda
05:29
water water water
112
314000
6000
voda voda voda
05:35
water water
113
320000
4000
voda voda
05:39
water.
114
324000
2000
voda.
05:41
DR: He sure nailed it, didn't he.
115
326000
2000
DR: Odlično je to svladao, za ne?
05:43
(Applause)
116
328000
7000
(Pljesak)
05:50
So he didn't just learn water.
117
335000
2000
I nije naučio samo vodu.
05:52
Over the course of the 24 months,
118
337000
2000
Tijekom 24 mjeseca,
05:54
the first two years that we really focused on,
119
339000
3000
prve dvije godine, na koje smo se najviše usredotočili,
05:57
this is a map of every word he learned in chronological order.
120
342000
4000
ovo je prikaz svih riječi koje je naučio, kronološkim redom.
06:01
And because we have full transcripts,
121
346000
3000
A budući da imamo potpune transkripte,
06:04
we've identified each of the 503 words
122
349000
2000
identificirali smo svaku od 503 riječi
06:06
that he learned to produce by his second birthday.
123
351000
2000
koje je naučio do drugog rođendana.
06:08
He was an early talker.
124
353000
2000
Rano je progovorio.
06:10
And so we started to analyze why.
125
355000
3000
I počeli smo analizirati zašto.
06:13
Why were certain words born before others?
126
358000
3000
Zašto su se neke riječi "rodile" prije drugih?
06:16
This is one of the first results
127
361000
2000
Ovo je jedan od prvih rezultata koje smo dobili
06:18
that came out of our study a little over a year ago
128
363000
2000
iz našeg istraživanja prije nešto više od godinu dana
06:20
that really surprised us.
129
365000
2000
i koji nas je zaista iznenadio.
06:22
The way to interpret this apparently simple graph
130
367000
3000
Na ovom naizgled jednostavnom grafu
06:25
is, on the vertical is an indication
131
370000
2000
na okomitoj osi prikazana je kompleksnost
06:27
of how complex caregiver utterances are
132
372000
3000
rečenica koje su izgovarali odrasli,
06:30
based on the length of utterances.
133
375000
2000
a temeljeno na dužini rečenica.
06:32
And the [horizontal] axis is time.
134
377000
3000
A na horizontalnoj osi je vrijeme.
06:35
And all of the data,
135
380000
2000
I sve smo podatke
06:37
we aligned based on the following idea:
136
382000
3000
poredali prema sljedećem principu:
06:40
Every time my son would learn a word,
137
385000
3000
svaki put kad bi moj sin naučio riječ,
06:43
we would trace back and look at all of the language he heard
138
388000
3000
pratili bismo podatke unatrag i gledali sav govor koji je čuo
06:46
that contained that word.
139
391000
2000
i koji je sadržavao tu riječ
06:48
And we would plot the relative length of the utterances.
140
393000
4000
i ucrtali bismo relativnu duljinu tih rečenica.
06:52
And what we found was this curious phenomena,
141
397000
3000
Tako smo otkrili ovaj zanimljiv fenomen:
06:55
that caregiver speech would systematically dip to a minimum,
142
400000
3000
govor odrasle osobe sustavno se svodio na minimum,
06:58
making language as simple as possible,
143
403000
3000
čineći govor što jednostavnijim,
07:01
and then slowly ascend back up in complexity.
144
406000
3000
a zatim bi ponovno postupno postao sve složeniji.
07:04
And the amazing thing was
145
409000
2000
Ono što nas je zadivilo
07:06
that bounce, that dip,
146
411000
2000
bio je taj pad, to pojednostavljenje,
07:08
lined up almost precisely
147
413000
2000
koje se gotovo potpuno poklapa
07:10
with when each word was born --
148
415000
2000
s trenutkom "rođenja" svake riječi --
07:12
word after word, systematically.
149
417000
2000
i tako sustavno, za svaku riječ.
07:14
So it appears that all three primary caregivers --
150
419000
2000
Tako izgleda da su sva tri primarna skrbnika --
07:16
myself, my wife and our nanny --
151
421000
3000
ja, moja supruga i naša dadilja --
07:19
were systematically and, I would think, subconsciously
152
424000
3000
sustavno i, rekao bih, podsvjesno,
07:22
restructuring our language
153
427000
2000
rekonstruirala jezik
07:24
to meet him at the birth of a word
154
429000
3000
kako bi se susreli s njim prilikom rođenja riječi
07:27
and bring him gently into more complex language.
155
432000
4000
i nježno ga uveli u složeniji govor.
