0:00:00.000,0:00:00.933 Hi, everyone.[br] 0:00:00.933,0:00:05.521 We know that most of you still working really hard on the last couple of assignments and problem sets[br] 0:00:05.521,0:00:12.110 but nevertheless our NLP class is coming to its close and so we thought we’ll both update you on a few of the issues.[br] 0:00:12.110,0:00:14.815 And first of all, everyone is always interested in numbers[br] 0:00:14.815,0:00:17.092 so let’s say a bit about the numbers.[br] 0:00:17.092,0:00:24.441 So, it seems like, as we come into week 7’s material of the problem sets and programming assignments,[br] 0:00:24.441,0:00:28.416 it seems like there are about 5,000 people still actively watching the videos[br] 0:00:28.416,0:00:32.423 and about 2,000 people doing all the work of the homeworks.[br] 0:00:32.423,0:00:37.186 But that’s, you know, overall it’s just been an enormous amount of stuff going on 0:00:37.186,0:00:40.626 so it now have well over a million video views and all.[br] 0:00:40.626,0:00:43.186 Yeah, it’s very exciting.[br] 0:00:43.201,0:00:47.464 Programming homework six, the parsing homework, was definitely our hardest homework so far[br] 0:00:47.464,0:00:50.468 and there were definitely problems with the code being under-commented[br] 0:00:50.468,0:00:53.571 so we appreciate very much those of you who helped out each other in the forums 0:00:53.571,0:00:56.674 and made things very clear for each other. 0:00:56.674,0:01:01.205 Those of you who managed to finish the homework have told us they got a lot out of it.[br] 0:01:01.205,0:01:04.366 But those of you who didn’t have the time to do all of it,[br] 0:01:04.366,0:01:07.318 we encourage you to do come back and do programming homework seven and eight[br] 0:01:07.318,0:01:11.178 which are much easier than homework six and should be a lot fun as well.[br] 0:01:11.178,0:01:14.526 One of the things that some people have been asking about is,[br] 0:01:14.526,0:01:18.831 “Hey, can we take those great parsers that we did write for programming assignment six[br] 0:01:18.831,0:01:23.762 and keep on working on them by posting them on Github or some other site like that?”[br] 0:01:23.762,0:01:26.964 There is a slight problem with that because, you know,[br] 0:01:26.964,0:01:29.943 we are hoping that we’ll be able to do this course again some time in the future[br] 0:01:29.943,0:01:32.272 and we’d like to be able to reuse some of the assignments[br] 0:01:32.272,0:01:37.723 so I’d really prefer that there just aren’t great solutions to all the assignments sitting on the open web[br] 0:01:37.723,0:01:41.870 and to keep some of that stuff on the forum sites for the class.[br] 0:01:41.870,0:01:46.526 So, it’s great that people want to keep exploring and doing better things for NLP[br] 0:01:46.526,0:01:50.135 but we’d really think it’s probably a better idea for you 0:01:50.135,0:01:53.745 to pick up one of the several open source NLP frameworks that’re out there 0:01:53.745,0:01:56.542 and look at ways in which you can contribute to that[br] 0:01:56.542,0:01:59.370 ’cause that both doesn’t sort of conflict with our assignments[br] 0:01:59.370,0:02:03.591 but that’s actually going to be more useful for you and people around the world in general[br] 0:02:03.591,0:02:05.941 if you’re helping along these open source frameworks.[br] 0:02:05.941,0:02:09.585 So there are good ones — For Python NLTK is the best-known one; 0:02:09.585,0:02:13.350 and there are several well-known Java NLP open source frameworks[br] 0:02:13.350,0:02:16.509 which certainly includes our own Stanford NLP tools[br] 0:02:16.509,0:02:20.636 but also other things like OpenNLP, the GATE NLP Framework;[br] 0:02:20.636,0:02:24.194 components: UIMA — a bunch of stuff out there if you look around.