1
00:00:00,000 --> 00:00:04,384
[MUSIC]

2
00:00:04,384 --> 00:00:11,971
Other examples of document retrieval.

3
00:00:11,971 --> 00:00:13,140
Okay?

4
00:00:13,140 --> 00:00:15,740
So we're gonna go through a few.

5
00:00:15,740 --> 00:00:19,450
For example,
let's take the singer Taylor Swift.

6
00:00:19,450 --> 00:00:23,642
So out of all my people,

7
00:00:23,642 --> 00:00:28,463
I'm going to select the one

8
00:00:28,463 --> 00:00:33,510
whose name is Taylor Swift.

9
00:00:33,510 --> 00:00:35,470
And so when I select her I can ask,

10
00:00:36,540 --> 00:00:41,130
let's retrieve some documents that
are similar to Taylor Swift's documents.

11
00:00:41,130 --> 00:00:43,960
In other words,
who are the people, just for

12
00:00:43,960 --> 00:00:47,068
the model, who are the people
who are closest to Taylor Swift?

13
00:00:47,068 --> 00:00:51,498
So she's a female singer and
you'll see other female

14
00:00:51,498 --> 00:00:56,520
singers kind of the same
generation like Carrie Underwood,

15
00:00:56,520 --> 00:01:00,274
Alicia Keys, Jordan Sparks, Leona Lewis.

16
00:01:00,274 --> 00:01:03,162
And just a little disclaimer here,
I had to go and

17
00:01:03,162 --> 00:01:07,821
look those people up in Wikipedia because
I really didn't know who they were.

18
00:01:07,821 --> 00:01:10,860
[LAUGH] Here we go.

19
00:01:10,860 --> 00:01:13,150
Let's take another example.

20
00:01:13,150 --> 00:01:18,028
So for example,
the actress Angelina Jolie.

21
00:01:18,028 --> 00:01:24,553
So if I take, out of all my people,

22
00:01:24,553 --> 00:01:32,888
I take the one whose
name is Angelina Jolie.

23
00:01:32,888 --> 00:01:37,230
And I'm now going to
query my nearest neighbor

24
00:01:37,230 --> 00:01:41,583
model to ask who is
closest to Angelina Jolie.

25
00:01:41,583 --> 00:01:44,814
And so who is closer to Angelina Jolie?

26
00:01:44,814 --> 00:01:49,840
Before I press enter here,
take a moment to think.

27
00:01:49,840 --> 00:01:53,078
If you know anything about
this actress Angelina Jolie,

28
00:01:53,078 --> 00:01:56,044
you might guess who is
the closest person to her, and

29
00:01:56,044 --> 00:02:00,530
if I run this, you'll see that
the nearest neighbor is Brad Pitt.

30
00:02:00,530 --> 00:02:03,700
And we all know about the relationship
between Angelina Jolie and Brad Pitt.

31
00:02:03,700 --> 00:02:08,680
And then you see Julianne Moore, Billy Bob
Thornton, George Clooney, these are all

32
00:02:08,680 --> 00:02:11,730
actors and actresses that have been in
movies actually with Angelina Jolie.

33
00:02:11,730 --> 00:02:13,600
So, It's pretty cool.

34
00:02:13,600 --> 00:02:15,710
We get some interesting retrieval here.

35
00:02:15,710 --> 00:02:19,940
So let me show you one last cool,
fun example.

36
00:02:22,080 --> 00:02:27,296
So you might be familiar
with a person called

37
00:02:27,296 --> 00:02:31,865
Arnold Schwarzenegger.

38
00:02:31,865 --> 00:02:36,911
The famous Terminator.

39
00:02:36,911 --> 00:02:41,967
So, I actually have to cheat here to

40
00:02:41,967 --> 00:02:47,023
make sure I get his last name right,

41
00:02:47,023 --> 00:02:50,747
Schwarzenegger, okay.

42
00:02:50,747 --> 00:02:52,030
Hopefully I got that right.

43
00:02:52,030 --> 00:02:58,260
And the question is, who are the nearest
neighbors to Arnold Schwarzenegger?

44
00:03:01,150 --> 00:03:01,650
Now.

45
00:03:02,790 --> 00:03:06,560
You may know that Arnold
had a double career.

46
00:03:06,560 --> 00:03:08,380
So first, he start as an actor, well,

47
00:03:08,380 --> 00:03:13,310
bodybuilder, then an actor,
then a politician, and became

48
00:03:13,310 --> 00:03:18,460
the governor of the state of California
in the United States, so several lives.

49
00:03:18,460 --> 00:03:22,680
So what are the nearest neighbors for
Arnold Schwarzenegger?

50
00:03:22,680 --> 00:03:25,580
And so you'll see Arnold Schwarzenegger.

51
00:03:25,580 --> 00:03:29,872
Then you'll see Jessie Ventura,
who is like a,

52
00:03:29,872 --> 00:03:36,760
Jessie Venture is a wrestler, an actor,
so similar to Arnold Schwarzenegger.

53
00:03:36,760 --> 00:03:42,040
And then a few politicians,
including former

54
00:03:42,040 --> 00:03:46,960
governors of Oregon, and of Rhode Island
and other states in the United States.

55
00:03:46,960 --> 00:03:52,050
So now we see that the nearest neighbors
for Arnold are politicians and

56
00:03:52,050 --> 00:03:55,220
actors and body builders, people like him.

57
00:03:55,220 --> 00:03:59,600
So this was a really cool example of just
building a very simple document retrieval

58
00:03:59,600 --> 00:04:02,050
system using TFIDF and nearest neighbors.

59
00:04:02,050 --> 00:04:06,192
Just like we learned in the module,
but now taking it to practice and

60
00:04:06,192 --> 00:04:10,274
finding some pretty cool documents
from this Wikipedia data set.

61
00:04:10,274 --> 00:04:10,774
[MUSIC]