07:31
And the implications of this -- there are many,
156
436000
2000
Pretpostavka je sljedeća -- ima ih više,
07:33
but one I just want to point out,
157
438000
2000
ali ovu želim istaknuti:
07:35
is that there must be amazing feedback loops.
158
440000
3000
postoje zadivljujuće petlje povratnih informacija.
07:38
Of course, my son is learning
159
443000
2000
Naravno, moj sin uči
07:40
from his linguistic environment,
160
445000
2000
od svoje jezične okoline,
07:42
but the environment is learning from him.
161
447000
3000
ali i okolina uči od njega.
07:45
That environment, people, are in these tight feedback loops
162
450000
3000
Ta okolina, ljudi, dio su petlji povratnih informacija
07:48
and creating a kind of scaffolding
163
453000
2000
i grade neku vrstu skele
07:50
that has not been noticed until now.
164
455000
3000
koja do sad još nije bila uočena.
07:54
But that's looking at the speech context.
165
459000
2000
Ali to se odnosi samo na govorni kontekst.
07:56
What about the visual context?
166
461000
2000
Što je s vizualnim kontekstom?
07:58
We're not looking at --
167
463000
2000
Sad gledamo --
08:00
think of this as a dollhouse cutaway of our house.
168
465000
2000
zamislite ovo kao presjek naše kuće, poput kućice za lutke.
08:02
We've taken those circular fish-eye lens cameras,
169
467000
3000
Uzeli smo kružne leće u obliku ribljeg oka
08:05
and we've done some optical correction,
170
470000
2000
i napravili smo neke optičke korekcije,
08:07
and then we can bring it into three-dimensional life.
171
472000
4000
nakon čega smo mogli stvoriti stvaran trodimenzionalni prikaz.
08:11
So welcome to my home.
172
476000
2000
Pa, dobrodošli u moj dom.
08:13
This is a moment,
173
478000
2000
Ovo je trenutak,
08:15
one moment captured across multiple cameras.
174
480000
3000
jedan trenutak zabilježen na više kamera.
08:18
The reason we did this is to create the ultimate memory machine,
175
483000
3000
Razlog zbog kojeg smo to učinili jest da bismo stvorili najbolji memorijski stroj,
08:21
where you can go back and interactively fly around
176
486000
3000
u kojem se možete vraćati i interaktivno se kretati
08:24
and then breathe video-life into this system.
177
489000
3000
i zatim udahnuti život ovom sustavu.
08:27
What I'm going to do
178
492000
2000
Sada ću vam prikazati
08:29
is give you an accelerated view of 30 minutes,
179
494000
3000
ubrzan pregled 30 minuta
08:32
again, of just life in the living room.
180
497000
2000
svakodnevnog života u dnevnom boravku.
08:34
That's me and my son on the floor.
181
499000
3000
To smo ja i moj sin na podu.
08:37
And there's video analytics
182
502000
2000
A ovo su videoanalize
08:39
that are tracking our movements.
183
504000
2000
koje prate naše pokrete.
08:41
My son is leaving red ink. I am leaving green ink.
184
506000
3000
Moj sin ostavlja crveni trag, a ja ostavljam zeleni trag.
08:44
We're now on the couch,
185
509000
2000
Sad smo na kauču,
08:46
looking out through the window at cars passing by.
186
511000
3000
gledamo kroz prozor automobile koji prolaze.
08:49
And finally, my son playing in a walking toy by himself.
187
514000
3000
I na kraju, moj sin se sam igra u hodalici.
08:52
Now we freeze the action, 30 minutes,
188
517000
3000
Sad zamrznemo snimku, 30 minuta,
08:55
we turn time into the vertical axis,
189
520000
2000
prikažemo vrijeme na okomitoj osi,
08:57
and we open up for a view
190
522000
2000
i dobivamo pregled
08:59
of these interaction traces we've just left behind.
191
524000
3000
interaktivnih tragova koje smo ostavljali.
09:02
And we see these amazing structures --
192
527000
3000
Vidimo te zadivljujuće strukture --
09:05
these little knots of two colors of thread
193
530000
3000
te male čvorove dviju boja,
09:08
we call "social hot spots."
194
533000
2000
koje nazivamo društvenim žarištima.
09:10
The spiral thread
195
535000
2000
Spiralni trag
09:12
we call a "solo hot spot."
196
537000
2000
nazivamo samostalnim žarištem.
09:14
And we think that these affect the way language is learned.
197
539000
3000
Mislimo da to utječe na način na koji se uči jezik.
