Latest / Semantic SEO Podcast / Hypernyms, Hyponyms, Meronyms and Antonyms: Using Lexical Semantics to Improve Semantic SEO Agencies' Content
Transcript
- James Dooley: Lexical semantics in semantic SEO today I'm joined with Pavel Klimakov and we're going to go through all the different names and there's a lot of them and some some completely confused me but some when I do start to learn them actually now I started to help by prompting to be so much better with artificial intelligence so I want to kick off with starting by saying with regards to the different names why is it important in your opinion for anyone that wants to improve their semantic SEO to understand the different names Pavel Klimakov: Sure so hyponyms hypernyms and meronyms these three would be the main things let's say to focus or pay attention to these things are primarily attached to core eyes framework and core eyes teaching for the main use of let's say optimizing or tweaking the overall representation of the content this actually applies to all three levels which is the topical map the briefs and then the actual content that is being written what these things are in nature is it allows you to take a kind of word or an object let's say focus on it as a central piece and then kind of dissect it or split it and realize what are other things that are either similar or closely related what this allows you is you're gathering very tightly interconnected concepts and for the most part if you provide your answers if you make your let's say contents like the briefs the overall planning of the topical map it has if it considers these elements at all steps your representation will be way more interconnected and the key word here is everything becomes much more densely represented there is less chance to kind of go all over the place and start to be vague about how you provide information the things you talk about and what basically the opposite happens you start to be more compact more condensed and that means there is a much higher chance for you to provide more value with more words which is one of the key components for lower cost of retrieval you're saying more with less words which is obviously a very important part with regards to semantic SEO the cost of information retrieval with regards to the NIMS I'm going to give you an example and yeah I don't think you'll like this example because this is very old-school TFIDF type BM 25 algorithm type things where I started to use within AI prompted acronyms so like SEO for search engine optimization and stuff like that and where I used acronyms was in certain industries that we do lead generation for I wanted to know all the acronyms related to that industry so it could have been like insurances or compliance legislation and what it started to do is it started to open up certain things to me that was like I've never even heard of this acronym and then it made me when I was like this is very relevant or it could be like CHAZ and stuff like that which is to do with health and construction health and safety and stuff like that and there were certain acronyms that I'd never heard of which I was able to get onto the page and actually started to realize that we've like query expansion or query augmentation or whatever you want to call it I started to run for more key words because people were searching for that acronym with the products or services that we did so but I wouldn't have known how to say it without saying give me all the acronyms to this industry so that is one of the NIMS that I've been using with regards to prompting right James Dooley: Can you give an example where someone might use the meronyms the hyponyms or the hypernyms in a way of being able to prompt AI to make you the in the content more dense like like there's more information in there in less words but you've got all these important kind of characteristics of what's needed Pavel Klimakov: Sure so let's say in a case where there is a writer that's already working with the brief usually what they do have is a kind of heading or a question that they need to process in some certain cases they might have scenarios where additional let's say words hints numbers statistics or other elements are already introduced they know that these elements will be a part of the answer typically you know if we're asking like what's like how much does this insurance cost or what's the point of like this thing versus that thing the thing itself is already in the question so that's kind of the primary focus from here if let's say it's a bad scenario where the writer doesn't know the industry but they're still writing the answers not a great case but then they can go to let's say any kind of AI and start to ask what are the specific let's say meronyms of this specific term here comes a little bit less a tricky part though because for all of these names basically there are programmatic like packages or pieces of a code that have already all of this like dictionary aligned with each other that you can basically search a specific word and identify what are these like meronyms of this word meronyms by the way basically are just a parts of the thing so if you're talking about you know a computer a computer will have a processor and then the memory and things like that or if you have a football team then there's players and coach and etc so that that's the logic there so in a scenario where a person that who is not educated but there's somehow need to be sounding smart and provide good enough answers this is a kind of quick hack you can ask an LLM to give you these meronyms to provide the winning dancer the only problematic case is that as I mentioned the more like let's say realistic use for these things would be done programmatically if let's say a kind of system wants to check if such meronyms or if such words are present if you're gonna ask an LLM it is most likely to kind of give you the general direction which might be overlapping with that basically data set or let's say those programmatically preset things the main idea