Y!Q: Adding Context to Search
Everytime we launch a new service, someone asks where the idea came
from and why we did it. There’s a good story behind this one, so I
thought I’d write it up here.
A bit over a year ago,
UK during the Christmas holiday. It was the
of the Tears
for Fears song Mad World. This caught his interest and
he wanted to know more.
Why is this song so popular? Where can I hear it? Is there a video?
Can I buy it?
He spent the next 30 minutes searching for answers to those questions and
more.
It turns out that the song’s popularity had a lot to do with the movie
Darko. The movie became a cult hit and the song was on the
soundtrack.
But I digress..
As a result of this experience, he posed a challenge to the Yahoo!
Search team: to build technology that makes it possible to accomplish
tasks like this in just a few minutes.
Not long after that, the team got wind of some contextual search
technology that Reiner
Kraft was building. Now, Reiner is one of our resident geniuses.
When he was at IBM, there was a patent attorney who did little more
than handle Reiner’s inventions.
him in their TR 100, a
list of technology innovators under the age of 35.
Reiner had realized that the current method of searching isn’t always
the most efficent way to get what you’re after. Most people aren’t
skilled in the art of choosing exactly which keywords to use when
searching.
Reiner’s technology was designed to help eliminate that problem. The
fundamental idea was to supplement search queries with
context. So instead of having to spend a lot of time
searching and assembling all the information you’re after, this
contextual search technology could incorporate that context (the stuff
you were reading at you moment you decided that you wanted to know
more) to find the most relevant results.
The team had a look at what Reiner was doing and immediately realized that
it could be used to meet Jeff’s challenge. They asked for a few
tweaks and two days later, Reiner had a working prototype. Excited by
what they saw, the team asked what it’d take to turn it into a full-blown
service. In no time Reiner had the help of some of our best product and engineering folks, as well as one of our DHTML wizards.
What they built is
Y!Q is a cool DHTML module that embeds contextual search directly into
a web page. We’re
in a test environment on Yahoo! News, but
can embedY!Q into their content (that embedding process is currently a bit klunky, but we’re working on that).
Y!Q is also available through the Y!Q
DemoBar, an Internet Explorer toolbar that brings Y!Q’s
contextual search functionality to any web page. Simply highlight
some related text on the page you’re reading and then perform a
search.
If you’re a Firefox user like me, don’t worry. You can also
Firefox.
In each implementation, Y!Q uses the context to help bridge the gap
between query and intent. This should help turn some of those 30
minute affairs into the 2-3 minute tasks they ought to be.
Give Y!Q a try and let us know what you think. It is
a beta product right now, so we’ll likely be tweaking things in the
near future.
Jeremy Zawodny
Yahoo! Search
P.S. If you’re wondering about the name, it’s play on “IQ”.
Knowledge often comes from combining information with the
relevant context, so it seemed like an appropriate name.
This is Awesome, i’m already lovin’ it :) I think this can help the users in missing search puzzle, search contextual, learn more, validate information.. Y!Q. thanks Y! folks
I was quite excited when I first read about this. Contextual search is something I’ve wanted to see for a while, i.e. the search engine looks at what comes back from the text search and makes guesses at what you’re looking for. An excellent example being Jeff’s search on the Mad World single.
But after playing with it for about half an hour I’m disappointed. Maybe it’s because I’m trying it on Linux (Konqueror and Firefox), so I don’t get all the whiz bang features. To me it looks like it does two things:
1. Take a web page and turn it into a set of meaningful search terms to find related content.
2. Take a search query and turn it into a meaningful set of search terms.
The first feature is great, but it looks like it takes a lot of effort to be able to use it (install plugins etc.). The second feature is less useful, but that’s probably because I’m pretty good at writing search queries in the first place.
The actual search results is what disappoints. I tried the Yahoo News demo, and clicked on the first story (Bush Sells Plan to Remake Social Security) and it derives the following search terms:
- president bush
- nest egg
- government benefits
- personal accounts
- term president
- campaigning
- social security
But all the search results are the same news story from different sites. Not hugely useful, I was hoping for some related content, maybe a profile on Bush, or what the current social security policy is, etc.
Next I try Jeff’s example and search for ‘Gary Jules Mad World’. The search changes it to ‘”Gary Jules” “Mad World”‘, which is good. But again the search results disappoint. There are a lot of links to lyrics, even one to a free MP3 download of the song (off someone’s personal university web space, legal?). The most useful was the Yahoo shortcut to the artist page and link to buy CDs. But that link only presented me with the choice to buy the album and not the single. So only one of Jeff’s original four questions was answered.
It’s a nice idea but looks like it has a long way to go. I think once the system can also attach semantic meaning to the search terms, this sort of thing will become a lot more useful.
hmm…how about adding a way to display the users searchstream as well? but yeah, as venki said, this is awesome :)
I don’t get it :-(
I added bookmark keyword for Firefox, it searches nicely, but it’s just a search. I don’t see the ‘contextual’ part. How this search is related to the page I’m reading?
I understand that a bookmark can’t get this information by itself, so may be this way provides only limited functionality. But this is an official way of adding Y!Q to Firefox and I wonder may be I’m missing something….