09:17
What we'd like to do
198
542000
2000
Htjeli bismo
09:19
is start understanding
199
544000
2000
početi shvaćati
09:21
the interaction between these patterns
200
546000
2000
interakciju između ovih uzoraka
09:23
and the language that my son is exposed to
201
548000
2000
i govora kojem je moj sin izložen
09:25
to see if we can predict
202
550000
2000
kako bismo pokušali predvidjeti
09:27
how the structure of when words are heard
203
552000
2000
kako struktura vremena u kojem se riječi čuju
09:29
affects when they're learned --
204
554000
2000
utječe na vrijeme u kojem se nauče --
09:31
so in other words, the relationship
205
556000
2000
odnosno, drugim riječima, vezu
09:33
between words and what they're about in the world.
206
558000
4000
između riječi i onoga što one znače u svijetu.
09:37
So here's how we're approaching this.
207
562000
2000
Evo kako mi pristupamo tome.
09:39
In this video,
208
564000
2000
U ovom videu,
09:41
again, my son is being traced out.
209
566000
2000
ponovno, pratimo mog sina.
09:43
He's leaving red ink behind.
210
568000
2000
On ostavlja crveni trag.
09:45
And there's our nanny by the door.
211
570000
2000
Ovo je naša dadilja, pored vrata.
09:47
(Video) Nanny: You want water? (Baby: Aaaa.)
212
572000
3000
(Video) Dadilja: Hoćeš vode? (Dijete: Aaaa.)
09:50
Nanny: All right. (Baby: Aaaa.)
213
575000
3000
Dadilja: U redu. (Dijete: Aaaa.)
09:53
DR: She offers water,
214
578000
2000
DR: Ona nudi vodu,
09:55
and off go the two worms
215
580000
2000
i dva crva odlaze
09:57
over to the kitchen to get water.
216
582000
2000
u kuhinju po vodu.
09:59
And what we've done is use the word "water"
217
584000
2000
Upotrijebili smo riječ "voda"
10:01
to tag that moment, that bit of activity.
218
586000
2000
da bismo označili taj trenutak, taj dio aktivnosti.
10:03
And now we take the power of data
219
588000
2000
A sada koristimo moć podataka
10:05
and take every time my son
220
590000
3000
i dobivamo svaki trenutak kad je moj sin
10:08
ever heard the word water
221
593000
2000
čuo riječ voda
10:10
and the context he saw it in,
222
595000
2000
i kontekst u kojem ju je vidio.
10:12
and we use it to penetrate through the video
223
597000
3000
Koristimo to kako bismo prošli kroz video
10:15
and find every activity trace
224
600000
3000
i pronašli svaki trag aktivnosti
10:18
that co-occurred with an instance of water.
225
603000
3000
koja se pojavila u istom trenutku kad i voda.
10:21
And what this data leaves in its wake
226
606000
2000
Ono što je nastalo kao posljedica tih podataka
10:23
is a landscape.
227
608000
2000
jest krajolik.
10:25
We call these wordscapes.
228
610000
2000
Ovo nazivamo "rječolikom".
10:27
This is the wordscape for the word water,
229
612000
2000
Ovo je rječolik za riječ voda,
10:29
and you can see most of the action is in the kitchen.
230
614000
2000
i možete vidjeti da je većina aktivnosti u kuhinji.♫
10:31
That's where those big peaks are over to the left.
231
616000
3000
Tu se nalaze ovi veliki vrhovi s lijeve strane.
10:34
And just for contrast, we can do this with any word.
232
619000
3000
Samo za usporedbu, tako možemo prikazati svaku riječ.
10:37
We can take the word "bye"
233
622000
2000
Možemo uzeti riječ "đenja",
10:39
as in "good bye."
234
624000
2000
kao u "do viđenja".
10:41
And we're now zoomed in over the entrance to the house.
235
626000
2000
I sad smo prebačeni pred ulaz u kuću.
10:43
And we look, and we find, as you would expect,
236
628000
3000
Otkrivamo, kao što bi se i očekivalo,
10:46
a contrast in the landscape
237
631000
2000
da postoji razlika u krajoliku,
10:48
where the word "bye" occurs much more in a structured way.
238
633000
3000
pri čemu se riječ "đenja" pojavljuje mnogo pravilnije.
10:51
So we're using these structures
239
636000
2000
Dakle, koristimo te strukture
10:53
to start predicting
240
638000
2000
kako bismo počeli predviđati
10:55
the order of language acquisition,
241
640000
3000
redoslijed usvajanja jezika,
10:58
and that's ongoing work now.
242
643000
2000
i još uvijek radimo na tome.
