Search Engine Watch
SEO News

Go Back   Search Engine Watch Forums > General Search Issues > Search Technology & Relevancy
FAQ Members List Calendar Forum Search Today's Posts Mark Forums Read

Reply
 
Thread Tools
Old 02-05-2005   #1
randfish
Member
 
Join Date: Sep 2004
Location: Seattle, WA
Posts: 436
randfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to all
Question Can IR Research Teach us Anything re: Spam Prevention or Anti-SEO Tactics?

Quote:
Originally Posted by randfish
For someone who is trying to understand the technology & theories to get a better understanding of how to use them in his/her favor, it's great to have IR scientists like yourself to help out. I'm wondering whether you have any insight into the other side of things - spam recognition/prevention, counter-SEO work, etc.
Quote:
Originally Posted by Nacho
Sounds like a great idea, you guys should start another thread on that.
Let me try to be even more specific about what I'm asking:

#1 Are there IR papers, research topics, scientists who have insight into preventing search spam from appearing in the indices of the major commercial search engines?

#2 Are there any of the above that focus on tactics commercial search engines can use to block/prevent/inhibit the practice of optimization used by SEOs to bring particular sites/pages into the top SERPs?

Last edited by randfish : 02-05-2005 at 03:32 PM. Reason: stray ] character
randfish is offline   Reply With Quote
Old 02-06-2005   #2
seobook
I'm blogging this
 
Join Date: Jun 2004
Location: we are Penn State!
Posts: 1,943
seobook is a name known to allseobook is a name known to allseobook is a name known to allseobook is a name known to allseobook is a name known to allseobook is a name known to all
Quote:
Originally Posted by randfish
#1 Are there IR papers, research topics, scientists who have insight into preventing search spam from appearing in the indices of the major commercial search engines?

#2 Are there any of the above that focus on tactics commercial search engines can use to block/prevent/inhibit the practice of optimization used by SEOs to bring particular sites/pages into the top SERPs?
http://airweb.cse.lehigh.edu/
__________________
The SEO Book
seobook is offline   Reply With Quote
Old 02-06-2005   #3
randfish
Member
 
Join Date: Sep 2004
Location: Seattle, WA
Posts: 436
randfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to all
From the site:
Quote:
the conflicting goals of accurate results of search providers and high positioning by content providers provides an interesting and real-world environment to study techniques in optimization, obfuscation, and reverse engineering, in addition to the application of information retrieval and classification.

The workshop solicits technical papers on any aspect of adversarial information retrieval on the Web. Particular areas of interest include, but are not limited to:

search engine spam and optimization,
crawling the web without detection,
link-bombing,
reverse engineering of ranking algorithms,
advertisement blocking, and
web content filtering.
They don't think very highly of us, do they?
randfish is offline   Reply With Quote
Old 02-08-2005   #4
orion
 
orion's Avatar
 
Join Date: Jun 2004
Posts: 1,044
orion is a splendid one to beholdorion is a splendid one to beholdorion is a splendid one to beholdorion is a splendid one to beholdorion is a splendid one to beholdorion is a splendid one to behold
Exclamation

Quote:
#1 Are there IR papers, research topics, scientists who have insight into preventing search spam from appearing in the indices of the major commercial search engines?
There are plenty of IR work in the field of spam identification based on frequency and length of nouns, verbs, adjectives, titles, etc.

We have developed a proprietary IR method for identifying spam based on on-topic analysis but it still in beta. If added to a search engine, it could be used as a spam filter.


Orion
orion is offline   Reply With Quote
Old 02-08-2005   #5
randfish
Member
 
Join Date: Sep 2004
Location: Seattle, WA
Posts: 436
randfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to all
Orion,

I had been noting that very basic stop-word frequency, sentence structure & noun frequency analysis could be used to root out a lot of the spam I was seeing in Google's results. Admittedly, I don't have the time or resources to test it over a large number of searches, but I was surprised at how many spam results would fall prey to something so simplistic.

Is your system something you plan on licensing? selling? any bidders?

Best of luck!
randfish is offline   Reply With Quote
Old 02-09-2005   #6
xan
Member
 
Join Date: Feb 2005
Posts: 238
xan has a spectacular aura aboutxan has a spectacular aura about
There are lots of methods about how to filter spam from indices, LSI was still used in 2003 actually.

