Synonym searching with Google
This morning we received the final article in Glen Cathey's Semantic Search series. If you haven't read it you should stop reading this right now and visit http://www.booleanblackbelt.com/
But make sure you come back because otherwise I'll get all lonely and ignored.
Ok. Sitting comfortably again?
Glen uses the tilde to great effect in his example strings, -~job -~jobs. This, combined with a thread on Recruiting Blogs, had me reaching for the keyboard (found under my fingers, no real surprise there), to spend a bit more time trying to understand all I can about synonym searches via Google.
Although there are a few sites which discuss the operator there don't seem to be a lot of definitive answers out there.
So I throw in ~job -job to show the synonyms which Google uses and hope to be able to work backwards.
By collecting the bold terms and removing them from my initial search I built up the following list: career, careers, employment, job, jobb, jobs, recruitment, resume, salary, vacancy, work.
Resume was an interesting term to find, but a search of resume -~job returned 79,000,000 results indicating that specifying a term cancels out the removal of it as a synonym.
Out of interest I changed the search to resume -job removing the synonym search which returned an additonal million and a half results. That's a lot of noise I don't want to hear in my searches.
Next I ran the same search with the term -~jobs to see how much crossover there was; employment, career, jobs, job, newspaper, positions, recruitment, vacancies, work.
I love that vacancy is pluralised when you pluralise the keyword, a very nice piece of design.
Comparing the pools, the search for resume -jobs -job retrieved 68,100,000 results where resume -~jobs -~job returns 66,800,000; a reduction of 1,300,000.
The results show that you can remove a lot of results from your searches and that most of those results will be nothing more than noise.
Having said that it's always going to be worth knowing what terms you are removing; imagine you're searching for a print journalist and use -~jobs without specifying newspaper, there's a chance (and I'm not checking it in case this example doesn't work and I have to think of another one which will only annoy me) that you'll miss out on candidates who have used newspaper journalist instead of print journalist.
Just as a point of interest, I put job and jobs into Google Sets to see if it uses the same set of synonyms. There is certainly cross over but Google Sets gives a lot more variation.
Perhaps one day we could see an option to specify the Sets synonyms in the standard search via a specific operator. If we ever do, I hope they use `, I hardly ever use that key...
But make sure you come back because otherwise I'll get all lonely and ignored.
Ok. Sitting comfortably again?
Glen uses the tilde to great effect in his example strings, -~job -~jobs. This, combined with a thread on Recruiting Blogs, had me reaching for the keyboard (found under my fingers, no real surprise there), to spend a bit more time trying to understand all I can about synonym searches via Google.
Although there are a few sites which discuss the operator there don't seem to be a lot of definitive answers out there.
So I throw in ~job -job to show the synonyms which Google uses and hope to be able to work backwards.
By collecting the bold terms and removing them from my initial search I built up the following list: career, careers, employment, job, jobb, jobs, recruitment, resume, salary, vacancy, work.
Resume was an interesting term to find, but a search of resume -~job returned 79,000,000 results indicating that specifying a term cancels out the removal of it as a synonym.
Out of interest I changed the search to resume -job removing the synonym search which returned an additonal million and a half results. That's a lot of noise I don't want to hear in my searches.
Next I ran the same search with the term -~jobs to see how much crossover there was; employment, career, jobs, job, newspaper, positions, recruitment, vacancies, work.
I love that vacancy is pluralised when you pluralise the keyword, a very nice piece of design.
Comparing the pools, the search for resume -jobs -job retrieved 68,100,000 results where resume -~jobs -~job returns 66,800,000; a reduction of 1,300,000.
The results show that you can remove a lot of results from your searches and that most of those results will be nothing more than noise.
Having said that it's always going to be worth knowing what terms you are removing; imagine you're searching for a print journalist and use -~jobs without specifying newspaper, there's a chance (and I'm not checking it in case this example doesn't work and I have to think of another one which will only annoy me) that you'll miss out on candidates who have used newspaper journalist instead of print journalist.
Just as a point of interest, I put job and jobs into Google Sets to see if it uses the same set of synonyms. There is certainly cross over but Google Sets gives a lot more variation.
Perhaps one day we could see an option to specify the Sets synonyms in the standard search via a specific operator. If we ever do, I hope they use `, I hardly ever use that key...
About
© 2009 Created by Andrea Mitchell on Ning. Create Your Own Social Network