11:00
In my lab, which we're peering into now, at MIT --
243
645000
3000
U mojem laboratoriju na MIT-u, koji sada gledamo --
11:03
this is at the media lab.
244
648000
2000
ovo je u medijskom laboratoriju.
11:05
This has become my favorite way
245
650000
2000
Ovo je postao moj omiljeni način
11:07
of videographing just about any space.
246
652000
2000
grafičkog videoprikaza svakog prostora.
11:09
Three of the key people in this project,
247
654000
2000
Troje ključnih ljudi u ovom projektu,
11:11
Philip DeCamp, Rony Kubat and Brandon Roy are pictured here.
248
656000
3000
Philip DeCamp, Rony Kubat i Brandon Roy prikazani su ovdje.
11:14
Philip has been a close collaborator
249
659000
2000
Philip mi je bio bliski suradnik
11:16
on all the visualizations you're seeing.
250
661000
2000
pri izradi vizualizacija koje gledate.
11:18
And Michael Fleischman
251
663000
3000
A Michael Fleischman
11:21
was another Ph.D. student in my lab
252
666000
2000
također je bio student na doktoratu u mojem laboratoriju
11:23
who worked with me on this home video analysis,
253
668000
3000
koji je sa mnom radio na ovoj analizi kućnog videa,
11:26
and he made the following observation:
254
671000
3000
i on je primijetio sljedeće:
11:29
that "just the way that we're analyzing
255
674000
2000
"način na koji analiziramo
11:31
how language connects to events
256
676000
3000
kako je jezik povezan s događajima
11:34
which provide common ground for language,
257
679000
2000
koji predstavljaju zajednički temelj jezika,
11:36
that same idea we can take out of your home, Deb,
258
681000
4000
tu istu ideju možemo primijeniti i izvan tvoje kuće, Deb,
11:40
and we can apply it to the world of public media."
259
685000
3000
i možemo je primijeniti na svijet javnih medija."
11:43
And so our effort took an unexpected turn.
260
688000
3000
I tako je naš rad krenuo u neočekivanom smjeru.
11:46
Think of mass media
261
691000
2000
Zamislite kako masovni mediji
11:48
as providing common ground
262
693000
2000
tvore zajednički temelj,
11:50
and you have the recipe
263
695000
2000
a vi imate recept
11:52
for taking this idea to a whole new place.
264
697000
3000
kako tu ideju prenijeti na neku višu razinu.
11:55
We've started analyzing television content
265
700000
3000
Počeli smo analizirati televizijski sadržaj
11:58
using the same principles --
266
703000
2000
koristeći iste principe --
12:00
analyzing event structure of a TV signal --
267
705000
3000
analizu strukture događaja TV signala --
12:03
episodes of shows,
268
708000
2000
epizode emisija,
12:05
commercials,
269
710000
2000
reklame,
12:07
all of the components that make up the event structure.
270
712000
3000
sve sastavnice koje čine strukturu događaja.
12:10
And we're now, with satellite dishes, pulling and analyzing
271
715000
3000
Sada, pomoću satelitskih tanjura, skupljamo i analiziramo
12:13
a good part of all the TV being watched in the United States.
272
718000
3000
dobar dio cjelokupnog TV programa koji se gleda u SAD-u.
12:16
And you don't have to now go and instrument living rooms with microphones
273
721000
3000
I sad ne trebate opremati dnevne boravke mikrofonima
12:19
to get people's conversations,
274
724000
2000
da biste čuli razgovore drugih ljudi,
12:21
you just tune into publicly available social media feeds.
275
726000
3000
samo se uključite u javno dostupne društvene medije.
12:24
So we're pulling in
276
729000
2000
Tako da prikupljamo
12:26
about three billion comments a month,
277
731000
2000
oko tri milijuna komentara mjesečno.
12:28
and then the magic happens.
278
733000
2000
I tada se događa čarolija.
12:30
You have the event structure,
279
735000
2000
Imate strukturu događaja,
12:32
the common ground that the words are about,
280
737000
2000
zajedničku točku svih riječi
12:34
coming out of the television feeds;
281
739000
3000
koje dolaze iz televizijskih emisija;
12:37
you've got the conversations
282
742000
2000
Imate razgovore
12:39
that are about those topics;
283
744000
2000
koji su o tim temama;
12:41
and through semantic analysis --
284
746000
3000
i kroz semantičku analizu --
12:44
and this is actually real data you're looking at
285
749000
2000
sad gledate prave, stvarne podatke
12:46
from our data processing --
286
751000
2000
iz naše obrade podataka --
12:48
each yellow line is showing a link being made
287
753000
3000
svaka žuta linija označuje vezu koja se stvara
12:51
between a comment in the wild
288
756000
3000
između nevezanog komentara
12:54
and a piece of event structure coming out of the television signal.
