Forum Moderators: open
This is happening more and more to me. The problem is that I've "tapped out" certain pet keyword areas of interest. The top 10-20 sites always return the same core sets of sites regardless of keyword variation.
An example: I've been doing a project for a client that turned into a project for me (plug:[webmasterworld.com...] ).
The site I was working for was a split linux/windows shop and needed something that would run in both with little problem. So, perl to the rescue. And oh by the way - could you make it gui with an install so I can distribute it to my affiliates, while you are at it?
So, enter the brave new world of perl tk cross platform gui programming. It was a steep learning curve. This should be easy for Google right?
Not so grasshopper. Regardless of the keyword variation, there was a group of core sites that just dominated the top 10. They were authoritative - yes. They were quality sites - yes. Did they have the info I was after - no.
Many times during that process, I did not find what I was looking for in the standard Google se. It was a process of trying to mentally/visually filter out those same sets of sites search after search (300-500 a day). I did go to other search engines to find what I was looking for. On the flip side, I did find some great starting info in the directory and some specific help in the Usenet groups. Unfortunately, the onfocus POPUPS (yes it pops up) Google is running in the groups, about drove me insane using it. It completely disrupts surfing patterns. That is especially true when Google usenet is having a slow day and you push 2-3 windows into the background to get some work done in the foreground.
Resolution:
Personal Filters. Send me a cookie with "-foo.com -bar.com -here.com" where there are default excludes applied to every search.
Personal Filters. If I've seen page X on three previous searches, then filter those out.
The other frustrating thing is when you want some information eg stainless steel tanks versus 'plastic' tanks - trying to search for something like this is a nightmare!
Of course trying to design a search engine that delivers this is no easy task. But if you could - what a step forward it would be.
Exactly. That's why I find that new Teoma option to "refine search" to be so powerful and time saving. It just relives you of all that hunt and peck guess work.
I have noticed that searches on technical data related to Perl, PHP and MySQL all tend to take 2 or 3 pages on average to find useful matching sites. I am not particularly good at writing queries - I just dump a bunch of keywords in and click "search". However, this is how most people search so it is worth stating.
I suspect it is the PageRank factor pushing irrelevent pages to the top of the pile, ahead of more useful ones.
There are many very informative sources out there but incoming links have not been vigorously persued as they would have been for a more commercial content.
Unfortunately Google with it's well intentioned PR algorithm often buries the result that I am looking.
If it wasn't for the preferences allowing a return of 100 results per page it would be even worse.
But I still rate Google streets ahead of the rest.
Resolution:
Personal Filters. Send me a cookie with "-foo.com -bar.com -here.com" where there are default excludes applied to every search.
Personal Filters. If I've seen page X on three previous searches, then filter those out.
Fabulous idea! I for one would love to see such a feature!
a) Commercial/products/shopping/etc... in the fullness of time as the numbers of advertisers increases this is what Adwords will surely become.
There certainly should not be any integration into the main returns (obviously) and no ranking boost as someone suggests. That would be a disaster.
b) Multi-keywords-domains do not receive a ranking boost... so leave 'em alone! I have always used hyphens simply because I find them more descriptive. I use them for no other reason than they help describe the site to the user.
I think a valid point is the one of prepositions and question adverbs.... and also past/present tense. I really am scratching the barrel a bit though with this.
For algorithmic improvement if I were Google I would focus on the Image returns. The ranking is fairly good, but certainly not as accurate as the other tabs. I know... it's a lot harder to define what an image is actually about... I do understand some of the difficulties.... but this is definitely the area that could be most improved (GG didn't ask to to state HOW to do it - thank goodness).
It might also start making more use of Alexa, where user popularity is a factor - although I would think it would have to be a minor tweak in the results, or new sites would never get listed on the first 100 pages.
I've since tried to find him non-biased engines/directories, and although google did mix with the Dmoz entries, it's trying to find results that are mostly non-commercial, and more ambiguous.
similar to the Mac-search and Unix-search - perhaps there could be an educational/learning one, or more specifically a non-commercial search?
I hope the themeing/topic grouping categorsition, such as the top results of the lab/glossary will help.
The main problem I see is not so much Google's results itself, but more in how to use Google.
so many search possiblities play a role.
- using more words or explicitly less
- the sequence and proximity of the words
- inclusions/exclusions
- similar results or other category results
- etc. etc.
We get about six trainees a month at our office, fresh from university. I am amazed how little they know of effective searching (some still only rarely use Google or understand its merits at all).
Google could put an interactive search tutorial on their site, or team up with schools and universities in setting up a short, free schooling educational search programme. Could even be a type of mini-exam (with high-scores-another hype) you could download.
1.
I agree with danny in the trouble of searching for the island java. I heared exactly that problem from a friend some weeks ago.
