WEBVTT 00:01.690 --> 00:02.620 All right, everyone. 00:02.680 --> 00:08.230 So the next topic, which we are going to learn in this video, the vocabulary now matching. 00:08.980 --> 00:16.370 So there are mainly two ways in a spacy you can do the matching like a rule-based matching and a phrase 00:16.380 --> 00:17.020 with matching. 00:17.680 --> 00:25.060 So we won't know right now into a rule-based matching or a face based mating in a python code. 00:25.540 --> 00:28.730 Instead of that, I will show you about one demo. 00:29.290 --> 00:30.790 So let me go to that Nemo. 00:31.330 --> 00:38.890 And the creator of this spacy day who created this demo of Rule-based Metcher Explorer. 00:40.090 --> 00:41.690 Now here, the writing biting cold. 00:41.800 --> 00:43.780 That won't be a very big deal. 00:44.260 --> 00:50.200 So the important thing is that how you define about some kinds of pattern, the pattern you define in 00:50.200 --> 00:51.520 case of regular expression. 00:51.850 --> 00:55.260 So this is like a little advance pattern. 00:55.390 --> 01:00.160 You can define with respect to different vocabularies of this natural language processing. 01:00.550 --> 01:07.450 So the important task is that how good you are at a defining pattern while you are searching for something. 01:07.930 --> 01:15.750 So this is the rule based matching pattern demo they created and will how little practice it? 01:16.220 --> 01:22.970 How you can define those pattern and often the pattern guard defining the match found close or not. 01:22.990 --> 01:25.710 That is the job of this spacy labeling. 01:26.320 --> 01:33.820 Now, if you want to go much more detail into this documentation of rule-based matching and a face based 01:33.820 --> 01:41.350 matching, you can refer to this spacy documentation that is a kukan measure that is nothing but a rule 01:41.350 --> 01:42.040 based measure. 01:42.520 --> 01:44.140 And there is a phrase mismatch. 01:44.920 --> 01:45.300 Quite right. 01:46.030 --> 01:48.660 So by default, they have given some tests. 01:48.760 --> 01:52.270 So out of this text, we are going to search for something. 01:52.840 --> 01:57.430 And our default model will be a small size of a core English. 01:57.550 --> 02:00.520 Martin, let me select this so token. 02:01.120 --> 02:04.120 Now, already there are a number of patterns they have already defined. 02:04.450 --> 02:07.840 So as of now, it just doesn't make any sense for us. 02:08.140 --> 02:09.730 So let me just remove this token. 02:10.710 --> 02:14.820 So let me be Lamarr will be match and. 02:15.670 --> 02:15.960 Horse. 02:16.330 --> 02:19.960 That means it's a part of speech, tagging will be known. 02:21.520 --> 02:23.350 And even let me remove this. 02:24.520 --> 02:28.150 Now you can see our pattern of variable is empty. 02:28.720 --> 02:31.210 So let's add our very first token. 02:32.170 --> 02:39.820 Let's say out of this particular full tax or I would say document, I want to search for something. 02:40.190 --> 02:44.070 So all those rule, you can define it as an accident. 02:44.650 --> 02:48.880 So something like you want to search for something in a lower piece. 02:49.210 --> 02:53.860 So all those rules will be applied on every single token. 02:54.610 --> 03:01.420 So first, this will be passed to this model and this model will create those token. 03:01.800 --> 03:08.710 And based on all those tacked on, we will define that, will search for those kind of back on inside 03:09.280 --> 03:10.750 all those tokens. 03:11.620 --> 03:13.300 So let's define our first. 03:14.640 --> 03:15.010 Pattern. 03:15.920 --> 03:19.490 So let's say there are a number of rules you can define here. 03:20.570 --> 03:25.610 Let's say we want to search for something is legit or not. 03:25.880 --> 03:29.860 In this particular stock and I'm just making selected. 03:30.830 --> 03:36.320 And you will be able to see one pattern and define in a dictionary for me like that. 03:36.320 --> 03:41.850 It will be like he's underscored digit will become true and fighters that have faced attacks. 03:42.670 --> 03:43.640 You'll be able to see. 03:44.700 --> 03:46.180 That is no match found. 03:46.560 --> 03:53.610 But suppose in between somewhere, I would just skip, let's say two, three, four and five fist attacks 03:54.110 --> 03:57.450 will be able to see it has able to find out. 03:57.930 --> 03:59.250 Two, three and a four. 03:59.550 --> 04:00.860 That means that is BGT. 04:00.900 --> 04:05.790 So if anyone a phone's digital in a token, it will give us. 04:06.290 --> 04:07.720 Suppose along with it. 04:07.780 --> 04:08.700 Two, three, four. 04:08.730 --> 04:11.500 Suppose I just make WUIS. 04:12.180 --> 04:13.710 So it will be a one gonna to do. 04:13.730 --> 04:14.920 It will be a different token. 04:14.970 --> 04:16.770 Let's see how it goes. 04:17.280 --> 04:20.820 So you can see it is considered as a single token. 