WEBVTT

1
00:00:00.000 --> 00:00:06.660
So in the last video we started to look at data visualization and data science applications.

2
00:00:06.660 --> 00:00:19.230
In this video we look at some some of the newer data science applications that come along in the last couple of years hopefully or will continue to develop develop in the next few years.

3
00:00:19.230 --> 00:00:22.760
So first I want to talk about is chatterbox.

4
00:00:22.760 --> 00:00:26.040
So I'm torn about a chart but we're thinking about a computer application.

5
00:00:26.040 --> 00:00:29.940
You can simulate conversation with a normal person.

6
00:00:29.940 --> 00:00:34.560
So it might be through text like in an actual chat.

7
00:00:34.560 --> 00:00:37.080
Or it could be through speech.

8
00:00:37.080 --> 00:00:40.380
And even one would would be a chapel.

9
00:00:40.380 --> 00:00:56.160
So these will rely on natural language processing and being able to train and train the application to be able to recognize whether that speech or sentence structure or various characteristics of chance.

10
00:00:56.160 --> 00:00:58.800
And it's still very very difficult.

11
00:00:58.800 --> 00:01:07.680
There is a simpler sort of model which can be done which is more rule or heuristic based which called for retrieval based models for chat bots where.

12
00:01:07.680 --> 00:01:16.320
Given a certain set of characteristics the application will take an appropriate response send that back to the user.

13
00:01:16.320 --> 00:01:23.940
And some of the more advanced models you got generative based models which are able to generate new responses from scratch.

14
00:01:23.940 --> 00:01:28.620
And depending on what the previous query or.

15
00:01:28.620 --> 00:01:31.470
Comment was from the person who used it.

16
00:01:31.470 --> 00:01:35.310
So there's still a lot of work to be done in this area.

17
00:01:35.310 --> 00:01:50.710
And if you stop trying to communicate with a chat bot outside of a specific domain then you'll realize the limitations very quickly but it's promising technology for the future and it has got backing from big companies such as Facebook.

18
00:01:50.710 --> 00:02:07.360
Facebook recently launched and chat bots for Messenger where you can communicate with you users to answer questions or to help them to pick products or even to help actually buy products themselves.

19
00:02:07.360 --> 00:02:23.760
Messaging is often used as preferred means of communication these days and having the additional assistance in order to be able to purchase products and makes the whole experience more frictionless for users means more likely that people will actually buying products.

20
00:02:23.760 --> 00:02:26.850
So speaking about the frictionless.

21
00:02:26.850 --> 00:02:37.930
Frictionless purchases Amazon go as a shop where there is no actual stuff you can simply go into the store pick up the items that you want and then leave.

22
00:02:37.930 --> 00:02:40.780
And it will be automatically charged on your account.

23
00:02:40.780 --> 00:03:02.560
So aside from from these sorts of applications looking at the direct application to being able to market and in order to improve the conversion rates or or to provide customized offers Internet of Things is an example which over the last few years have become incredibly popular.

24
00:03:02.560 --> 00:03:08.450
So this is the idea that it's not just computers which are connected to the Internet.

25
00:03:08.450 --> 00:03:20.860
You've also got mobile phones you've got tablets and an increasingly large amount of sensors around the place which were able to detect movement or temperature or or anything.

26
00:03:20.860 --> 00:03:22.300
Basically.

27
00:03:22.300 --> 00:03:23.650
And.

28
00:03:23.650 --> 00:03:38.890
By using these strategically it is possible to identify extra information about customers whether they're in this in a similar area or whether there's a part of a bricks and mortar store where there are lots of customers going or nobody's going.

29
00:03:38.890 --> 00:03:39.640
And.

30
00:03:39.640 --> 00:03:46.520
If you want to use these in real time than location based on fertilising takes full advantage of.

31
00:03:46.520 --> 00:03:50.780
The devices being connected to the Internet and tracking location.

32
00:03:50.780 --> 00:04:03.800
So this is often done with smart phones or smart watches which people like to carry around and will often have agreed in order to get the application to transmit information about their current location.

33
00:04:03.800 --> 00:04:21.110
So by using this as a user approaches a particular store or a particular part of a store the application can can can push the user a prompt to say you've got a customized offer here which you might like to use based on a previous purchase.

34
00:04:21.110 --> 00:04:33.210
Maybe they like a particular kind of coffee which you can offer for free or maybe they particularly enjoyed a certain DVD and just pushing these things to the users when they're in the right place.

35
00:04:33.210 --> 00:04:36.920
And there are some still some challenges associated with this.

36
00:04:36.920 --> 00:04:40.220
A lot of the time the response data isn't granular enough.

37
00:04:40.220 --> 00:04:42.260
Or quick enough to be useful.

38
00:04:42.260 --> 00:04:47.630
But these are things which you're going to hopefully continue to improve over the next few years.

39
00:04:47.630 --> 00:04:56.360
And then a final example is that of wearable technology which I guess is a subset of the Internet of Things.

40
00:04:56.360 --> 00:04:57.500
Where.

41
00:04:57.500 --> 00:05:09.230
Clothing which people wear or other sort of devices which they attach themselves like a Fitbit will transmit information or allow some form of connectivity.

42
00:05:09.230 --> 00:05:28.730
So an example which was fairly recent was Nike developed NBA fans jerseys which had a specific code allowing the user to be able to connect their smartphone to this code in order to download or so in order to access a customized player experience.

43
00:05:28.730 --> 00:05:29.760
And.

44
00:05:29.760 --> 00:05:39.770
The continued interaction with this application gives them access to more information and allows them to continue to think about the brand and at all times.

45
00:05:39.770 --> 00:05:48.490
As an aside the jerseys for the players themselves were actually developed using senses where.

46
00:05:48.490 --> 00:05:49.600
The.

47
00:05:49.600 --> 00:05:54.370
The results coming back determine which areas.

48
00:05:54.370 --> 00:06:01.100
For example the players required a great range of movement unable to tailor the shows accordingly.

49
00:06:01.100 --> 00:06:19.020
So so that's it for our examples of data science applications and there are plenty more and we encourage you to try and think about other applications which you could use or which you know about in the digital marketing context and how you think they can be used now or in the future.