Named Entity Recognition - Natural Language Processing With Python and NLTK p.7

Key Takeaways

what's going on everybody Welcome to part seven of our python with nltk for natural language processing tutorial video series in this video we're going to be talking about named entity recognition with python uh Python's nltk module so let's go ahead and scroll on down here and we really don't need any more of really any of this uh we'll comment this out for now uh you might want to return to it later but actually I'm going to get rid of it uh cuz I'm not using anymore in this tutorial so I'll get rid of it anyway now what we're going to do is uh talk about some named dentity recognition so we've got tagged that still has our parts of speech and all that and then what we can do is we can say named n for named entity and that's going to equal nltk NE chunk and then we want a chunk via named entity tagged and then we can do [Music] named. draw not let save and run that close this okay so here you can see part of what is causing some problem immediately uh we've got white house so facility white organization house person Eric but not catching Draper but I think it doesn't catch Draper in this instance possibly cuz there's no space between this these words um anyway continuing along no named entities found okay served America found that at least thank you and uh before we go any further I probably should um show you guys what these words even mean so here are I'll just paste them right in here these are named entity type examples so you can have an organization a person location a date time money uh percent facility gpe this is just geographical kind of location is so this would be a location but this is more like a general location uh so anyways those are the types let's go back over to what we're showing up here no named entities there none there whoops misclicked come on give us another one driv me nuts okay our Union is strong okay so that was a a correct grab there uh so let's close this though let's go over here close this okay and what you can also do though if you don't care uh it'll still pick these up but if you don't care about these very specifically what you can do is you can say this and then binary equals true and what this does for you is it just classifies everything is a named entity period it doesn't actually tell you like any like organization person location anything like that cuz like I said sometimes it's just inaccurate uh unexpected indent what have we done it's probably just angry at this I don't know I don't see anything else that should give me an unexpected indent yeah okay so here you can see that the named entity chunk was correct at least in this instance to chunk White House together okay uh so that was good it probably would have got Eric Draper if if that space wasn't an issue but anyway it's finding these named entities now um I think we had a few to go before we found any more Nam entities so I just want to kind of X out a few Applause I don't think we've made it past that I can't remember shoot let's clicked again H I'm not really seen any more um identities I can't believe we're not finding see if we can find an identity protect our people maybe people should have been one oh we found some though here's United States his named entity and so on so anyways good enough uh we'll close this out and uh let's go back one more time to the original version um right so if we look we'll see here that these are separated when you did without binary okay so remember this would be before it was binary equals true this one not to take out my trash someone do that for me uh facility organization so it's it's calling white and house separate right they're not actually being called the exact same like they're not part of the same entity as opposed to when we called this to be binary true you'll find that they were not separate see they are part of the exact same figure out how to get that little tag thingy to show up I like that give me it whatever jerk there we go got it uh anyway it's classifying them as the same named entity so that might be useful for some of you if you are hoping to get those together same thing with Eric Draper if that space wasn't there it would pick it up uh but I think if it wasn't binary uh or if it yeah if it wasn't binary it would probably say to anyway it can get kind of annoying so uh nltk named entity recognition can be really useful but it's like false positives are kind of high and its error rate is also kind of high so you'd want to have something else in place to maybe check the N dentity recognition um but you know that's it for the basics of the named dentity recognition it's actually pretty useful to have but I've sometimes I go back and for sometimes I'll use the name d recognition but sometimes I'll just look for nouns and that almost works just as good so anyway uh that's that you can also combine the two though keep that in mind uh so if you find if you use name deny recognition you can search for all nouns and then also if you find uh a named entity that has two named entities back to back chances are they actually should go together why nltk doesn't have that kind of understood I don't really know the I'm sure the people there are much smarter than me so I'm sure there's a really good reason why um but anyway that works for me at least so maybe if you're looking for something specific like that uh you might want to try that out so that's it for the name entity recognition if you have any questions or comments please feel free to leave them below otherwise as always thanks for watching thanks for all the support and subscriptions and until next time