WEBVTT 00:00.420 --> 00:01.380 All right, everyone. 00:02.390 --> 00:07.820 So the next topic, which we are going to learn is named entity recognition. 00:08.720 --> 00:12.660 So for illustration purposes, I kept you three sentence. 00:13.220 --> 00:19.040 Let's say Apple is looking at my UK startup up for one billion dollars. 00:19.850 --> 00:26.690 So name recognition is something like whenever you encounter, let's say, some organization, some 00:26.840 --> 00:35.030 in some places, if you can go to the species documentation inside that there is a linguistic feature 00:35.780 --> 00:37.970 so named entity recognition. 00:38.540 --> 00:46.580 So this spacy named entity recognition can recognize, like a company location, organization products 00:46.730 --> 00:47.690 very much easily. 00:48.050 --> 00:55.340 And out of all those tax, if you have already an idea that if you are looking for Apple as an organization, 00:55.370 --> 00:59.710 you can even take Veigar, this is a for further analysis of your tax. 01:00.080 --> 01:05.570 So name entity recognition is one of the very crucial and an important step for understanding of your 01:05.810 --> 01:06.230 tax. 01:06.770 --> 01:13.670 So I woke up here three documents or I would say three sentence saying we are going to find what are 01:13.670 --> 01:15.830 the name entities that are associated with them. 01:16.340 --> 01:17.900 So let me execute Lisburn. 01:19.060 --> 01:26.350 And I'm going to Laura, a small size model in English language, and let's apply this sentence one 01:26.350 --> 01:29.480 by one and create a knock object out of. 01:32.220 --> 01:32.900 Brian LPT. 01:34.340 --> 01:39.530 Let's say Esslin and I'm going to assign you to, let's say, park one. 01:41.970 --> 01:45.330 And just like earlier, we help grab this. 01:47.010 --> 01:51.660 What I would say limited edition and tokenization saved me. 01:51.840 --> 01:59.130 This penalty is not just to go squill for getting some particular text analytics process. 01:59.160 --> 02:06.760 Instead of that, it's a very generic function which will pass through your text or any kind of full 02:07.170 --> 02:13.380 document which will return us a very high level document object. 02:14.060 --> 02:20.790 Like a lot of things, tagging, passing, cost, tagging, naming BD recognize just tokenization. 02:20.790 --> 02:23.940 Lemrick, I design everything it will take to extract it. 02:24.300 --> 02:27.660 So this is like an a very common Tarpeena especially. 02:29.600 --> 02:34.040 All right, now, to get the entities we can just simply use, like a E.A., yes. 02:34.740 --> 02:36.120 And let me run it. 02:36.390 --> 02:37.610 Let's see what it was written. 02:38.130 --> 02:45.150 So you can see it has identified that there are three entities are available in this particular statement. 02:45.750 --> 02:52.140 Now, that is Apple and we have a UK and one billion dollar. 02:52.680 --> 02:55.830 Now, other all like an envelope or a known. 02:56.840 --> 03:03.560 So what we can do if we want to get detailed explanation about this entities, we can decide later what 03:03.960 --> 03:04.260 part. 03:06.320 --> 03:14.290 Let's E.A. in, let me bring this E.A. tech sector. 03:14.800 --> 03:23.200 Let's say I want to grab ENFP thought, the label name, that Apple is like an organization or it will 03:23.200 --> 03:24.190 be kind of poison. 03:24.400 --> 03:25.510 So what is the entity? 03:25.540 --> 03:28.490 It belongs to so label undisclosed. 03:29.260 --> 03:33.730 And if I just run, it will get a little bit more detailed explanation. 03:34.030 --> 03:36.280 So Apple is or do your arguments. 03:36.290 --> 03:37.440 It's an organization. 03:37.940 --> 03:40.540 UK Teckman say it's a geographic location. 03:40.570 --> 03:42.940 Cheapy and one billion dollar. 03:42.970 --> 03:44.410 That means it's a kind of money. 03:44.710 --> 03:50.140 If you want to get a little more detailed explanation about this entity type. 03:50.510 --> 03:52.570 We can just simply use SDR. 03:53.440 --> 04:00.190 So just wrapping around that string function as their function and we can use like us Plessey. 04:02.260 --> 04:02.760 Explain. 04:03.140 --> 04:09.970 And here I can pass E.A. not label it so little bit more descriptive stuff. 04:10.580 --> 04:10.980 Hurricane. 