04:21.240 --> 04:22.470 And that's like this. 04:22.620 --> 04:25.480 He's dead bodies in this digital phase. 04:25.890 --> 04:29.670 And it is unable to find out any match. 04:31.000 --> 04:31.300 All right. 04:31.800 --> 04:32.790 So that is one thing. 04:33.780 --> 04:35.280 Let's see something as. 04:36.290 --> 04:37.790 Let's say entity type. 04:38.690 --> 04:41.420 So let's say entity type organization. 04:42.440 --> 04:47.120 So is there any entity, diabetes organization in this particular tax? 04:47.690 --> 04:49.110 So most probably one one. 04:49.490 --> 04:52.820 So let's just keep let's say, the company Google name. 04:53.420 --> 04:54.460 Let me keep it here. 04:54.470 --> 04:56.290 Even a specialist, I am not sure. 04:56.300 --> 04:59.660 Species of organization or not. 05:00.230 --> 05:01.820 And let referenced attacks. 05:02.850 --> 05:09.830 And you'll be able to see Google got found, Google match found, that means Red River, that entity 05:09.880 --> 05:13.170 type his organization in any of the tokens. 05:13.670 --> 05:15.310 That will be a match phone. 05:15.630 --> 05:16.950 But SBC has. 05:17.620 --> 05:20.120 It's a high voltage. 05:20.980 --> 05:21.330 Wrong. 05:21.570 --> 05:22.950 So let me refresh this. 05:25.310 --> 05:29.480 No, no species, not a organization in that database might be. 05:31.360 --> 05:31.720 All right. 05:32.060 --> 05:35.830 So that is like a very simple single attribute we have, right? 05:36.560 --> 05:37.220 But suppose. 05:38.610 --> 05:44.980 Along with the hurricane, let's just go somewhere else, let's say lemme so lemme Morozova. 05:45.540 --> 05:46.900 So there is a B.. 05:47.490 --> 05:52.800 So any token who's Lamma was and will be a B, it will find out. 05:53.100 --> 05:54.210 Spillages refreshing. 05:55.240 --> 06:01.690 And you can see he's lamb of LBB, high lamb of LBB, so Vedova and me, Lhamo, so will we be. 06:02.670 --> 06:04.280 So it has found a match. 06:04.580 --> 06:05.880 He's hot. 06:06.280 --> 06:06.790 And a meat. 06:08.650 --> 06:11.000 All right, so let's try something else. 06:12.200 --> 06:12.690 Let's see. 06:13.710 --> 06:15.350 We will keep Leiker a stop. 06:15.800 --> 06:17.960 So is there any stoppers are not. 06:18.680 --> 06:19.790 Let me refresh you. 06:20.840 --> 06:23.470 And you'll be able to see all those stopwork. 06:23.660 --> 06:24.770 It is able to fine. 06:25.670 --> 06:27.700 Now let's hear something else. 06:29.760 --> 06:37.100 So let's say lower and we want to search for something in the lower, let's say small wooden. 06:37.830 --> 06:39.140 So small. 06:41.420 --> 06:44.640 And then you'll be able to see lower will become small. 06:45.330 --> 06:46.850 Let's add one more token. 06:47.670 --> 06:49.500 It will be a lower and. 06:50.700 --> 06:52.380 Let's say we are searching for one. 06:53.970 --> 06:56.280 So the Lord wasn't of any. 06:57.510 --> 06:59.280 Will be find out. 06:59.970 --> 07:03.960 Let's refresh it and you will be able to see small wooden. 07:04.500 --> 07:04.990 Find out. 07:06.480 --> 07:09.420 Now, suppose, along with the small wooden. 07:11.370 --> 07:11.970 Somewhere. 07:12.190 --> 07:13.100 I'll just make it. 07:13.980 --> 07:17.300 Capitals on capitals, capitals, small. 07:17.550 --> 07:17.720 Whoo! 07:18.450 --> 07:20.230 And let's see. 07:21.380 --> 07:23.610 And let's see whether it is able to find out or not. 07:25.180 --> 07:25.570 All right. 07:25.610 --> 07:28.960 So you can see small more then also got final. 07:29.690 --> 07:37.310 Now let's have one more thing I want to tag along with a small wooden in between, that has to be some 07:37.310 --> 07:38.300 magic bullet. 07:39.170 --> 07:42.380 So we have to keep it some digital here. 07:42.470 --> 07:43.700 So let me remove it. 07:45.000 --> 07:45.840 Let's add. 07:47.490 --> 07:47.910 Hoopes. 07:49.470 --> 07:51.270 So Lowitt, let's say. 07:52.660 --> 07:54.670 Small and. 07:56.320 --> 08:01.120 Let's add one more token, let's say the Jeep Hayes digit. 08:02.020 --> 08:08.090 Let's add one more Pokemon like Lowitt and it will be wooden. 08:09.070 --> 08:09.460 All right. 08:09.700 --> 08:11.190 And let's Surfest attacks. 08:11.920 --> 08:12.290 All right. 08:12.290 --> 08:19.930 So you can see now it isn't it has not found those match like a small wooden, but it is searching for 08:19.930 --> 08:20.800 small then. 08:20.890 --> 08:21.620 Is there any d.g? 08:21.910 --> 08:24.380 And then there is a wooden token. 08:25.300 --> 08:25.720 All right. 08:25.740 --> 08:31.970 So you can do all those kind of mysterious stuff with this viler finding the patterns. 08:32.290 --> 08:35.890 And you can see this way, you can define those patterns. 08:36.190 --> 08:38.260 Now, how to do it in our patent code? 08:38.380 --> 08:40.330 We will see in the next video.