PCWorld Forums: Google's Search Algorithm Challenged - PCWorld Forums

Post your comments for Google's Search Algorithm Challenged here

The problem with searches is that you get thousands of hits that have nothing to do with your search words. Even if you say must contain this phrase they still pull the words out and give you anything. I've had hits first in line that didn't contain one word of my search parameters in it.

It will take more than an algorithm to take over Google's market share.

Too late, Google is a search giant and it has a massive brand. It is used by practically every internet user out there. Besides, what other RADICAL changes can you make to a search engine to entice a large numbers of users to use it? Besides tweaks here and there, I don't see anything revolutionary coming up for the future of search engines, it just does one job: search.

Num2, on 19 November 2011 - 02:20 PM, said:

That's when you pick "Verbatim" under "More search tools" in the left hand nav.

SethathiMorokole, on 19 November 2011 - 04:39 PM, said:

Sounds to me like this will be the next Cuil.

does he have the world data like Google has?

i think people should not be limited to one algorithm for searching. at this time its easier for people to simply refine their search criterion to ascertain a different set or refined set of results. but i would rather click a button and let the server allocate a different algorithm for me.

Fsd, on 19 November 2011 - 05:23 PM, said:

governments are always looking for improved technology. the new algorithm will likely de-throne google in foreign markets and even with our own covert sub governments that not only require faster super computers but faster software to conduct searches of their databases, which i think are larger than googles, bings and yahoo combined.

NOTHING IS IMPOSSIBLE...
TECHNOLOGY IS NOT CONSTANT...
WHO NOW..GOD ONLY KNOWS WHETHER WE ADAPT TO NEW SEARCH ENGINE IN FUTURE OR NOT...
TRYING IS MORE BETTER THAN BEING IDLE...

Fsd, on 19 November 2011 - 05:23 PM, said:

Search engines aren't doing a great job. I come daily upon Google searches that bring no relevant results. If someone can come up with a better solution, not just a little better but significantly better, I'm sure I will use it, and I'm sure a lot of others will. Google may be a massive brand, but if a much better search engine comes along, either Google will buy it or another big player (such as Microsoft) will buy it, and it will be used.

I'll wait to see what this guy comes up with, but I can only hope he does have something nice up his sleeve.

The name itself, Volunia, is DOA.

Well, when the U.S. starts forcing Google, Yahoo, etc. to block internet hits of 'banned' content (like China does), underground search engines that DO NOT block that content will get some hits. Until they're blocked.

Secure proxy services that grant access to the 'whole' internet, through servers hosted in countries that haven't become a corporatist/fascist paradise will become a normal fixture of internet use.

Fsd, on 19 November 2011 - 05:23 PM, said:

Never say never

Shouldn't be too difficult to beat Big G's SERPs re. relevancy.

The problem, though, lies in the fact that the majority of Big G's users are not sufficiently discriminating; for them, quantity impresses. And, that they are there owing to the freebies, which are just good enough to attract and addict them, that Big G offers.

Well, actually I'm just used to how it works.

When they change it around, it's annoying, because the expected KINDS of search results change with it. Though the 'fix' they did a year or two ago where they got rid of the big 'PAY US TO ANSWER THIS' sites off the first page or so of hits was welcome. Saving me from having to page past them.

Adding and removing words from a search is just routine. It's not having a 'million' results for something that impresses me. It's finding what I want at, or near the top with relatively few tries.

But this does, in fact, take some slight amount of skill and experience. For instance, knowing something about computers makes finding information about computers VERY easy. Though having all of your results dump to ads for retail products and services seems to satisfy most clueless fools.

Unfortunately, you can't really automate that knowledge away trivially. Knowing WHAT to ask, and HOW to ask it is important in any kind of research.

What annoys me about Google results is the repetition of the same content from the same site in multiple matches. I do a lot of searching on PCWorld.com (because I work here) and I see dozens of repetitions of (for example) the same headline where it appears as teaser text in a "Related Articles" box. This happens even though Google assures me that it has excluded many matches from its reported results because they are substantially identical.

I know that Google wants to block repetitious content, but it seems stymied by at least two things: an inability to distinguish between the main content on a page and peripheral content (which leads it to see the content of pages with duplicate sidebar heads as different because the unrelated main bars on the page are different) and a problematic weighting of metadata (which people viewing the page from the outside don't see, but which may cause the same main-page content to seem different to a search engine that treats the metadata variations as highly significant).

I also don't understand why the number of Google results you get increases (hugely, in many cases) when you try to refine your search by excluding certain terms. For example. I just now searched for the exact word or phrase "mausoleum" and Google reported 11.6 million matches. Then I searched for the exact word or phrase "mausoleum" but with none of the words "velociraptor"--and Google reported 19.2 million matches. Hunh?

Anyway, I welcome a fencer's foil in place of (or as a supplement to) a bludgeon, if Marchiori's group can pull it off.

I think the a future generation of search will need some AI to evaluate those links while it's crawling over it. Or people doing that.

Here's another interesting search engine toy...
http://www.wolframalpha.com/

I remember using metacrawler as an engine, once upon a time.
http://www.metacrawler.com/

Metacrawler, and several other meta-SEs, are still around, and still useful.

Yargs, on 20 November 2011 - 04:32 PM, said:

Hm-mm; not really.

What is wants to do is to rank multiple occurrences of multiple copies of content in the SERPS according to the context within which they appear. Think of it as being analogous to your ranking various covers of the same musical recording.

Yargs, on 20 November 2011 - 04:32 PM, said:

You may find, though, as you drill deeper into the SERPs, that such numbers decrease dramatically.

At least part of this strange behavior owes to the fact that the initial reported totals are estimates; Google does not actually construct a full listings of all possible SERPs, but rather extends them as needed when called for by the user.

Also, remember that we now have the case of the double-quotes and plus-sign operators having been seemingly conflated, with unknown consequences.

This post has been edited by deepsand: 20 November 2011 - 08:04 PM