Forum Moderators: open
<mumbles>Arrgh, gotta go back and put city, state, zip on the couple of hundred money pages that I just finished updating at 4am<mumbles>
Easiest way to spam?
Based on the results I saw earlier, this is going to need some work. How involved will Applied Semantics be?
Or if you wanted to see places that had Guinness on tap in New York.
...well for that you could just ask me : )
time to get my database of zip codes out ;)
put city, state, zip on the couple of hundred money pages
Easiest way to spam?
As several here have already pointed out, sites will now need to put some geo marker stuff in their pages then hope they get spidered.
Even then, as many have said, one wonders how much more useful Google geo-search will be than a local Internet yellow pages already is.
Investing millions to draw maps with a few new pizza parlors? Cute, but what does this do for search?
What GG does *not* say is how limited their concept of geo is. They demand you think in terms of city/state, for example, like any online Yellow pages does.
Google is clearly thinking from the point of view of a college student looking for nearby pizza, not a person who needs a web designer from somewhere in the same general area. A family needing furnishings for a new home wants the best place to find it in that area, without being limited to an arbitrary radius.
The first true combined web search/geo search tool solves this dilemma with a more searcher-friendly method.
In the original such tool, as with this new Google feature, one can enter key terms, and the geo targeting info. But in the first tool, if one enters a ZIP code, tel area code, or ANY town name, the engine is smart enough to find where you want to search, and goes on to find *all* matches in the surrounding metro area. NOT just in a fixed-range diameter (which of course is very simplistic in complex or large metro areas, especially those which cross state lines.)
As to the design of the Google answer to Overture, there are a dozen ways to plot physical locations, and if one has the cash, like Google, one can then pay Mapquest or some other company to supply little maps.
But what Google and most of the posters in this thread are overlooking is that simply adding address info to a webpage does nothing for the great majority of geo-oriented web searches.
This is not any “fault” of Google, or any other crawler engine: spidering, by itself, can never accurately determine how a web page relates to a physical location.
One needs a mechanism where the website owner can simply tick off the geo-markers that ought to apply to his or her site or page. If the site describes a physical location, then a ZIP or postal code is great. But if the site is simply *about* the area in question, address info is meaningless. EG, if I have a site that is *about* travel in France, I need to associate that site in a rational way with potentially hundreds of areas, not a few addresses. In the conventional search engine operator mind-set, this would seem to many like SPAMming - but in this instance it is useful to the searcher to find this site no matter which French geo target she happened to enter.
The first true combined web search/geo search tool works this way already.
And then, the engine needs to be clever enough to “know” that if a searcher wants to find information about, say, Indian handicrafts in Taos, New Mexico, she may want related content on the web that has nothing to do with that specific location.
The first true combined web search/geo search tool works this way already.
I am sure there will be many more improvements and refinements in these tools, for, as several have noted, geo-search is certainly “hot”.
Glad to see you are trying to catch up, Google, but you have a ways to go yet in this particular space.
They demand you think in terms of city/state
Would you like lat/long better? ;)
SN
Compare [research.overture.com ] to the cached snapshot.
No thanks! Latitude and longitude for Mr and Ms American consumer? LOL
(See my example of how the first tool defines a metro area, for example, *without* relying on simplistic L/L centering.)
What I am saying is that a friendly geo search lets the user supply a wide variety of target terms, in less rigid forms, and then does what it takes to pin down the target area, if it is not immediately evident.
The first combo website and local area search tool does this.
It also then *remembers* the user's geo area past searches, so it can default to a "home" area, for example, when the user wants to turn the geo dimension option "on".
All I am saying is that when we add the geo-dimension to web search the complications are massive. From a search user's perspective, reducing web search to one's local area is a great goal. Some have worked on it for years, and their solutions thus reflect the true complexity of the problem.
The first Overture and Google attempts do not deal with the full implications, that's all. And cannot, IMO, so long as one relies on spidering.
first true combined web search/geo search tool
Which one are you talking about? Or you saying about a hypothetical one?
But this is the Google News thread, not a thread for discussion of alternative tools. I only mentioned the other site to give the forum participants some specific ideas of how hard this aspect of search is, when faced fully, and how limited the Google and Overture attempts are, to date.
Incidentally, for the thread’s general info on the hot geo-search area:
To see some of the business as well as technical implications of web-based geo search you may be interested in the upcoming conference on the collision of web search and local-area search like the IYP which will be conducted by the Kelsey Group.
Check out the agenda topics at [kelseygroup.net...]
“Digital Directories & Interactive Local Media Summit”
(No, I am not advertising for this conference - could not afford it anyway! - LOL - But to see just how “hot” this area is, all of a sudden, look at the presenters and the attendees...)
The search results gave me Miami Hotels, a furniture magazine and several other irrelevant and unsatifying results.
