1
00:00:04,340 --> 00:00:07,540
Hey everyone. My name is Anna.

2
00:00:07,540 --> 00:00:13,040
And I'm here to present our new online course about natural language processing.

3
00:00:13,040 --> 00:00:16,235
NLP tasks are everywhere around us;

4
00:00:16,235 --> 00:00:21,515
suggest in search, automatic Gmail replies, machine translation.

5
00:00:21,515 --> 00:00:27,845
But one task which is especially popular today is chatbots or dialogue systems.

6
00:00:27,845 --> 00:00:32,445
It could be a bot that tries to hold the human-like conversation with you.

7
00:00:32,445 --> 00:00:36,715
Or it could be a bot that assists with some particular tasks,

8
00:00:36,715 --> 00:00:39,350
like comparing two vacation packages,

9
00:00:39,350 --> 00:00:42,920
or setting alarms, or answering your questions

10
00:00:42,920 --> 00:00:47,545
about credit in the bank instead of a representative in a call center.

11
00:00:47,545 --> 00:00:50,240
Do you know how this things work?

12
00:00:50,240 --> 00:00:51,835
So what is inside?

13
00:00:51,835 --> 00:00:56,465
Some deep learning magic or a rule-based approach or something else.

14
00:00:56,465 --> 00:01:00,985
Well, you will build your own chatbot in our course project.

15
00:01:00,985 --> 00:01:03,730
But before that, we will cover also

16
00:01:03,730 --> 00:01:07,285
some other interesting tasks like text classification,

17
00:01:07,285 --> 00:01:11,875
name entity recognition, duplicates detection and many more.

18
00:01:11,875 --> 00:01:14,190
Our course is rather advanced.

19
00:01:14,190 --> 00:01:16,490
We expect you're already familiar with

20
00:01:16,490 --> 00:01:19,035
some methods of machine learning and deep learning,

21
00:01:19,035 --> 00:01:22,465
and now you want to apply them to texts.

22
00:01:22,465 --> 00:01:28,210
Our goal is not to stack several black box techniques to build a dialogue monster.

23
00:01:28,210 --> 00:01:31,065
On the opposite, we will see what is inside

24
00:01:31,065 --> 00:01:34,800
each box and cover it with a good amount of mathematics.

25
00:01:34,800 --> 00:01:38,205
So does that thing look familiar to you?

26
00:01:38,205 --> 00:01:40,240
Are you scared with this letter?

27
00:01:40,240 --> 00:01:42,790
Well, you're about to see a lot of this kind of

28
00:01:42,790 --> 00:01:46,510
formulas in the upcoming weeks. So, be prepared.

29
00:01:46,510 --> 00:01:49,960
We will discuss how to represent pieces of text with

30
00:01:49,960 --> 00:01:54,955
sound vectors so that we can compute similarity between vectors.

31
00:01:54,955 --> 00:01:58,255
For example, how do you teach a machine that

32
00:01:58,255 --> 00:02:02,540
leaky faucet and tap is dripping are kind of the same thing.

33
00:02:02,540 --> 00:02:05,395
Personally, I am passionate about NLP.

34
00:02:05,395 --> 00:02:07,255
I'm finishing PhD. on that,

35
00:02:07,255 --> 00:02:10,150
teaching NLP at Yandex School of Data Analysis and

36
00:02:10,150 --> 00:02:15,185
Moscow State University and doing internships and companies from time to time.

37
00:02:15,185 --> 00:02:18,100
I have never done online courses though.

38
00:02:18,100 --> 00:02:21,500
So, I am very excited to have this experience now,

39
00:02:21,500 --> 00:02:24,570
and I will be happy to see you on board.