289
759000
3000
i dijela strukture događaja koja dolazi iz televizijskog signala.
12:57
And the same idea now
290
762000
2000
I ta se ista ideja
12:59
can be built up.
291
764000
2000
sad može nadograditi.
13:01
And we get this wordscape,
292
766000
2000
Tako da dobivamo ovaj rječolik,
13:03
except now words are not assembled in my living room.
293
768000
3000
osim što riječi sada nisu složene u mojem dnevnom boravku.
13:06
Instead, the context, the common ground activities,
294
771000
4000
Umjesto toga, kontekst, zajedničke aktivnosti,
13:10
are the content on television that's driving the conversations.
295
775000
3000
sadržaj su na televiziji koji stvara razgovore.
13:13
And what we're seeing here, these skyscrapers now,
296
778000
3000
Ovo što ovdje vidimo, ovi neboderi,
13:16
are commentary
297
781000
2000
jesu komentari
13:18
that are linked to content on television.
298
783000
2000
koji su povezani s televizijskim sadržajem.
13:20
Same concept,
299
785000
2000
Isti koncept,
13:22
but looking at communication dynamics
300
787000
2000
osim što se gleda dinamika komunikacije
13:24
in a very different sphere.
301
789000
2000
u vrlo različitoj sferi.
13:26
And so fundamentally, rather than, for example,
302
791000
2000
Tako da, u biti, umjesto da, primjerice
13:28
measuring content based on how many people are watching,
303
793000
3000
mjerimo sadržaj na temelju broja gledatelja,
13:31
this gives us the basic data
304
796000
2000
ovo nam daje osnovne podatke
13:33
for looking at engagement properties of content.
305
798000
3000
za razumijevanje poticajnih svojstava sadržaja.
13:36
And just like we can look at feedback cycles
306
801000
3000
I baš kao što možemo gledati povratne cikluse
13:39
and dynamics in a family,
307
804000
3000
i dinamiku u obitelji,
13:42
we can now open up the same concepts
308
807000
3000
sad možemo proširiti iste koncepte
13:45
and look at much larger groups of people.
309
810000
3000
i obrađivati mnogo veće grupe ljudi.
13:48
This is a subset of data from our database --
310
813000
3000
Ovo je podskup podataka iz naše baze --
13:51
just 50,000 out of several million --
311
816000
3000
samo 50.000 od nekoliko milijuna --
13:54
and the social graph that connects them
312
819000
2000
i društveni graf koji ih povezuje
13:56
through publicly available sources.
313
821000
3000
kroz javno dostupne izvore.
13:59
And if you put them on one plain,
314
824000
2000
Ako ih stavite u jednu ravninu,
14:01
a second plain is where the content lives.
315
826000
3000
na drugoj ravnini nalazi se sadržaj.
14:04
So we have the programs
316
829000
3000
Tako imamo programe
14:07
and the sporting events
317
832000
2000
i sportske događaje
14:09
and the commercials,
318
834000
2000
i reklame
14:11
and all of the link structures that tie them together
319
836000
2000
a sve strukture povezivanja koje ih međusobno vežu
14:13
make a content graph.
320
838000
2000
zajedno čine graf sadržaja.
14:15
And then the important third dimension.
321
840000
4000
I sada, važna treća dimenzija.
14:19
Each of the links that you're seeing rendered here
322
844000
2000
Svaka od poveznica koje ovdje vidite prikazane
14:21
is an actual connection made
323
846000
2000
stvarna je veza koja je nastala
14:23
between something someone said
324
848000
3000
između nečega što je netko rekao
14:26
and a piece of content.
325
851000
2000
i dijela sadržaja.
14:28
And there are, again, now tens of millions of these links
326
853000
3000
A sada ponovno vidimo desetke milijuna tih poveznica
14:31
that give us the connective tissue of social graphs
327
856000
3000
koje sačinjavaju vezivno tkivo društvenih grafova
14:34
and how they relate to content.
328
859000
3000
i načina na koji su povezani sa sadržajem.
14:37
And we can now start to probe the structure
329
862000
2000
Sad možemo početi ispitivati strukturu
14:39
in interesting ways.
330
864000
2000
na zanimljive načine.