SEO's are not popular no, this is because the general idea is that they manipulate our data so that our retrieval efforts are always flawed. It's frustrating. If you look up white papers on spam and indices, SEO's are not considered angels! Imagine someone your links changed all the time on your sites, so your performance was always flawed. maybe not a great example, but you get my drift? I'm sorry to say that. I suppose the next thing is to work together and maybe get to know each other.

This is a paper you might find of interest:

Damn spam by microsoft

IBM

ISYS

You will find many more as well.

Last edited by xan : 02-09-2005 at 09:14 PM.
xan is offline   Reply With Quote
Old 02-10-2005   #7
orion
 
orion's Avatar
 
Join Date: Jun 2004
Posts: 1,044
orion is a splendid one to beholdorion is a splendid one to beholdorion is a splendid one to beholdorion is a splendid one to beholdorion is a splendid one to beholdorion is a splendid one to behold
Exclamation

Randfish, here is one of such papers.

http://www.stanford.edu/class/archiv.../rdg12-afw.pdf

Hope can see some of you at AIRWeb, which is part of the W3C Conference (www2005, Japan). I was invited to present.

Orion
orion is offline   Reply With Quote
Old 02-13-2005   #8
claus
It is not necessary to change. Survival is not mandatory.
 
Join Date: Dec 2004
Location: Copenhagen, Denmark
Posts: 62
claus will become famous soon enough
no different from email spam

SERP spam is no different from email spam, in the sense that it is entirely possible to train bayesian filters or the like to identify spam pages - even based entirely on on-page factors.

For those that like to play around, here's a nice How-To article from IBM (using PHP). There's also the PHPBayes thingy - page is in French, but it runs okay on my box anyway (that was a joke)

If you're more of a Perlophile, Ken Willams has developed some nice stuff to try out.

--
Added: Here's some powerpointy stuff on the concept (in html though) - you will need to know that "weiter" is German for "Next" - it's from Uni Stuttgart.

Information Retrieval and text mining 1

And here's a pdf (also in German): Maschinelle Sprachverarbeitung in Theorie und Praxis

Both of the above are focused on the email variety, but that doesn't really matter, as text is text. This one mentions the web page variety a bit more:

Information Retrieval and text mining 2 (around page 29+)

Last edited by claus : 02-14-2005 at 04:09 AM. Reason: added some + fixed a stupid typo
claus is offline   Reply With Quote
Old 02-13-2005   #9
xan
Member
 
Join Date: Feb 2005
Posts: 238
xan has a spectacular aura aboutxan has a spectacular aura about
I asked around to see if anyone had any up to date insight on this and I these papers were thrust at me. The general consensus:

Threshold has to be tuned and its bad at estimating probabilities. Bayes is a supervised learning algo so to improve it needs to be exposed to a lot more training data somehow.

Spam filtering is a pure text classification task. Bayes has been used for a long time to classify things. Indexing is also a classification task - the problems are very similar. Methods can work the same for both.

There's stuff on that here:

Freshmeat

Density-based spam detection is being studied, there was a paper written in 2004:

Industry/government track papers: Density-based spam detector - Kenichi YOSHIDA, Fuminori ADACHI, Takashi WASHIO, Hiroshi MOTODA, Teruaki HOMMA, Akihiro NAKASHIMA, Hiromitsu FUJIKAWA, Katsuyuki YAMAZAKI
August 2004
Proceedings of the 2004 ACM SIGKDD international conference on Knowledge discovery and data mining

They seem to look into the problems we currently have with Bayes. They say they have promising results. They use document space density to determine which emails are spam. They use a "mass mail detector" which monitors packets between servers. Their "Vectorizer" sticks all the text into a vector representations (Ngram, TF, etc...). All the values are given hash values. They solve the problem of the learning engine having to backtrack all the time to handle a single email by making a new unsupervised learning algo. The hash values are used and mapped to the cache. Its faster apparently. Definately worth the time to look at it, this paper shows possible solutions.

Another worthy piece of research is:

Traffic characterization and SPAM: Characterizing a spam traffic
Luiz Henrique Gomes, Cristiano Cazita, Jussara M. Almeida, Virglio Almeida, Wagner Meira
October 2004
Proceedings of the 4th ACM SIGCOMM conference on Internet measurement

They look at the problem form a different angle. They work to identify a "spam signature".
They look at e-mail headers, sizes, number of recipients per mail,popularity and temporal localityamong recipients are seen as they key differences between spam and non-spam (but I am simplifying).

