Sunday, 18 March 2012

Order matters with Google advanced search commands

The great thing about running search workshops is that you have so many people experimenting with advanced commands that someone is bound to spot an anomaly that you haven't. We've become used to seeing different results when changing the order in which we enter keywords but not when using advanced search commands. During one of my workshops we had a couple of people playing around with Google's allintitle command. This tells Google to look for all of the keywords following allintitle in the title of a document.

The search that was initially used was allintitle:diabetic retinopathy and came back with 277,000 results. Restricting the search to UK academic sites by using allintitle:diabetic retinopathy site:ac.uk reduced the number to about 2,190 and gave sensible results. But changing the order of the commands to site:ac.uk allintitle:diabetic retinopathy gave  two very bizarre results:


Both results are from academic sites but the allintitle as a search command seems to have been ignored. The first entry includes intitle, diabetic and retinopathy and the second has allintitle, diabetic and retinal. Using the Verbatim option from the menus on the left hand side of the results page gave us zero!

Next we tried combining allintitle with fieltype:pdf.
allintitle:diabetic retinopathy filetype:pdf

gave us 3490 results of which at least the first 100 were relevant.

Switching the order to :
filetype:pdf allintitle:diabetic retinopathy

gave 495,000 results some of which were relevant but many did not contain all of our terms nor did they contain both diabetic and retinopathy in the title. Google was also looking for variations on our terms.


Using Verbatim on this search gave us zero again.


When we looked at the advanced search screen Google had put everything in the right boxes. If we used the advanced search screen to enter our terms afresh the search worked with Google putting the allintitle command at the start of the search.

Was this a general problem or just with allintitle? We then played around with the intitle command.

intitle:diabetic intitle:retinopathy site:ac.uk - 2220 sensible results (slightly more than our original allintitle search)

site:ac.uk intitle:diabetic intitle:retinopathy - 2220 sensible results identical to those above

intitle:diabetic intitle:retinopathy filetype:pdf - 3480 sensible results

filetype:pdf intitle:diabetic intitle:retinopathy - 3480 sensible results same as previous search

We then tried using a phrase after intitle:

intitle:"diabetic retinopathy" site:ac.uk - 2130 sensible results

site:ac.uk intitle:"diabetic retinopathy" 2130 sensible results identical to previous search

Following a suggestion made by Tamara Thompson of PIBuzz ( http://pibuzz.com/) changing the search slightly to site:ac.uk "intitle:diabetic intitle:retinopathy" gave exactly the same results.

Just to make sure that it wasn't just us in the UK seeing this I asked fellow members of AIIP (http://www.aiip.org/) to run the original two allintitle searches. They saw exactly the same thing.

Its seems, then, that there is a problem when allintitle is not the first command in a search. The intitle alternatives appear more reliable. If you prefer to use the command line rather than fill in the boxes on the Advanced Search screen remember that order sometimes matters.

Does this affect other combinations of commands? I left it at allintitle and intitle but I wouldn't be at all surprised.

Wednesday, 14 March 2012

Use more than Google

If you need more evidence - other than me telling you! -  that you need more than Google then take a look at The Disruptive Searcher (Sanity checking Google http://disruptivesearcher.wordpress.com/2012/02/27/sanity-checking-google/):

"if I hadn’t searched across more than Google for data on a small, new company that I was asked to research recently, I would have missed out on some very significant information that Google just wasn’t showing me."

So take a look at Bing (http://www.bing.com/), DuckDuckGo(http://duckduckgo.com/) and Blekko (http://blekko.com/) for starters. The Disruptive Searcher also mentions Dogpile (http://www.dogpile.com/), which combines results from Google, Bing and Yahoo.

Friday, 2 March 2012

Google+ overrides search settings

I've noticed this strange behaviour for a while but have only now had to time to try and find out what is going on. When I'm signed in to a Google account that has Google+ associated with it I cannot display more than 10 search results per page. This is despite having Instant switched off and specified 100 results per page in my search settings. I checked another Google account that does not have Google+ and my settings are respected as they are when I am signed out. I have cleared cookies and web cache, and tried different browsers. The same problem occurs. So it seems that if your account has Google+ associated with it Google overrides what it likes! I already have two browsers open all the time: one signed in to my main Google account for gmail and other stuff, and one signed out for search. Sometimes, though, I want to run the same search within Google+ and then on web search so I have to go to the effort of copying my search across to the signed out browser.


This may sound like a minor issue but if Google is ignoring this user setting one starts to wonder what else it is choosing to ignore.

Thursday, 1 March 2012

Clear your YouTube history

Now that Google has merged all your personal data, time to double check that you've removed stuff you do not want Google to use. As well as the steps mentioned in my previous posting (Google personalisation: web history isn’t the only problem http://www.rba.co.uk/wordpress/2012/02/22/google-personalisation-web-history-isnt-the-only-problem/) you might want to clear out your YouTube history as well.

Sign in to your account and go to  http://www.youtube.com/my_history  and  http://www.youtube.com/my_search_history. The first is your viewing history and the second your search history. If you don't want Google to use this information to "enhance your search experience" in its other products clear both of the histories and pause to stop it gathering future activity.

And another reminder that if you are fed up with Google trying to personalise your ads you can opt out at http://www.google.com/ads/preferences. Note that this does not get rid of Google's ads altogether it just stops Google using your past searching and browser behaviour to decide which ads to display. To opt out of targeted advertising from other networks you should also pay a visit to  http://www.aboutads.info/choices/ and http://www.networkadvertising.org/managing/opt_out.asp.

Update

I should have said that the opt-out of targeted advertising is done via cookies on your computer and so is computer and browser specific. If you or your system periodically clears cookies then you will have to opt out again. You do not have to be signed in to a Google account to do this. There is also a  Keep My Opt-Outs plugin for Chrome.

Further update

I've just checked a very old Google account of mine that was used for news alerts but does not have a Gmail account or anything else associated with it  (it was set up pre-Gmail). I can log in to YouTube with it but when I try to access the My History page to delete search and view history it insists that I set up a channel, which I do not want or need, before I can do it. Tell you what, Google, I'll just delete the whole account.