WEBVTT 00:01.370 --> 00:01.980 Hey, everyone. 00:02.010 --> 00:09.860 So next Project Cough and LP, which we're going to work upon, is a tax simulation and for illustration. 00:09.950 --> 00:12.510 But it was kept here one big tax. 00:13.080 --> 00:15.350 So what is tax summarization is. 00:15.780 --> 00:20.250 So we all know that we are all bombarded with lots of lots of data. 00:20.640 --> 00:25.920 There is lots of articles we keep on reading on a daily basis. 00:26.310 --> 00:28.490 So there are lots of data around us. 00:28.980 --> 00:30.460 Mostly in a tax forms. 00:30.990 --> 00:33.260 I won't say all because I in a tax form. 00:33.750 --> 00:37.110 But many times we keep on releasing a tax forum. 00:37.560 --> 00:40.830 And it consumes lots of lots of a long time. 00:41.280 --> 00:45.790 It may happen that there will be, let's say, 5000 words article. 00:45.960 --> 00:50.090 And maybe the important point will be just five or 10 important points today. 00:50.520 --> 00:56.940 So even if you just can perhaps somehow those five important points, that will be sufficient for you 00:57.600 --> 00:59.910 to grasp those whole article. 01:00.450 --> 01:07.800 So the idea behind this automatic tax symbolizes and is the same that whether you can pick up all those 01:07.800 --> 01:16.380 important 10 important statement on a sentence or some kind of somebody out of it, can you grab it 01:16.950 --> 01:21.330 through automatic this natural language processing related techniques? 01:21.900 --> 01:25.080 And that is what the idea behind this tax summarizes. 01:27.110 --> 01:33.270 Something like a finding all those useful information out of huge amount of tax. 01:33.860 --> 01:37.020 And even it will reduce your reading time also. 01:37.420 --> 01:40.430 So you can even possess more and more documents. 01:40.490 --> 01:46.660 So you can be very much choosy in choosing or selecting some kinds of article. 01:46.910 --> 01:51.230 Because now you can process more and more article while reading. 01:51.530 --> 01:56.870 So mainly these are some of the big advantage of this tax summarizes an application. 01:58.290 --> 02:06.200 And as I told you, I already created this particular tax so you can see Maria set up for a related 02:06.200 --> 02:07.710 one article, is that so? 02:07.710 --> 02:14.500 I would highly suggest to you that you just pause this video and read this whole article, which I kept 02:14.500 --> 02:16.770 putting a factual string. 02:17.400 --> 02:22.200 Let me executive and let's see how many characters are there inside this. 02:28.010 --> 02:34.540 So you can see we have fifteen hundred sixty three characters out there inside this particular stream. 02:35.000 --> 02:43.100 So instead of writing such a huge amount of text, can we just come up with some limited set of sentences 02:43.160 --> 02:46.930 or words like a hundred maybe or maybe 200? 02:47.540 --> 02:49.580 And that will just easily summarize it. 02:50.090 --> 02:55.520 So, as I told you, just pause this video and read this full article. 02:55.550 --> 03:00.820 So when we find a summarization of this particular article, you'll get to know about. 03:01.040 --> 03:06.620 How good of a summarization is because as a human, you can create those somebody easily, as far as 03:06.620 --> 03:12.470 you can understand, those cyclical tax or any huge amount of tax you are trading on. 03:13.100 --> 03:13.450 All right. 03:13.520 --> 03:18.600 So let's proceed ahead with importing this minimum library. 03:19.130 --> 03:26.750 And then we'll see what techniques and mechanism we are going to apply to choose all those important 03:27.320 --> 03:29.290 information out of this article. 03:32.150 --> 03:33.630 So let me create. 03:35.130 --> 03:37.130 Let's import less Pesi Lively 03:39.830 --> 03:41.630 and Fromm's Tracy. 