I would say, as always we need to be better at dealing with linguistics, but in the face of current techniques, it might not be such a crucial issue. One thing for sure is that naive bayes is going out of fashion...

Fighting SEO in the search engines comes back to my blog article. Something has got to happen eventually.

SE's tend to stay on top of what people know and what they're doing regarding SEO. If they think think its a serious problem, they will counteract, like they did with people inflating the word density on their pages. SEO shouldn't bother us, it should encourage good practice which means our indexes are cleaner.

Sorry randfish, I didn't understand what kind of stuff you were looking for. I hope this helps.

If you have trouble finding the papers, I will email a copy to you.

Last edited by xan : 02-13-2005 at 08:23 PM.
xan is offline   Reply With Quote
Old 02-13-2005   #10
randfish
Member
 
Join Date: Sep 2004
Location: Seattle, WA
Posts: 436
randfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to all
Thanks to all of you. I'm going to try and read through all of this and come back with what I can get from these papers. Xan - thanks for the offer to email me, I may take you up on that if I can't find them tonight.
randfish is offline   Reply With Quote
Old 02-14-2005   #11
claus
It is not necessary to change. Survival is not mandatory.
 
Join Date: Dec 2004
Location: Copenhagen, Denmark
Posts: 62
claus will become famous soon enough
Quote:
Originally Posted by xan
Threshold has to be tuned and its bad at estimating probabilities. Bayes is a supervised learning algo so to improve it needs to be exposed to a lot more training data somehow.
Yes, you do need human interaction. Really. IMHO, this is where the major SE's are coming to now - there's a limit to what can be done automagically when it comes to quality, as quality can be spoofed.

As to "a lot more" - i'd say it was continuous training and maintenance. Still, i'd risk the statement that you can get a serious amount of "bang for the buck" by employing relatively few, relatively low-skilled people to assist in the training. Even with a corpus of billions of pages.

Quote:
Originally Posted by xan
Spam filtering is a pure text classification task. Bayes has been used for a long time to classify things. Indexing is also a classification task - the problems are very similar. Methods can work the same for both.
We are talking about the specific subclass of web pages popularly referred to as "SPAM" (however wrong that term may be - or not). Ie. not the email kind of spam, but web pages in stead. As stated in my post above, i agree that it's a case of "same, same - only different" - you should be able to use same/similar methods on both problems.

Quote:
Originally Posted by xan
I would say, as always we need to be better at dealing with linguistics, but in the face of current techniques, it might not be such a crucial issue.
As applied to spam (email + web) i agree. Linguistics, semantics, and such might improve "natural search" but to deal with this specific subclass of texts it doesn't seem necessary at all to be able to understand any of it. It's a simple task and a simple classification is the only thing that's needed. So, simple tools would be preferred, imho.

Quote:
Originally Posted by xan
One thing for sure is that naive bayes is going out of fashion...
Not that i care a lot about what's in fashion or not, but still - would you care to emphasize a little on that statement?

What makes you think that this simple and efficient tool has lost value (if that's what you are proposing)?

Quote:
Originally Posted by xan
Fighting SEO in the search engines comes back to my blog article. Something has got to happen eventually.
It's not really about "fighting SEO" per se - SEO's do a lot of different stuff and far from all of it can be classified as spam (except by the most extreme "purist" metrics) - some SEO (some would say most SEO) is even useful for the SE's as it enables them to index the web better.

Anyway, that's an age-long and sometimes heated discussion, let's not go there. Regarding your blog article - i didn't see a link anywhere in this thread, could you post it or PM me?

Last edited by claus : 02-14-2005 at 04:04 AM.
claus is offline   Reply With Quote
Old 02-14-2005   #12
randfish
Member
 
Join Date: Sep 2004
Location: Seattle, WA
Posts: 436
randfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to all
The dialogue between SEs, engineers and SEOs is extremely critical at this point in time. In order to stop the kind of SEO (spamming) that makes SEs think of us as a conflicting entity, we have to think in reverse.

SEO as an industry and a methodology of marketing/advertising (which is what it really is) needs to be better publicized, more open and more connected with commercial search engines. If Google, Yahoo!, MSN & Ask were to all have an FAQ on ranking and point to a single organization that creates standards and regulates the SEO industry, we could all come to a much better place.