04:12.840 --> 04:19.610 Now you can see Happel is organization and its legal description will be something like it's a company 04:19.730 --> 04:21.260 as INSEE institution. 04:22.100 --> 04:23.510 This is G.P payments. 04:23.630 --> 04:27.080 It's like a country's cities, states. 04:27.290 --> 04:32.800 And one billion dollars is associated with money, monetary values, including even you need also. 04:33.680 --> 04:34.070 All right. 04:34.490 --> 04:36.280 So that is about the Hasman. 04:36.740 --> 04:41.150 Now, it has not left any particular over, which is not at all. 04:41.330 --> 04:48.680 I mean, with this entity and it has not recognized in FUSCUS, let's try to apply same thing for S2 04:49.220 --> 04:51.470 and let's see how it goes. 04:52.310 --> 04:56.650 So let's say high school and all this Mikie document to 04:59.580 --> 05:05.750 now has to is like San Francisco considers banning sidewalk delivery robots. 05:06.230 --> 05:13.460 So from this particular document or a sentence, it has just identified San Francisco as a country apart 05:13.460 --> 05:14.200 from all other. 05:14.300 --> 05:17.490 They are not considering as an entity like. 05:17.870 --> 05:19.900 OK, let's call it the third one. 05:21.370 --> 05:29.140 So let me paint when that three first and let me make it documentary. 05:32.050 --> 05:33.050 And let's run it. 05:33.400 --> 05:33.790 All right. 05:33.820 --> 05:40.480 So you can see it as they're displayed DST, that it's a Facebook is hiring a new vice president in 05:40.490 --> 05:40.900 US. 05:41.290 --> 05:48.250 But from this sentence, it has just identified us, has a name entity, and that is nothing better 05:48.430 --> 05:48.910 countries. 05:48.940 --> 05:51.940 And that's perfectly fine, but has a leader. 05:51.940 --> 05:54.200 It has identified Apple as an organization. 05:54.250 --> 05:57.460 But here in this case, it has not identified. 05:57.640 --> 06:04.630 Facebook has an organization or any kind of neat even it has not identified as a name entity. 06:05.050 --> 06:11.850 So that is a provision in a spacy to add some extra tokens as and name entity also. 06:11.860 --> 06:16.240 And we can even assign those things as a specific entity. 06:16.480 --> 06:18.150 So how we can do it? 06:18.440 --> 06:23.640 So what we can do, we how to create a new objects, let's say. 06:25.420 --> 06:28.640 Not the not entity. 06:30.190 --> 06:33.910 If he were displayed here, there will be only one entity that is the US. 06:34.180 --> 06:41.740 So here we need to add one more thing so we can just simply use it like this and let's create some new 06:42.370 --> 06:42.880 entity. 06:43.990 --> 06:46.060 Let me assign it like a list. 06:48.510 --> 06:49.360 So we are ready. 06:49.620 --> 06:56.100 One more angry and this entity we are going to define like Facebook is like an Apple. 06:56.220 --> 06:58.560 So that will be an organization. 06:58.800 --> 07:01.920 Companies, agency or institution kind of naming. 07:03.950 --> 07:09.340 And this thing, I'm just going to assign it to Doctrine ENP. 07:09.800 --> 07:10.130 All right. 07:10.190 --> 07:13.850 So now our objective is to define this new entity. 07:14.330 --> 07:16.490 So let me create a few more subtle. 07:18.580 --> 07:26.100 And to define this new underscore entity, we are going to use the span so span. 07:26.740 --> 07:29.050 And this particular span requires. 07:30.570 --> 07:31.650 Offshore arguing. 07:32.110 --> 07:37.090 So first, the argument is from which document you are looking for some entity. 07:37.120 --> 07:38.580 So that will be a No. 07:38.830 --> 07:45.340 Three, and that will be a very first token only or from which token to which token you are looking 07:45.340 --> 07:45.520 for. 07:45.550 --> 07:46.510 So only for token. 07:46.930 --> 07:48.490 So zero to one. 07:49.120 --> 07:53.110 And what other types of this new entity you are referring to? 07:53.620 --> 07:57.300 So that will be a legacy label and it will be equal. 07:57.670 --> 07:59.380 He cites an organization. 07:59.830 --> 08:02.350 Now this audio is so you need to define it. 08:02.720 --> 08:02.930 So. 08:04.690 --> 08:05.620 How will define it? 08:05.770 --> 08:11.020 So we are going to use these Lochley, not Rocket Springs. 08:13.140 --> 08:19.820 And let me add here, or, you know, this particular span, we have to import it. 08:20.380 --> 08:23.680 So before this new entity, I'm just going to import. 