[br] 0:02:24.963,0:02:31.394 Okay and that brings up the issue of what’s going to happen with this class website after the class ends. 0:02:31.394,0:02:33.806 It is going to stay open and available 0:02:33.806,0:02:37.656 and all you guys will be able to keep on looking at stuff and referencing things 0:02:37.656,0:02:43.021 and even beyond that, what we’re also going to have is this going to be available in an archive mode. 0:02:43.021,0:02:46.153 So, people who haven’t been registered in the class 0:02:46.153,0:02:50.227 will also be able to look at the content in terms of the videos that are up there 0:02:50.227,0:02:52.605 but not the programming assignments 0:02:52.605,0:02:56.030 which we are kind of going to keep to the people who were enrolled in the class. 0:02:56.030,0:02:58.387 But you shouldn’t worry about things going away. 0:02:58.387,0:03:00.413 Yes and those of you who are enrolled can still — 0:03:00.413,0:03:02.412 the forum will stay up — 0:03:02.412,0:03:05.375 we encourage you to continue to talk to each other on the forums. 0:03:05.375,0:03:07.819 Those of you who never got to finish some of the homeworks, 0:03:07.819,0:03:09.313 keep working on the homeworks, 0:03:09.313,0:03:11.733 go ahead and post them. The site will stay up. 0:03:11.733,0:03:16.055 And then for new people we will eventually be teaching the class again 0:03:16.055,0:03:18.300 but meanwhile we’ll be leaving the videos up 0:03:18.300,0:03:21.068 so if your friends want to watch the videos, we encourage that. 0:03:21.068,0:03:24.583 And then eventually we’ll actually teach the class again. 0:03:24.583,0:03:29.747 In fact for everybody, it’s been really helpful on the forums there — 0:03:29.747,0:03:33.168 we mentioned this earlier about helping people out with the code, 0:03:33.168,0:03:35.923 helping people out with basic NLP. 0:03:35.923,0:03:39.960 It’s been very supportive community and we really appreciate that and we think it’s really great. 0:03:39.960,0:03:42.808 We’d like help from you in suggesting — 0:03:42.808,0:03:46.983 you have already given us suggestions for next year but I’ve put up a forum just now 0:03:46.983,0:03:51.926 that we’d like you, give us specific suggestions about ways we can make the course better for next year. 0:03:51.926,0:03:53.815 We know Homework six is one of them 0:03:53.815,0:03:57.377 but there’s lot of suggestions you’ve given about improving the problem sets 0:03:57.377,0:03:59.907 or other reading material we can suggest 0:03:59.923,0:04:03.559 or anything else that will help us improve our course for the second time we teach it 0:04:03.559,0:04:05.740 that would be great. 0:04:06.356,0:04:12.866 We all have dreams of the future where there are computers with full natural language understanding 0:04:12.866,0:04:16.473 and in our research we still variously hoped to work to those goals 0:04:16.473,0:04:21.992 but in terms of the practical NLP that’s deployed around the world at the moment, 0:04:21.992,0:04:25.391 really you guys are just seeing the ton of the stuff that are actually used. 0:04:25.391,0:04:30.590 Things like text classifiers, building sequence classifiers, named entities and other things,[br] 0:04:30.590,0:04:34.530 parsers, questions answering, machine techniques,…[br] 0:04:34.530,0:04:37.429 So you really should feel good if you’ve made your way through this class 0:04:37.429,0:04:41.974 that you are a competent practitioner of the kind of useful NLP techniques 0:04:41.974,0:04:47.515 that’re leading to a new class of more intelligent language-wielding computer applications 0:04:47.515,0:04:51.188 and so we hope that you’ll be able to take these ideas and knowledge, 0:04:51.188,0:04:56.227 and go off and apply them in many different places. 0:04:56.227,0:05:01.481 We clearly are in this world now where human language material is just everywhere over the web 0:05:01.481,0:05:06.929 as part of the new move to the content authoring and social computing that’s going on everywhere. 0:05:06.929,0:05:09.378 So much of that is about language use. 0:05:09.378,0:05:13.088 So now you have a good toolkit to be able to go off and do things with that material 0:05:13.088,0:05:16.011 and we hope that you’ll be able to find good things to do. 0:05:16.011,0:05:21.154 Yeah. Thanks for taking the class and look forward to seeing you wherever and whenever we see you. 0:05:21.154,9:59:59.000 Thanks a lot.