To refine the search like in teoma is a cool thing. Also the idea of searching only sites included in dmoz.
But the best thing would be to do a search in a category in dmoz, and by option also in all sites linked from the sites included in dmoz. Imagine, you could just do different steps away from dmoz: For example: First you do a search for java and select the dmoz-category "island java". Now all sites included in dmoz under this category and containing the search phrase are shown.
Then you click "one step more" and all pages linked directly from dmoz sites _and_ containing the search phrase are shown. And so on. Just better would be, if you can do that also for yahoo.
2.
And I agree fully with the trouble of multiple-keyword-domain-spammers. It's also a problem in my business. But of course the solution should not be sorting them out because of having multiple keyword domains. But just look on the pages, they have nearly no text on it, as I know them. Then some kind of forwarding technique follows.
I often thought about the possibility for google of randomly rechecking pages without the googlebot-user-agent and the google-ip. Then this cloaking-activities would have an end. (Sorry for all you cloakers out there). Would there be a problem with the spider policy? Hmm, I think not, because you visited and spidered the domain before not anonymous, and therefore the site owner allowed the spidering and knows of it. Why not send an "cloaked" spider after that to an "cloaked" site?
3.
Expanding google to do shopping searches and price-comparison is not a good idea, I think. Except you can be shure, that all commercial sites are in the selection by some kind of artificial intelligence. But doing again this paid-inclusion-thing is unfair. For the user and for the small dealers. I privately don't use price-comparison, because I can't be shure that all dealers are in there. Anyway, it's the thing of the dealers to be sure to be found when people search for "shop" "prices" or "order online". Just include this words on your page.
4. Why not doing some funny things with people (by option, of course)? For example, you could bring together people searching for the same thing at the same time for some chat. Maybe you can opt to wait for example 30 minutes, until somebody searches for the same thing. And then chat. Or just look what he selects then. Or the possibility to leave some hints or greetings for users, that come after me searching for the same thing. Or some kind of collaborative searching. Trouble with all oft that would be always commercial abuse, thats the sad thing.
another interface for arranging your desired "advanced" Google search in a one stop shopping.
Unfortunately I can't find answers on other search engines either-- so the scary thing is I often come to the conculsion that their just isn't any content regarding specific keyword xyz. It's almost as if when I use Google I cross my fingers and hope I get a search result that fits, otherwise I'm doomed cos I know its not anywhere else!
Regardless I'm definetly proficient in searching so its not that, its just like Brett explained with his perl tk gui programming- its so new its hard to find specific content on it yet.
(edited for spelling)
Google Rocks.
However, I think the refine search feature has some merrit. I use google everyday but a couple of recent searches left me wishing there was an easier way.
Example 1: I needed a "portable outdoor play pen" for my 1 year old son because our yard is huge and unfenced. First I performed various iterations of this search and found scads of results for simple playpens which claim to be useable outdoors but which are basically indoor playpens. Taking the time to read through several of the sites I discovered these things are sometimes referred to as playyards. Searches for that were still unsuccessfull as playpens are marketed by retailers as playyards :-(. Reading through these sites (at which point I was beginning to lose hope) I discovered there is such a thing as a superyard xt :-) Searched for that and found a site I could actually buy one, and I did. The whole process took about an hour. Same trouble to drive to toysrus and ask the flunky assistant ;-)
Example 2: I needed some simple freebee customized <hr> graphics. Searching for this gave me several graphics design companies and a few human resources companies, but none having what I wanted. Again, taking the time to read through the pages I discovered hr stands for "horizontal rule" and modifying my search to contain this result produced the desired results. This time I only spent about 30 mins reading sites and "learning" what I should be searching for.
If google were able to provide related keywords to include/exclude during my search, then I would not have had to read the text of pages to "learn" more about possible terms for what I was looking for. In example 2 could have excluded the human resources links (though i can do this with minus sign, most peeps don't know this), and immeadiately been able to intuit that the google suggestion "horizontal rule" was probably what i wanted.
Just my .02
Maybe an image submit. I submit a page with an image that has text within, and then it gets indexed. That is all I can think of.
however, im wondering if googles perception of webmasters and who they choose to PR0 may possibly be tweaked to cure some of its current "problems".
I wonder how many sites are "blacked out" by PR0 or "browned out" by poor PR. Then, if relevance could be measured....how much of an effect perfection would make by having a "proper" representiation of each page (ie this utopia google is looking for)
i see my site has went down for a few keywords....must mean that googles results are getting less relevant, or google chooses to define my site as less important than it was a month ago (the other sites in the SERPs are the same also)
I agree. I search Google for Zips, phone numbers, addresses..etc.....
works great.
I also use google a lot when I get a call on my cell phone and I don't recognize the area code.