14:41
So if we, for example, trace the path
331
866000
3000
Ako, primjerice, pratimo put
14:44
of one piece of content
332
869000
2000
jednog dijela sadržaja
14:46
that drives someone to comment on it,
333
871000
2000
koji potiče nekoga da ga komentira,
14:48
and then we follow where that comment goes,
334
873000
3000
i zatim slijedimo put kojim taj komentar ide,
14:51
and then look at the entire social graph that becomes activated
335
876000
3000
i zatim pogledamo cijeli društveni graf koji se aktivira
14:54
and then trace back to see the relationship
336
879000
3000
i zatim se vratimo kako bismo vidjeli odnos
14:57
between that social graph and content,
337
882000
2000
između tog društvenog grafa i sadržaja,
14:59
a very interesting structure becomes visible.
338
884000
2000
pojavljuje se vrlo zanimljiva struktura.
15:01
We call this a co-viewing clique,
339
886000
2000
To zovemo ekipom za zajedničko promatranje,
15:03
a virtual living room if you will.
340
888000
3000
virtualni dnevni boravak.
15:06
And there are fascinating dynamics at play.
341
891000
2000
Odvija se fascinantna dinamika.
15:08
It's not one way.
342
893000
2000
I to nije jednosmjerno.
15:10
A piece of content, an event, causes someone to talk.
343
895000
3000
Dio sadržaja, neki događaj, potakne nekoga da govori.
15:13
They talk to other people.
344
898000
2000
Ta osoba razgovara s drugim ljudima.
15:15
That drives tune-in behavior back into mass media,
345
900000
3000
To potiče ponovno uključenje u masovne medije
15:18
and you have these cycles
346
903000
2000
tako da postoje ciklusi
15:20
that drive the overall behavior.
347
905000
2000
koji potiču cjelokupno ponašanje.
15:22
Another example -- very different --
348
907000
2000
Drugi primjer -- vrlo različit --
15:24
another actual person in our database --
349
909000
3000
druga stvarna osoba u našoj bazi --
15:27
and we're finding at least hundreds, if not thousands, of these.
350
912000
3000
a pronalazimo barem stotine, ako ne i tisuće takvih.
15:30
We've given this person a name.
351
915000
2000
Dali smo toj osobi ime.
15:32
This is a pro-amateur, or pro-am media critic
352
917000
3000
On je pro-amater, pro-am, medijski kritičar
15:35
who has this high fan-out rate.
353
920000
3000
koji dopire do velikog dijela javnosti.
15:38
So a lot of people are following this person -- very influential --
354
923000
3000
Tako da mnogi ljudi slijede tu osobu -- vrlo utjecajnu --
15:41
and they have a propensity to talk about what's on TV.
355
926000
2000
i imaju sklonost komentiranju sadržaja na TV-u.
15:43
So this person is a key link
356
928000
3000
Ta je osoba ključna poveznica
15:46
in connecting mass media and social media together.
357
931000
3000
između masovnih medija i društvenih medija.
15:49
One last example from this data:
358
934000
3000
Posljednji primjer iz ovih podataka:
15:52
Sometimes it's actually a piece of content that is special.
359
937000
3000
Ponekad je poseban upravo dio sadržaja.
15:55
So if we go and look at this piece of content,
360
940000
4000
Pa ako promotrimo taj sadržaj,
15:59
President Obama's State of the Union address
361
944000
3000
govor predsjednika Obame o stanju Unije
16:02
from just a few weeks ago,
362
947000
2000
od prije samo nekoliko tjedana,
16:04
and look at what we find in this same data set,
363
949000
3000
pogledajte što pronalazimo u tom istom skupu podataka,
16:07
at the same scale,
364
952000
3000
na istoj razini,
16:10
the engagement properties of this piece of content
365
955000
2000
poticajna svojstva tog dijela sadržaja
16:12
are truly remarkable.
366
957000
2000
zaista su izuzetne.
16:14
A nation exploding in conversation
367
959000
2000
Narod koji odjednom počinje komentirati
16:16
in real time
368
961000
2000
u stvarnom vremenu
16:18
in response to what's on the broadcast.
369
963000
3000
potaknut onime što se emitira.
16:21
And of course, through all of these lines
370
966000
2000
I naravno, kroz sve te linije
16:23
are flowing unstructured language.
371
968000
2000
teče nestrukturiran jezik.
16:25
We can X-ray
372
970000
2000
Možemo napraviti rentgensku snimku
16:27
and get a real-time pulse of a nation,
373
972000
2000
i dobiti stvarni puls naroda,
16:29
real-time sense
374
974000
2000
stvarne podatke
16:31
of the social reactions in the different circuits in the social graph
375
976000
3000
o društvenim reakcijama u različitim krugovima društvenog grafa
16:34
being activated by content.
376
979000
3000
koje aktivira sadržaj.