Imagine:

SEs could learn to better fight spam and search issues by having a one-on-one dialogue with the people who watch their results even more obsessively than they do. SEOs could get issues like banning, 302 hijackings, indexing issues, etc. fixed quickly without having to complain for months until an SE rep. notices it on a board like this one.

There is so much critical interaction that should be taking place, and I think most of it, in the end, will benefit the SEs more than the SEOs. But, even so, I'm more than ready and willing to say it should happen. Critical mass may not be here yet, but it will be soon, and people on both sides of this fence need to put aside the differences they perceive and learn to see that they should be working towards a common goal.

The best SEO is the one who is helping the search engines find and rank the best content. After all, the future (and much of the present) industry of SEO is actually creating great content - and personally, I much prefer it to testing keyword density formulas.

BTW - Still haven't gotten to the papers; I'll try to tomorrow.
randfish is offline   Reply With Quote
Old 02-14-2005   #13
xan
Member
 
Join Date: Feb 2005
Posts: 238
xan has a spectacular aura aboutxan has a spectacular aura about
What makes you think that this simple and efficient tool has lost value (if that's what you are proposing)?

There's a lot of research showing that naive bayes is not efficient enough, for reasons similar to those I gave in my post. Naive bayes has been around for a long time, almost 10 years, things move on. Support-vector machines are being used by a lot of people now. They're very powerful. k-Nearest Neighbors are also very powerful classifiers, and are commonly used. Naive bayes did win the KDD-CUP, and is particularly efficient in classification tasks, however it doesn't cope with higher order interactions, if features do not belong to clear classes, they don't get modelled.

Bayes classifiers are not bad or unefficient, but new models exist that prove better at the task of classification.

My blog article simply addresses the different points of view between SEO's and IR professionals:

Search science

Last edited by xan : 02-14-2005 at 02:30 PM.
xan is offline   Reply With Quote
Old 02-15-2005   #14
claus
It is not necessary to change. Survival is not mandatory.
 
Join Date: Dec 2004
Location: Copenhagen, Denmark
Posts: 62
claus will become famous soon enough
warning: long post follows
--------------------------------------------------------

Thanks for the reply Xan - i didn't know about vector support machines, but a little reading quickly made the point of those clear to me. Interesting concept, really. It seems to me that it's somewhat a combination of the Bayesian stuff, geometry, and standard min-max thinking.

Now, i'm not so sure we disagree that much, but i have to tear your post apart anyway. I've got a feeling that you don't fully understand the concept of what "spammy web pages" is, and that you confuse it with email spam, which is something entirely different.

For that reason, i also personally don't like to use the term "spam" about web pages, but we lack a better term, and the SE's use it. Also, there's no clear definition of what a "spammy web page" is as opposed to "unsolicited commercial email".

---
Vector support machines:

I might be wrong here, but in dealing with very large data sets (as in 4-10 billion pages) and the specific problem of "spam vs. non-spam web pages" (with a high degree of noice) it seems to me that this technique is very hard to employ relative to Bayes, and that it needs a far greater amount of maintenance, and control. Also, too many parameters have to be estimated, imho (and it seems there are problems with scale, but perhaps that's just me).

As for n-Nearest, well, why not? That comes closer to the simplicity that is not only desirable, but often required when dealing with very high volumes of data. You will still have the problem of choosing the optimum value of "n" which - due to all the noice - might not be the same across the full data set, and/or across different timeslices.

I'm not trying to be impolite, but there's always a tradeoff in the choice of tools, and the main points to seperate these from bayes seem to be (a) novelty, and (b) more maintenance.

So, back to Bayes:

Quote:
Originally Posted by xan
There's a lot of research showing that naive bayes is not efficient enough, for reasons similar to those I gave in my post.
Okay, and these were:

Quote:
Originally Posted by xan
Threshold has to be tuned and its bad at estimating probabilities. Bayes is a supervised learning algo so to improve it needs to be exposed to a lot more training data somehow.
As for training, i've commented on this already. A data set of some billions of pages should provide adequate data, i'd say. Theshold tuning is no different from the techniques you mentioned, it's even simpler than in the case of SV. IMHO, probabilities are not a required feature for this specific task, as:

Quote:
Originally Posted by xan
Spam filtering is a pure text classification task.
Next:

Quote:
Originally Posted by xan
Bayes has been used for a long time to classify things.
That a tool has been used before hardly makes it less useful

Quote:
Originally Posted by xan
Density-based spam detection is being studied, there was a paper written in 2004:

(link - requires an ACM Web Account. )

They seem to look into the problems we currently have with Bayes.
I don't have such an account, so i couldn't read the paper. In the abstract the main benefit they emphasize is that their method is un-supervised. That's a good thing. They do require "with a short white list" though, so it's not that much of a difference to having students train filters. Still, that's only one problem, not problems, so i still miss the rest.