Great idea, but I think it needs quite a bit of work before it is of much use.
Easiest way to spam?
Not at all.
One of my sites lists about 300 widgets by street address, however, the city and state are understood to the site visitor and are only included in titles and heads where appropriate.
From what I've seen of the Google search by location results its indexer is looking for recognizable city/state combinations or a zip code. Why should I make its job more difficult?
Geo-search is developing and this is a first, simple step to make pages more friendly to Google's product. Why not?
OK, so, what does this key on? City/state or zip info in the page itself or some kind of <address> tag in the head or something? Will zip code, alone, be sufficient to get it to work (for those of us who don't want to be posting our organization's street address online for whatever reason)?
Yah. I couldn't agree more with ibizwiz and others that there are many ways to improve this first demo. But I think Googlers were playing with this and they realized that this geosearch demo already lets you do new types of searches that you couldn't really do before. Around here for example, it's easier to do a distance-based search than trying to do a search like
"salsa lessons" "mountain view"¦sunnyvale¦"palo alto"¦"santa clara"¦milpitas. You get the idea.. :)
As far as the future, my hunch is that just like any feature, the amount of effort that we put into will depend on the interest and utility that we see for users. I'm pretty encouraged from the interest that we've seen so far. :)
...still, there has to be a way to precisely define a 'location' - Zip codes are more or less pointed but ... sounds like a great challenge for Google. Any way I can participate? Would love to help if I can.
the amount of effort that we put into will depend on the interest and utility that we see for users
Methinks there might also be a few guys with eyeshades and adding machines in the back room totting up adwords bids for "hotel 90210". Hoo boy, however many zips or communities exist I guess that's how many virtual 'mini-Googles' would be running around out there, all with ad space for sale.
What a change for the marketplace that would be, geo-search with contextual ads.
I think this has been in beta for some time. I recall seeing it perhaps a month ago or so...
I swear I brought up this idea in the Google Suggestions for 2003 (?) thread, but can't find this discussion-- how can I crow about it without anything to point to?!
Unfortunately, when I ran a few searches in the city I live in, plus other cities in the area, the results were far less than awesome. For example:
The top hit on "recreation" in my city was the nearly-useless cookie-cutter directory page for a nearby city. Seven of the top 10 results were links to the city website (this is a good resource on local recreation, but not if you consider that it's just one site and most of the hits were to minutes of city council meetings). Most of the results listings included "related" links to national or regional "directory" sites with a page for every city. I also got hits on low-value pages deep in the US Masters Swimming site and a national restaurant-listing site.
When I searched on "methodist" my results were somewhat better, in that the results were mostly listings for institutions with "Methodist" in their name. However, the top 10 results for my town included 3 listings for the local Methodist-affiliated hospital (with related deeplinks to other hospital units), two listings for one local Methodist church (there are 3 Methodist churches with websites, but the other two churches did not appear in the top 10 results), a listing for a radio station in a nearby city, and several listings for Methodist churches in cities in the surrounding region (including one church that's about 100 miles away). The website for the distant church was a particularly content-rich site (probably good if you really like Methodist churches), but the other non-local church links were to yellow-pages-level directory listings. "Related" links included several obituaries from the local newspaper and the homepage for a local public elementary school.
This may have promise, but it hasn't delivered yet. ;)
I liked your example of "methodist"--I assume you were looking for churches, not hospitals? Just to pick a city at random, let's use pigeon forge, TN. If you search for "methodist church" nearby that city, you get higher precision than if you just typed "methodist." I wouldn't claim that it finds every methodist church in that area, but it does find a nice local directory site that you can then find some really good answers from. So the search result doesn't immediately give perfect precision and recall, but it does point you to some resources that you might not find just poking around by yourself. I would say that local people definitely can find places where the recall is lacking, but it already provides a nice solution for people outside the area to start with their research. Applying old power-searcher tricks such as tacking on a "united" or "first" to the methodist church query lets you possibly unlock more potential results too.
Anyway, I enjoyed your post. I think if people have the right expectations (right now it can't replace a local person's experience, but it can be better than what an outside visitor would find on their own), it can be pretty fun to play with this demo. :)
Your search for Pub near Mountain View CA did not match any documents within 15 miles.
How do you survive their, Googleguy ;)
Each system, Google's and the geo techniques have their advantages (and weaknesses).
I'd love to see them integrated somehow.
It would be great to have this return lat/long co-ordinates and/or street addresses
as well as an interface to the Google api (or xml feed).
I mean, if I'm living in Tsunohalia and I want to buy Yanahootic widgets, I can search a) all over the world because I don't care about price or shipping b) In Yanahoot because they of course will have a larger selection or c) in my Tsunohaliastic neighbourhood because I need them NOW and/or I like to see what I buy. In short, page location will bring web shopping back to resembling normal human 'shopping behaviour'.