Now I wonder if its possible to take the links clicked from the related cluster, and re-build the clusters again, applying the user drive n data as another layer of social relevancy.
This way user driven data can be used to refine the initial query set based on what is actually popular e.g. Yes the DVD & MP3, but not the picture or poster etc.
This would reduce the margin of error, and could also be used to re-define the initial algo that extracted the clusters from the initial result set.
Now for a DSL broadband user on a fixed IP, you could show related Local services…..
A further use could also be to add orientation into the equation (e.g. mobile) and the related cluster can change at to geo location of where you are.
For the specific example of Geoff, at home he could also be told which of his local stores he could be the album, price comparisons, or potentially even better, whilst he was travelling work.
Alternatively he could you call the UK office and send someone out to buy it for him……
Well done guys, let mne know if you need any more ideas :)
There’s a little mistake here – when the search string is too long, the error says, “Contextual Search disabled because your query is longer that the supported length.” ‘That’ should be changed to ‘than.’
The results seem ok so far!
I have been playing with it today and I thought Iwould give my feedback. First of all, blinkx did this a while ago now and it is much more automatic. Y!Q involves a lot of manual search and I never seemed to get linked information. I just dont get it, I highlighted a section and it brought me to irrevlant infomation.
I don’t really see how this search differs from any other search. If it is contextual, isn’t it supposed to already understand what I am looking for? Why must I highlight and then refine the search by adding additional terms in the search box? It seems rather more difficult than helpful.
Great work guys! YQ is much more flexible than the Blinkx tool that Sarah mentions and gives more relevant results.
I like what ID:entity is proposing too. Localization and contextual relevance pulled together and delivered in a personalized manner appropriate to your current access device. Yahoo’s ability to call on a mass of media content should serve it, and us, well here.
Hey I had a better idea and I suggested it to some one in the text analytic group at Y! India who was interviewing me for a position.
It goes like this.
Let the user enter his search for the first time. He gets some guery.
The SE returns some link. Now the user is going to go to only those links that he finds more or less relevant from the synopsis of the page. I do the same :D
Keep tracking the pages he has gone to. Extract keywords out of the pages the user has gone to. Extracting keywords from the synopsis (of the cliked links) would also be a nice idea.
All these keywds + the original key words make up the context to refine the search. Now pop up sumthing showing u have betters results to return. Search using the modified keywords. and retrun the resutls.
Voila, i dont have to surf thru 1,000 links to get some info.
How does that sound Jeremy ?
Jeremy / VibHanshu
the tools that Vibhanshu, is talking about above has already been developed at MIT in prototype form many years ago. They published a paper on reconnaissance agents. There were two kind if agents local, and global.
The local agents were a product call Letvia or something similar, they scanned the page you were reading and looked ahead at this links on the page, and recommending other pages ahead (based on the keyword analysis within the inverted index (e.g. how content match / ad sense) already works.
Global on reconnaissance agents would actually query Yahoo! regarding the pages you were viewing in your browser and come up with suggested links, this is more in kind with Y!Q.
The name of the global reconnaissance agent was powerscout.
The paper can be found here (No 1 result)
http://search.yahoo.com/search?p=reconnaissance+agent+%2B+MIT+%2B.pdf&fr=FP-tab-web-t&toggle=1&ei=UTF-8
I cant get it to work! I downloaded it and am highlighting, but it never seems to bring me back what I am looking for. Also, what is the difference between highlighting and typing it in? I need something that will just search for me.
Daisy~ I don’t agree. I find that Blinkx is way more flexible than Y!Q search. I have not YET received any relevant information from my searches. I even went ahead and compared the two engines. Blinkx was much more intuitive AND, i got back relevant data. Maybe you just don’t know how to use Blinkx. If you did, you’d agree with me.
Will it be possible to add site specific search to YIQ.
I see this as a hack to show all my related posts of my blog in a post.
Thanks,
Amit Agarwal
http://labnol.blogspot.com
Hi Daisy, it sounds very presumptious of you to state the just because I don’t agree with your opinion that it must be because I “just don’t know how to use Blinkx”. How arrogant you are! I found blinkx to frequently return irrelevant results. Coupled with its inflexibile approach to searching my conclusion was that is was pretty hopeless. Despite drawbacks, to be expected in a new test release, Y!Q by comparison seemed a breath of fresh air.
You guys at Yahoo! keep up the good work. With all the recent new features you’ve added, Yahoo Search is getting better by the day. Thanks!
I added your new search Y!Q on my site for example see this page: http://www.fafzali.com/ttak/archives/000065.html but at results i saw unknown characters
why?
thanks
bye
Anyone remember or heard of FizzyLab … a Seattle-based company, in operation back in 1999-2001?
Content Relevator … Ad Relavator? OK, may be the product names weren’t exactly cool, but the products certainly were. Content Relevator was Y!Q and Ad Relevator was Google’s AdSense.
I do love Y!Q but one think is odd to me.Why yahoo given browser button for it.They could simply add it to toolbar.That woyld be easier.
I think this clustering one should be a good start http://www.topgist.com
how can i add a yhoo search on my webpage