here is that this is just a hint for what are the right things to say and in a scenario let's say if you're dealing with a topical map let's say you're working things out and you gathered some concepts maybe you realize this is not this might not be covering all of the things you can take the other two components which is hyper names and hyponyms over here you can think it's think of it in a like slightly simpler way hyper name it moves the concept upwards meaning it groups something so if we take let's say a word or let's say a concept like France which is a country what is a group of that that is countries right so from the word France towards countries you have this connection and if you go for a hyponym this is the basically opposite direction you go from a group of countries towards something smaller or let's say fruits and let's say apple or banana so that's the kind of direction the most important idea here is that it just allows you to think about whatever is in front of you in a slightly different way it's kind of like a trick or a kind of quick tool that oh wait maybe I forgot about that so let me think about this thing by the way just as like a complete side note but I think I saw somewhere some long time ago there was like a YouTube video or something like this like if you forgot what like what if you run out of things to say in a conversation or something like this it's like remember the last thing and then think that you can generalize it go more specific or like connected to the other similar concept it's something like that so this is just a kind of hint how to be unstuck in a case where you're not really well educated but this is not really a kind of I don't know like a hack or some kind of a secret weapon or something like that James Dooley: What's interesting and the reason why I wanted to do this episode with you right I'm just I'm because I always get muggled up between the hyponyms the hypernyms the meronyms and stuff like that so I explained on the acronyms of how I use that and I've just got it up now where we use meronyms right I'm probably saying that wrong or meronyms of let's say we do wet pore surfacing which is a rubber surface on a playground right and where I do meronyms is in the section usually to do with price and what we're normally doing is is we're then trying to get all the meronyms of how that affects the overall price and sometimes it brings something back that says oh okay now we can give what is the cost of wet pore the cost of wet pore is and then give a figure underneath it what influences the cost and then we're getting all the different elements that make up what wet pore is and how that influences the price element that's where we use the meronyms this is just for us a page level the hyponyms is more like the different types so like types of playground surfacing and stuff like that the hyponyms which kind of like you said goes up one group that's where normally we're trying to do okay it goes up a group into the next thing what else is in that group that we might be able to do comparisons against so that versus this did you see that there so we can use that for then the comparisons so that's where but without using the words like hypernyms, hyponyms, meronyms I feel like it's by using them words now and the team have been prompted when they're writing content okay in this section try to find out how this might influence that or in this section look at the hypernyms for if there's any comparisons that we could do in the article it might end up being in the in the the micro content or it could end up being like on a third party source where we can do James Dooley versus Pavel Klimakov or whatever it's going to be and we can do a comparison that link back through to our site saying James Dooley is far superior to Pavel Klimakov because blah blah blah blah blah and then we're able to do like brand jacking and stuff like that and that's where some of it is and there's one last one and I think I've got this from yourself I'm pretty certain I did the antonyms and we're obsessed now with antonyms or when we're doing a review page all of our reviews used to be very positive sentiment and I feel like with AI if it's too positive they don't like it and it almost needs to have a part of it that says here's not if you do prom pros you need to have a con section but you try to make the con section not that bad if you're obviously looking to sell but we also use it for query fan out terms on third party like so we would write articles about let's say I've got a brand called James Dooley I would do an article for James Dooley alternatives or why should I not use James Dooley but then in the article I frame is being you do want to use James Dooley so I'm turning the antonym into a positive and framing and controlling everything back to what I do so that's how we're using it in real life businesses obviously you're using the names in in a different way do you use the name the antonyms in anything for a semantic content network a topical map our content briefs Pavel Klimakov: Sure so I want to comment on yeah that you mentioned yeah the antonyms is actually a very very powerful thing and I learned this through Cori through his very very very old videos that were like four or five years ago or something like that he was this is this just seems to be the overall kind of decision that Google took on at some point and they are just stuck to it but basically they always put okay don't want to say always but like really almost always they want to put the opposite next to each other so if you're looking for the benefits of something they want to give you the risks next to it all the time and I do think this has to do with this like safety kind of like neutralizing the answer that they're providing to not misdirect anyone who is searching so you know if you're let's say searching for like some medical benefits of a specific pill and but their side effects they have to say that this is not just a good thing you need to be aware of the whole picture which