Google Blogoscoped

Forum

Restricting the Search Space  (View post)

Elias Kai [PersonRank 10]

Friday, June 22, 2007
17 years ago4,747 views

An additional restriction:
Search within a file format ? any terms + .pdf or .doc
Video search for funny videos
Use it in Google.com
site:youtube.com fun

Search within a certain country domain extension:
site:.es online
site:.co.uk online

James Xuan [PersonRank 10]

17 years ago #

<<Search within a file format ? any terms + .pdf or .doc>>
This can be achieved with http://exalead.eu

Matt Cutts [PersonRank 10]

17 years ago #

Puzzle for the day:

Do a Creative Commons search on Google. Examine the resulting url closely. Can you discover something that Google's Creative Commons search can do that (say) Yahoo's can't?

Ionut Alex. Chitu [PersonRank 10]

17 years ago #

Google includes only pages that link to a CC license and use a rel="license", while Yahoo includes all the pages that link to CC (like Google Blog: http://search.yahoo.com/search;_ylt=A0geu7_7_3tGIREAk.hXNyoA?p=google+blog&fr=moz2&vm=p&cc=1).

Philipp Lenssen [PersonRank 10]

17 years ago #

Google's "as_rights" parameter seems to accept a list of inclusions/ exclusions of licenses you want to search for, like:
(cc_publicdomain|cc_attribute|cc_sharealike|cc_
nonderived).-(cc_noncommercial)

Perhaps this can be mixed freely... Yahoo on the other hand has a boolean "cc" parameter that is then mixed with an array of "ccs" parameters. Hmm, I guess you can't specifically mix this with your own exclusions like you can with Google?

TheRaveN 2.0 [PersonRank 1]

17 years ago #

Take a look at this Google Advanced search Mock-up:

http://www40.brinkster.com/israelpics/Google_advanced_search_Mock_Up.htm

TheRaveN 2.0

Philipp Lenssen [PersonRank 10]

17 years ago #

Hah, I love that mockup. I think I'll try one too.

Rue [PersonRank 1]

17 years ago #

Until they offer the ability to restrict your search to a certain topic, all these other features are pretty lame.

What do I mean by "a certain topic"? I mean that if you are interested in "automobiles", you should be able to specify that. And then the engine will automatically insert all those domains, data types, etc. that are relevant to that topic, and automatically restrict your search to those areas, without you having to manually specify it.

You could argue that Google personalized search is already trying to do this. When I type "jaguar", and Google personalized has already detected that I look at a lot of car sites, Google will then restrict my results to the "car" topic.

The problem is that there is no way of specifying the topic manually. So if -this time- I want to search for jaguar the animal, and personalized search does not peg me as an "animal" person, the only way to do it is to manually go in and add every single domain and document type myself. Again, that's pretty lame.

Stan James [PersonRank 1]

17 years ago #

One more addition is to search through a **personality**.

That's what we do here at Lijit:
Bundle up all the content sources of a person (blog, bookmarks, photos, diggs) and that of their social networks (blogs they read, mybloglog communities, delicious fans, etc..) and roll all that into a search.

E.g. search VC Brad Feld about "term sheets":
http://www.lijit.com/pvs/bfeld?q=term+sheets

We automagically discover the content sources of a blog; you can see it chew on Fred Wilson's blog here:
http://www.lijit.com/signup/content?uri=http%3A%2F%2Favc.blogs.com%2F

A social network is the oldest vertical search there is. :)
My post about it: http://wanderingstan.com/2006-12-26/personal_network_search

-Stan

Philipp Lenssen [PersonRank 10]

17 years ago #

> What do I mean by "a certain topic"? I mean that if you
> are interested in "automobiles", you should be able
> to specify that.

Google actually once had a tool out there that did just that (they removed it since, though perhaps their personalized search efforts of today implicitly go in this direction):

http://blogoscoped.com/archive/2004_03_29_index.html#108055392361730949

Rue [PersonRank 1]

17 years ago #

Philipp, it appears that the service is still there: "www.google.com/alerts"

But it does not appear that they are really doing "topic" selection. Yes, you can enter a few keywords, and be alerted to updates to any page that contains these keywords. But that is simply the notion of "push" (rather than "pull") web search, aka document routing and filtering. You're still just getting back pages that you asked for, in the keywords. You're not actually doing any conceptual topic detection.

For example, what if I typed "jaguar" into this Google alerts service? Would I get updates related to the car or to the animal? It's unclear. Because you're not actually specifying a topic; you're specifying keywords. Do you see the difference?

Philipp Lenssen [PersonRank 10]

17 years ago #

Rue, I should have clarified; what I was talking about was not the Web Alerts feature, which still exists, but the Personalized Web Search feature of 2004, which doesn't exist in this form anymore. Back then you could select a couple of topics of interest from a list of categories. E.g. Technology -> Science. Then, the results of this special search – not the default web search though, it was neatly separated :) – could be shifted using a slided (minimum to maximum personalization). While there's a screenshot of the search result with the slider...

http://blogoscoped.com/files/personalized.gif

... I can't locate a screenshot of the categories selection page anymore...

Rue [PersonRank 1]

17 years ago #

Ok, I understand you now; thank you.

The question remains, though: Even back in 2004 when you could do this topic selection.. were the topics/categories themselves manually created? Or were they automatically generated from the data? My guess is that the topics were more Yahoo-like, i.e. manually created.

That is not as interesting as it could have been. More interesting would be to have automatically generated topics. Moreover, it would be even more interesting to first type in a query, have it generate a number of topics to which this query belongs, and -then- let you further restrict your results. If these topics are generated on the fly, then you'll get a very different (more targetted, more relevant) set of possible domains on which you can restrict your query.

It's a shame Google never went in this direction.

Forum home

Advertisement

 
Blog  |  Forum     more >> Archive | Feed | Google's blogs | About
Advertisement

 

This site unofficially covers Google™ and more with some rights reserved. Join our forum!