Anyway, you wrote a few details:

Quote:
Originally Posted by xan
They use a "mass mail detector" which monitors packets between servers.
Does not apply to this particular problem.

Quote:
Originally Posted by xan
Their "Vectorizer" sticks all the text into a vector representations (Ngram, TF, etc...). All the values are given hash values.
Not extremely different from Bayesian tokenization is it?

Quote:
Originally Posted by xan
They solve the problem of the learning engine having to backtrack all the time to handle a single email by making a new unsupervised learning algo. The hash values are used and mapped to the cache. Its faster apparently.
So, they put all their discriminators in memory, and that way they don't have to query the database? Nice thought but unlike email, speed does not have to be a crucial factor in web spam filtering as you can do this prior to querying, or even prior to indexing. I didn't see any references to Bayes in the abstract.

Quote:
Originally Posted by xan
Another worthy piece of research is:

(link - requires an ACM Web Account. )

They look at the problem form a different angle. They work to identify a "spam signature".
They look at e-mail headers, sizes, number of recipients per mail,popularity and temporal locality among recipients are seen as they key differences between spam and non-spam (but I am simplifying).
Again, i could only read the abstract. This is very specific to the email-variety of spam, though. I didn't see any references to Bayes.

Next:

Quote:
Originally Posted by xan
I would say, as always we need to be better at dealing with linguistics, but in the face of current techniques, it might not be such a crucial issue. One thing for sure is that naive bayes is going out of fashion...
I agree on the first part. After all this work, the fashion part is still a mystery to me though.

So, back to your latest post:

Quote:
Originally Posted by xan
Naive bayes has been around for a long time, almost 10 years, things move on. Support-vector machines are being used by a lot of people now. They're very powerful. k-Nearest Neighbors are also very powerful classifiers, and are commonly used.
Yes, lots of stuff has been around for a long time. That's not a valid argument in my book. Yes, there are other powerful tools, and yes, these are also used. Still, i maintain that for the specific task of filtering web page spam i don't see those two you mentioned as superior to bayes.

Quote:
Originally Posted by xan
Naive bayes did win the KDD-CUP, and is particularly efficient in classification tasks, however it doesn't cope with higher order interactions, if features do not belong to clear classes, they don't get modelled.
You seem to have a somewhat ambiguous realtionship with the Bayes method *lol*

Higher order interactions are not necessarily part of this particular problem scope (although the amount of noice could suggest that it woud be beneficial to consider those as well). I will maintain, that:

Quote:
Originally Posted by claus
you can get a serious amount of "bang for the buck" by employing relatively few, relatively low-skilled people to assist in the training. Even with a corpus of billions of pages.
Lastly,

Quote:
Originally Posted by xan
Bayes classifiers are not bad or unefficient, but new models exist that prove better at the task of classification.
Perhaps, due to the higher-order stuff you mentioned, the accuracy level of some other methods might be better, or even extremely good. This is not a case where very high accuracy is required, though, as you have a secondary system (the PR-process, or ranking system) that will limit the impact of false negatives.

You don't always need the most sophisticated tool. What you need in this case is a "serious workhorse" that will catch as much as possible from a very high volume data set with some small (tolerable) error margin, and a low maintenance level. Once the training data set is starting to evolve, you are in that exact situation.

The key to having more false negatives than false positives lies in the selection of training data. For that reason well-instructed humans must be employed, and the model training should start with the most blatant examples.

I hope the above somewhat clarifies why i suggested the Bayesian approach, and still think it's a very valid suggestion for this particular topic. Bayesian spam filters have been are employed succesfully on emails for a long time, so they also have a solid track record from a related field.

---
On a more personal note:

Thanks for the blog link, that was interesting reading. I had a hard time doing the actual reading though, as white on black is very hostile to my eyes, especially with a font size as small as that. It is a barrier to getting your message out, i think. Anyway, i use firefox, so i could just override your layout and colors, so if you really like it that way don't bother changing it for me.
claus is offline   Reply With Quote
Old 02-15-2005   #15
claus
It is not necessary to change. Survival is not mandatory.
 
