Wordnet & SEO - Using the lexical database to research

<removed>

very interesting stuff, rand. i'll be playing with this for a while. thanks!!!

You can tell they're looking at the senses of the words (and combinations), only in an effort (maybe) to get the bigger picture.

That is interesting, because it could mean that I'd better trade links for my table tennis site with sites about dominoes (which is also played on a table) instead of badminton (which is in fact a highly related sport in real life).

If anything, it shows that algos don't think as humans do, even though they aspire to.

Thanks for the article randfish... You don't happen to know of a Dutch version, do you?

Wit,
You are somewhat correct, although my guess is that logically speaking, both are equally valuable. Where something like Wordnet becomes really valuable is when it points out sites/themes/pages that you would never consider getting links from.

Seriously, though, just spend ten minutes entering your keyword phrases and see how much Wordnet knows about the topics, the hyponyms/hypernyms/etc. It's fascinating stuff and it really can give you an insight into where search is or how it might progress over the coming years.

Rand, THIS is a hugely beneficial tool! Sure beats doing it manually.

Another one of your great posts I have printed out for "learning" and "research"


Don't know about anyone else, but by golly I sure do appreciate the expansion of my own personal "future" understanding of word relationships. (since I have no clue right as of about now...all the Research papers I read last week have my brain fried)

Cheers

Hey gang,

I just finished a terrific research paper from the Czech Republic (where I once lived for 6 months back in college). www.cs.cas.cz/semweb/download.php?file=05-01-Snasel-Moravec-Pokorny&type=pdf - Here is the PDF.

It is exceptionally technical in some areas, but I was still able to pull out some very important concepts to share with you:

1. LSI/A, which I have been somewhat dismissive of in the past is talked about in-depth in the paper. The engineers note that WordNet's database can be applied in place of some of the more computationally expensive parts of LSI, making the use of it more practical and possible for commercial search engines. (Although, based on the writings of the 3 IR guys at SEW, I still think LSI/A is an unlikely candidate for use).
2. A EuroWordNet does exist with the oncologies in Dutch, Italian, Spanish, German, French, Czech & Estonian. http://www.illc.uva.nl/EuroWordNet/
3. The hypernym "ontology" is discussed as a method for structuring topics and themes. Once again, this is a great way for SEOs to see how their keyword phrases/terms fit into a computer's 'understanding' of the language and apply it.

They give an example of a topic structure built through WordNet:

> Psychological Feature
> Cognition, Knowledge, Noesis
> Content, Cognitive Content
> Knowledge Domain, Knowledge Base
> Discipline, Subject, Field
> Humanistic Discipline
> Philosophy

If I were to create a new website, no matter the type of client, I would use this kind of careful construction of topics to build the architecture for the site. The internal link structure and internal anchor text alone would give the kind of on-page/on-site boost that would give me a serious edge up on the competitors.

I think that this kind of application based material is tough for many SEOs to understand and apply right now. But, as you read more and more, you can pick this up very quickly. Good luck to everyone, and please, do play around with WordNet sometime - it's a great learning experience and it might even help you jump up a few positions in the SEs.

Thanks for the Dutch link randfish. I came across a couple of other Dutch WordNet derivates, but they were mainly concentrating on the 'popular dictionary' application and very limited.

Since the English "original" is very comprehensive, I'll be better off just translating... unless I want to build a site for "measuring instruments" (hmmmm ;) I might...)

A quick note from the researcher...

we've been using wordnet for a very very long time (1985) as its a machine readable dictionary and thats kinda what it was designed for. CyC is along the same lines as this as well.

I wonder why wordnet came up all of a sudden and why no one found it before.

It groups synonyms into synsets. That way it can produce semantic similarities. It keeps distinguishes between nouns, verbs, adjectives and adverbs in different categories. The problem with it is that it isn't very good at irregular verbs and the database isn't very fresh because its hard to keep up to date.

All you foreign language speakers, don't forget you can use Euronet!

Xan -

I'm so glad you're here - now someonw who really knows what they're talking about can critique my posts. Why haven't SEOs used wordnet in the past?... That's a good question and I don't have a good answer. Maybe SEO in the past has never been difficult enough to warrant switching from the status quo of techniques. Hopefully, the future will change that.

Hi Xan, glad to see you made it.

Thank you for the welcome all!

I think maybe people using wordnet suddenly must be because of a post or something, perhaps one of mine or someone elses, but one thing for sure is that SEO's even paying attention to it means they are changing.