16:37
So, to summarize, the idea is this:
377
982000
3000
Dakle, da sažmem, ideja je sljedeća:
16:40
As our world becomes increasingly instrumented
378
985000
3000
Kako nam je svijet sve više automatiziran
16:43
and we have the capabilities
379
988000
2000
i imamo mogućnosti
16:45
to collect and connect the dots
380
990000
2000
prikupljati i povezivati točkice
16:47
between what people are saying
381
992000
2000
između onoga što ljudi govore
16:49
and the context they're saying it in,
382
994000
2000
i konteksta u kojem to govore,
16:51
what's emerging is an ability
383
996000
2000
pojavljuje se i mogućnost
16:53
to see new social structures and dynamics
384
998000
3000
da vidimo nove društvene strukture i dinamiku
16:56
that have previously not been seen.
385
1001000
2000
koje prije nismo mogli vidjeti.
16:58
It's like building a microscope or telescope
386
1003000
2000
To je kao kad sklapate mikroskop ili teleskop
17:00
and revealing new structures
387
1005000
2000
i otkrivate nove strukture
17:02
about our own behavior around communication.
388
1007000
3000
našeg vlastitog ponašanja tijekom komunikacije.
17:05
And I think the implications here are profound,
389
1010000
3000
I mislim da ovo ima duboke implikacije,
17:08
whether it's for science,
390
1013000
2000
bilo za znanost,
17:10
for commerce, for government,
391
1015000
2000
trgovinu, vladu,
17:12
or perhaps most of all,
392
1017000
2000
ili možda, najviše od svega,
17:14
for us as individuals.
393
1019000
3000
za nas kao pojedince.
17:17
And so just to return to my son,
394
1022000
3000
I samo da se vratim na svog sina,
17:20
when I was preparing this talk, he was looking over my shoulder,
395
1025000
3000
dok sam pripremao ovo predavanje, gledao mi je preko ramena
17:23
and I showed him the clips I was going to show to you today,
396
1028000
2000
i pokazao sam mu videozapise koje sam mislio pokazati vama danas
17:25
and I asked him for permission -- granted.
397
1030000
3000
i tražio sam ga dopuštenje -- i dobio ga.
17:28
And then I went on to reflect,
398
1033000
2000
Nakon toga sam razmišljao,
17:30
"Isn't it amazing,
399
1035000
3000
"Nije li to fantastično,
17:33
this entire database, all these recordings,
400
1038000
3000
cijelu ovu bazu, sve snimke,
17:36
I'm going to hand off to you and to your sister" --
401
1041000
2000
predat ću tebi i tvojoj sestri",
17:38
who arrived two years later --
402
1043000
3000
koju smo dobili dvije godine kasnije.
17:41
"and you guys are going to be able to go back and re-experience moments
403
1046000
3000
"A vi ćete se moći vratiti i ponovno proživjeti trenutke
17:44
that you could never, with your biological memory,
404
1049000
3000
kojih se inače, sa svojim biološkim pamćenjem,
17:47
possibly remember the way you can now?"
405
1052000
2000
nikako ne biste mogli sjećati kao što to možete sada."
17:49
And he was quiet for a moment.
406
1054000
2000
On je zašutio na trenutak,
17:51
And I thought, "What am I thinking?
407
1056000
2000
pa sam pomislio, "Što je meni?
17:53
He's five years old. He's not going to understand this."
408
1058000
2000
Ima pet godina. Neće razumjeti ništa od ovoga."
17:55
And just as I was having that thought, he looked up at me and said,
409
1060000
3000
I upravo u trenutku kad sam to pomislio, pogledao me i rekao,
17:58
"So that when I grow up,
410
1063000
2000
"Znači, kad narastem,
18:00
I can show this to my kids?"
411
1065000
2000
mogu ovo pokazati svojoj djeci?"
18:02
And I thought, "Wow, this is powerful stuff."
412
1067000
3000
I pomislio sam "Opa, ovo je moćna stvar."
18:05
So I want to leave you
413
1070000
2000
Za kraj vam želim pokazati
18:07
with one last memorable moment
414
1072000
2000
posljednji trenutak za pamćenje
18:09
from our family.
415
1074000
3000
iz naše obitelji.
18:12
This is the first time our son
416
1077000
2000
Ovo je prvi put da je naš sin
18:14
took more than two steps at once --
417
1079000
2000
napravio više od dva koraka odjednom --
18:16
captured on film.
418
1081000
2000
zabilježeno na filmu.
18:18
And I really want you to focus on something
419
1083000
3000
Želim da obratite pažnju na nešto
18:21
as I take you through.