Join Date: Dec 2004
Location: Copenhagen, Denmark
Posts: 62
claus will become famous soon enough
Quote:
Originally Posted by claus
there's no clear definition of what a "spammy web page" is as opposed to "unsolicited commercial email".
On a related note, i am working on using the Bayesian approach on a link checker script. I will look into the other suggestions as well. I'm having the problem of maintaining a large collection of topical links that are also publicly available as a directory-type site. While i can check for 404 links (page not found) automatically, i can not check for pages that change scope.

One particular problem that annoys me a lot (in regard to the directory, otherwise not) is the people buying popular domain names and exchanging the content on those with so-called "search portals" having only a lot of pay-per-click links and no "real" content. You could say it was a variety or a subclass of the web page spam issue, i guess. These will not get caught automatically as they do not return the 404 error, but they offer no value in my link collection, so i purge them whenever i see them.

Pages like that seem like they are ideal candidates for such an approach,at least to me. Making the bayesian link checker is a very slowly progressing project because i'm always extremely busy with other projects, so if anyone knows of an existing OSS link checker bayesian script, i'd be glad to know about it

Last edited by claus : 02-15-2005 at 07:16 AM.
claus is offline   Reply With Quote
Old 02-15-2005   #16
xan
Member
 
Join Date: Feb 2005
Posts: 238
xan has a spectacular aura aboutxan has a spectacular aura about
Well, you can't please them all!

I have average results using naive bayes, so maybe I'm biased.

All valid comments claus. "out of fashion" is a UK expression meaning to become extinguished I supposed. I meant being used less and less.

SVM's perform beautifully for many different applications. Most of my collegues are using these now.
xan is offline   Reply With Quote
Old 02-17-2005   #17
xan
Member
 
Join Date: Feb 2005
Posts: 238
xan has a spectacular aura aboutxan has a spectacular aura about
Quote:
Originally Posted by randfish
Let me try to be even more specific about what I'm asking:

#1 Are there IR papers, research topics, scientists who have insight into preventing search spam from appearing in the indices of the major commercial search engines?

#2 Are there any of the above that focus on tactics commercial search engines can use to block/prevent/inhibit the practice of optimization used by SEOs to bring particular sites/pages into the top SERPs?
Hello my friend.

Article on search-science on its way inspired by your questions. I found them interesting. As always you have good threads. I will mention you on the article of course. Did you want a link or somthing?
xan is offline   Reply With Quote
Old 02-17-2005   #18
randfish
Member
 
Join Date: Sep 2004
Location: Seattle, WA
Posts: 436
randfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to allrandfish is a name known to all
Xan,

Just your presence here is a gift. Thanks for your contributions. I'll definitely seek out your site to take a look. If you're interested, I do blog on my own site - seomoz.org - regarding search, seo, etc. I am still looking for a few free hours when I can read all of the papers recommended to me in this thread.

I think that SEOs still have a long way to go before they realize how advanced spam detection is and can/will be in the future. Perhaps by illuminating this know we can help the industry to avoid the nasty path of banning and penalization.

Xan, I'm wondering if you can tell me if there are better ways to find research papers than through Google's search. Should I be using specific academic search engines to find these papers - often Google will only have the papers listed through Portal, which requires a membership.

Thanks!

Last edited by randfish : 02-17-2005 at 11:55 AM.
randfish is offline   Reply With Quote
Old 02-17-2005   #19
xan
Member
 
Join Date: Feb 2005
Posts: 238
xan has a spectacular aura aboutxan has a spectacular aura about
Hi Randfish,

there are lots of research sites out there, especially one very popular repository: Citeseer.IST

The IEEE, the ACM amongst other are quite expensive and really only worth it if you are into hardcore equations and things. Citeseer offers some ACM papers, and is an excellent place to go fishing. Maybe try Google scholar?

If you have any comments about my article, please say so and I will post your response.

Happy fishing!
xan is offline   Reply With Quote
Old 02-17-2005   #20
xan
Member
 
Join Date: Feb 2005
Posts: 238
xan has a spectacular aura aboutxan has a spectacular aura about
Ok, its posted, but its not technical or anything like that. Just basic banter.
xan is offline   Reply With Quote
Reply


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off