08:25.330 --> 08:29.210 From spacy dark dawkins'. 08:30.940 --> 08:31.460 Import. 08:33.400 --> 08:33.870 Span. 08:34.180 --> 08:40.450 OK, so we have defined a new or R d organization kind of object. 08:41.170 --> 08:44.050 If you just execute it for you will be defined. 08:44.620 --> 08:49.630 Now from this basically tokens, that is a span function is there. 08:49.690 --> 08:53.350 So this band function will create a new entity for us. 08:54.580 --> 09:00.670 From this documentary object, and now we are going to head this new entity to our existing and to do 09:00.670 --> 09:03.080 list in this documentary, Rocky. 09:03.910 --> 09:09.520 So inside this documentary, very well, wherever this Facebook will appear, it will be identified 09:09.520 --> 09:10.760 as a name entity. 09:12.010 --> 09:15.980 All right, so now let us please play Lockheed or even Pierce. 09:18.640 --> 09:25.250 All right, so now you can see Facebook karma, you dartez that means to name and delete has recognized 09:25.610 --> 09:26.270 if you can. 09:27.080 --> 09:29.570 Let's say around this one again. 09:29.930 --> 09:30.300 All right. 09:30.430 --> 09:30.980 You can see. 09:33.110 --> 09:33.400 Hopes. 09:33.500 --> 09:33.970 I hope. 09:34.270 --> 09:36.160 I mean, executed this part again. 09:36.220 --> 09:37.800 That's why it is just that. 09:38.380 --> 09:39.370 So what we can do. 09:39.390 --> 09:41.020 We can just execute this one again. 09:41.890 --> 09:42.460 All of them. 09:43.960 --> 09:52.010 So it will add one more entity, like a Facebook and let me remove this one and let dissipate over everything 09:52.010 --> 09:53.950 that entities in this Dockley. 09:54.490 --> 09:58.300 And you'll be able to see there are two entities it has recognized. 09:58.350 --> 10:01.480 Facebook and the US, Facebook, his organization. 10:01.930 --> 10:04.400 US is like in a country's geographic location. 10:06.400 --> 10:11.650 Now, let's say same thing, if you want to do some kind of Zuli, you want to display. 10:13.220 --> 10:18.830 And for that, we can use from a sea import display. 10:20.850 --> 10:22.000 Actually, it's a display. 10:22.060 --> 10:22.450 See? 10:22.780 --> 10:23.320 All right. 10:23.740 --> 10:28.720 So we can use like in this place, we know, let's say, a vendor. 10:31.330 --> 10:34.030 Now, here we can pass multiple arguments. 10:34.230 --> 10:38.110 Well, it will be against the first argument will be dock's. 10:38.950 --> 10:44.230 So what dogs we are referring to, let's say we want to display Forder knock one. 10:45.040 --> 10:45.890 We our learoyd. 10:45.890 --> 10:54.180 So define this document that is corresponding to this Apple string Apple is looking for and style. 10:54.830 --> 10:55.240 So. 10:55.690 --> 10:56.620 Let's just make it. 10:57.400 --> 10:58.620 What other possible styles. 10:59.380 --> 11:04.210 And you can see there is the one BP heading there to go to their documentation. 11:04.630 --> 11:06.510 But I only know that it's in a. 11:07.590 --> 11:08.250 ENFP. 11:10.150 --> 11:16.440 And let's make it Jupiter easy, equal to prove otherwise, it one this. 11:16.670 --> 11:16.920 Yes. 11:18.570 --> 11:18.920 All right. 11:18.990 --> 11:21.830 So you can see in a very beautiful, nice manner. 11:21.990 --> 11:23.210 It has a scientist. 11:23.230 --> 11:28.730 Apple has organization, UK is a GP and one billion dollar that is like money. 11:29.860 --> 11:35.590 Now, suppose out of this entity, you will just want to find some specific entity only. 11:35.940 --> 11:37.510 So the same function, you can use it. 11:38.000 --> 11:40.580 But one more extra argument you can add. 11:40.900 --> 11:43.360 Let's say option an option. 11:43.360 --> 11:45.130 You can it like on a dictionary. 11:46.210 --> 11:47.790 Let's say d.a.'s. 11:50.130 --> 11:59.910 And let's see, I just want to display organization finally and let me put it a comma like this display. 12:00.790 --> 12:01.140 All right. 12:01.170 --> 12:05.050 So you can see only organization, it has displayed Bud Light. 12:05.160 --> 12:12.780 So that is the whole story about the name and recognition and how to do very easily distinguish functions 12:12.780 --> 12:14.470 are really well inside the space. 12:15.100 --> 12:15.510 All right. 12:15.540 --> 12:16.870 See you in the next video.