03:43.360 --> 03:47.420 Not Lange, not even. 03:50.220 --> 03:54.850 Dutch Topo's We are going to import stop. 03:56.230 --> 04:01.780 And one more thing from this ruling class, we are going to import increased punctuation. 04:05.150 --> 04:05.600 All right. 04:06.040 --> 04:06.340 Let me. 04:06.530 --> 04:06.880 Security. 04:09.220 --> 04:11.640 Let's load our small size, Martin. 04:12.490 --> 04:17.950 So it will be NLB, spacey, dark lord. 04:19.920 --> 04:20.800 An endless court. 04:20.850 --> 04:21.170 Court. 04:22.390 --> 04:24.760 And this put a verb in this quote, test him. 04:25.480 --> 04:26.380 Let me run. 04:30.730 --> 04:39.940 Let's apply this whole tax honor and be model so far that we are going to use this in L.P tax. 04:41.640 --> 04:44.180 Let me assign it to some dark object. 04:47.300 --> 04:52.280 And legislate played over every single token between will cook in. 04:53.230 --> 04:56.780 No, let's bring them to conduct tax. 04:57.620 --> 05:00.680 And let us keep it as a list comprehensive. 05:02.030 --> 05:04.360 So that will be netting what tokens? 05:08.360 --> 05:12.230 Let me print pawprints. 05:16.460 --> 05:16.820 All right. 05:16.850 --> 05:20.810 So these are the tokens with which we are going to work up on. 05:21.590 --> 05:26.180 All right, so one more thing is if you just display this punctuation. 05:29.130 --> 05:30.910 So these are all contrition marks. 05:31.770 --> 05:36.400 Now, what we are going to do, we are going to add one more contrition that will be slashed. 05:36.590 --> 05:40.140 And so new line and let me. 05:40.140 --> 05:42.540 Assigning to same contrition. 05:47.730 --> 05:49.260 And if you displayed. 05:51.570 --> 05:55.200 You can see now less and also exist as a punctuation mark. 05:55.920 --> 06:01.980 All right, so deserted some of the minimum stuff while reading your actual article, which we are going 06:01.980 --> 06:03.960 to use for the somebody purpose. 06:04.650 --> 06:05.010 All right. 06:05.970 --> 06:06.990 So there is a first step. 06:08.360 --> 06:08.780 Hopes. 06:11.890 --> 06:14.690 I kept here the first Luol heading. 06:14.950 --> 06:17.030 Let me make it here. 06:17.200 --> 06:18.540 So firstly, 130. 06:21.960 --> 06:26.480 Next, is tax planning really do and then sentenced tokenization. 06:27.180 --> 06:27.870 And before that? 06:29.020 --> 06:34.030 Let me discuss what idea on which we are going to create this summarization. 06:34.780 --> 06:41.860 So there is a one basic technique which we can apply, like we can try to do the score of individual 06:42.000 --> 06:45.580 sentence so we can do the sentence tokenization. 06:46.180 --> 06:53.140 And to each particular, let's say, sentence, we will give us some score and we can try to find out. 06:54.390 --> 07:00.510 The highest court or first, let's say 20 or 30 percentage of those standards, and that will be nothing 07:00.540 --> 07:01.940 but a tax summarization. 07:02.910 --> 07:09.560 But now the question is how to view those scores to individual sentence for that. 07:09.570 --> 07:15.690 There is a one basic idea you can apply, like we can first create a word frequency. 07:16.170 --> 07:22.800 And those frequency will give us the information that how many times each individual vocabulary word 07:22.860 --> 07:24.330 here appears. 07:24.900 --> 07:34.370 And based on that, we can give some score to individual words in a sentence and score. 07:34.470 --> 07:39.810 We can just add it up according to individual voice appeared in a sentence. 07:40.170 --> 07:46.380 So as we process along, you will get idea in sooner text leaning part in a sentence, tokenization. 07:47.010 --> 07:50.540 In the next video, we will see how to achieve those things. 07:50.640 --> 07:54.080 First, we'll try to create those word frequency contact. 07:54.610 --> 07:54.930 All right. 07:55.000 --> 07:56.420 So see you in the next video.