420
1086000
2000
dok vas vodim kroz snimku.
18:23
It's a cluttered environment; it's natural life.
421
1088000
2000
To je natrpana okolina, svakodnevni život.
18:25
My mother's in the kitchen, cooking,
422
1090000
2000
Moja majka je u kuhinji, kuha,
18:27
and, of all places, in the hallway,
423
1092000
2000
i baš u hodniku
18:29
I realize he's about to do it, about to take more than two steps.
424
1094000
3000
shvaćam da će to učiniti, napraviti više od dva koraka.
18:32
And so you hear me encouraging him,
425
1097000
2000
Čujete me kako ga ohrabrujem,
18:34
realizing what's happening,
426
1099000
2000
shvaćam što se događa,
18:36
and then the magic happens.
427
1101000
2000
i tad nastupa čarolija.
18:38
Listen very carefully.
428
1103000
2000
Slušajte pažljivo.
18:40
About three steps in,
429
1105000
2000
Nakon oko tri koraka,
18:42
he realizes something magic is happening,
430
1107000
2000
on shvaća da se događa nešto čarobno.
18:44
and the most amazing feedback loop of all kicks in,
431
1109000
3000
I uključuje se zadivljujuća petlja povratnih informacija,
18:47
and he takes a breath in,
432
1112000
2000
on udahne,
18:49
and he whispers "wow"
433
1114000
2000
i prošapće "opa"
18:51
and instinctively I echo back the same.
434
1116000
4000
a ja instiktivno odgovorim isto to.
18:56
And so let's fly back in time
435
1121000
3000
Vratimo se nazad u vremenu
18:59
to that memorable moment.
436
1124000
2000
do tog trenutka za pamćenje.
19:05
(Video) DR: Hey.
437
1130000
2000
(Video) DR: Hej.
19:07
Come here.
438
1132000
2000
Dođi ovamo.
19:09
Can you do it?
439
1134000
3000
Možeš li?
19:13
Oh, boy.
440
1138000
2000
Ajme meni.
19:15
Can you do it?
441
1140000
3000
Možeš li?
19:18
Baby: Yeah.
442
1143000
2000
Dijete: Da.
19:20
DR: Ma, he's walking.
443
1145000
3000
DR: Mama, on hoda.
19:24
(Laughter)
444
1149000
2000
(Smijeh)
19:26
(Applause)
445
1151000
2000
(Pljesak)
19:28
DR: Thank you.
446
1153000
2000
DR: Hvala.
19:30
(Applause)
447
1155000
15000
(Pljesak)
ABOUT THE SPEAKER
Deb Roy - Cognitive scientistDeb Roy studies how children learn language, and designs machines that learn to communicate in human-like ways. On sabbatical from MIT Media Lab, he's working with the AI company Bluefin Labs.
Why you should listen
Deb Roy directs the Cognitive Machines group at the MIT Media Lab, where he studies how children learn language, and designs machines that learn to communicate in human-like ways. To enable this work, he has pioneered new data-driven methods for analyzing and modeling human linguistic and social behavior. He has authored numerous scientific papers on artificial intelligence, cognitive modeling, human-machine interaction, data mining, and information visualization.
Deb Roy was the co-founder and serves as CEO of Bluefin Labs, a venture-backed technology company. Built upon deep machine learning principles developed in his research over the past 15 years, Bluefin has created a technology platform that analyzes social media commentary to measure real-time audience response to TV ads and shows.
Follow Deb Roy on Twitter>
Roy adds some relevant papers:
Deb Roy. (2009). New Horizons in the Study of Child Language Acquisition. Proceedings of Interspeech 2009. Brighton, England. bit.ly/fSP4Qh
Brandon C. Roy, Michael C. Frank and Deb Roy. (2009). Exploring word learning in a high-density longitudinal corpus. Proceedings of the 31st Annual Meeting of the Cognitive Science Society. Amsterdam, Netherlands. bit.ly/e1qxej
Plenty more papers on our research including technology and methodology can be found here, together with other research from my lab at MIT: bit.ly/h3paSQ
The work that I mentioned on relationships between television content and the social graph is being done at Bluefin Labs (www.bluefinlabs.com). Details of this work have not been published. The social structures we are finding (and that I highlighted in my TED talk) are indeed new. The social media communication channels that are leading to their formation did not even exist a few years ago, and Bluefin's technology platform for discovering these kinds of structures is the first of its kind. We'll certainly have more to say about all this as we continue to dig into this fascinating new kind of data, and as new social structures continue to evolve!
Deb Roy | Speaker | TED.com