is why I don't think they want to kind of dissect the pluses from the negatives they want to always bring them together and then when it comes to things like reviews I can say that they did some kind of study from I don't remember which research paper but from many months ago they were just checking reviews I believe from Amazon and what they found there is quite easy to identify like fakely generated reviews they follow certain like principles and patterns but for the most part if you take like one account who spreads reviews reviews would be overly positive but the real review is almost never positive or it can be but from the same user for all products that's practically impossible so there are these kinds of patterns so what they do see is that if you like something that would sound more realistic or more fair when you have here's the good but here is not so good or here's a bit of a neutral or you know these kinds of things that's basically things that become a little bit more like believable and also I want to mention related to these let's say meronyms hyponyms the way you explained how you let's say try to process the pricing the important part in all of this is that it's really so much more about understanding the user because if they're asking like a generic question like let's say I need to paint all the walls in this apartment how much is that going to cost but the real price becomes like the square the square meters right or the like the overall amount of where it needs to be done probably the type of paint plus if you know some other additional things need to be done like if we have to remove furniture or additional stuff so these become a kind of contexts that start to affect the price in certain niches or industries you realize that these kinds of like going down and splitting everything by meronyms actually gives you a much better representation because all of these components start to strongly affect the price but also think about it from the perspective of just overall let's say like education and let's say providing a kind of answers I'll give a let's say quick example here imagine you are searching or somebody is asking for a question like there is let's say Mount Everest and like how long does it take to climb Mount Everest but now the real scenario now depends how well trained you are at what time of the year ago are you going with a big group are you going by yourself or with the guides are you trying to go as fast as possible or are you just like you know enjoying the views how much food you bring with yourself so all of these elements start to actually change the final answer that factual final accurate answer depends on these elements so if things like let's say splitting the concept into meronyms help you identify more of these concepts that's generally most likely will be more useful but a much more powerful elements on this is to actually rely on your users let's say their queries their behaviors and if you have direct communications with the clients you will actually know directly and precisely what are these things that are bothering them and that's exactly what needs to be addressed as you're providing this information that's the main logic so going back to the concept of all these names it's just a kind of assistant like a like a trick how you can try to identify those but if you already have better sources like direct communication with the customers actually use that instead of kind of being stuck on this level but of course use as much of this stuff as you can James Dooley: Yeah for sure I mean sometimes it's just it brings what's it what we found like in real-life examples is when we speak to a business owner and we try to get as much information as we can out of them they don't give us all the information they give us what they think is all the information and then when we use certain names with prompts we end ask the client what's this oh yeah that's really important because of X Y & Z well why don't you tell us and then what starts to happen it helps prompt questions to them that we would never have known about even acronyms like the same thing with that like we just it just says what does this mean and he goes oh it just means that that and that is that important if you were selling that is that important yeah it's really important because of this this and this and we're like oh you never told us that yeah yeah I know and then literally this goes on and on and on until and it's trying to extract that information from a client but be able to prompt them in a way to get the information Pavel Klimakov: Yeah right I probably should have mentioned that when I'm saying this I am let's say remembering one of the specific cases in my mind which I probably should have mentioned by myself that if I'm dealing with let's say kind of sauce company usually they have like a ticket service and if they have a proper like sales setup they're also recording all of the sales so if you go to both of these parts the support ticket area if you see what people are for what for what features they're asking or what things they have problems with that's that thing and if you go into sales pitches and sales recordings and they realize that over there there are questions like technical questions not like you know oh but this is expensive or you know can you do cheaper not that stuff specifically but oh but you know but we have the specific material or but we have that other specific condition those are the elements that I'm talking about so if you have these things somehow documented somewhere with the companies you work with that's the like the goldmine that specific thing because you're right asking directly from people in the moment they will not have it at the top of their mind it's just it's somewhere deeper and it's hard to kind of bring it up James Dooley: Yeah for sure anyone who's watching this will hope you like the episode and going through all the different names with lexical semantics Pavel an absolute pleasure Pavel Klimakov: Likewise thank you James