Domain: ijs.si
Stories and comments across the archive that link to ijs.si.
Comments · 57
-
Re:20-30 ms is massive
Easy to tell if you run NTP (pretty much certain if you have servers). For example, take a look at the image here - you can clearly see the effect of route changes to remote servers.
-
Re:Dr Eliza
-
Re:O RLY?
I love your eyes Roxxy
roxxy: I love your eyes too, bill.
You make me hot!
roxxy: do you like it that i make you hot?
What's your favorite sex move!?!
roxxy: What about your own favorite sex move, bill?
I like pinching your nipples.
roxxy: Can you elaborate on that, bill?
I like pinching them and making you squeal!
roxxy: Do you wish to like pinching them hard so me, bill?
-
Eliza
-
Virtual Medical License
Doesn't a "virtual doctor visit" necessarily imply that there are virtual doctors? If so, please send me a check for $45 and proceed to http://www-ai.ijs.si/eliza/eliza.html
Thanks.
-
Re:Lots of tweaking required (in general)
I used this: http://www-ai.ijs.si/eliza/eliza.html
-
Re:I have a way of dealing with this,
I still think Eliza is a real person.
-
Re:Exaggeration?
You are looking at it wrong.
Not every one has your experience, and your experiences aren't the only legitimate ones.
IN your example, just web crawl this:
http://www.ijs.si/software/amavisd/amavisd-new-docs.html
Bam, documenting is done.While documenting can be held to an unreasonable standard, your example is not it.
For example, developers should never do end user documentation. Hire a professional becasue users need it laid out differently then developers, and most developers aren't trained to think that way.
"Not to mention that when you look at how some of these people code and/or setup systems/databases, it's obvious just WHY they need so much documentation. The darned things don't make any sense. Without some ancient codex you can't make heads or tails of the system."
That explains a lot.
When you do it it's obvious and everyone should know, but when 'they' do it it's not obvious.
You need to think about your ego and expectations a little. It will do you good."I think far too often "Josh" might simply be getting mud slung his way due to him being an ass."
There, fixed that for you.Sorry, couldn't resist after reading your sig.
Yeah, and I am scared of being 'Dealt with' from someone like you~ -
Eliza
Wow. Just wow. Eliza was more convincing than that.
-
Re:Simple, cheap and high quality 3D home theater
Now all I have to do is glue two video cameras together and sell it as a "budget 3-D video camera"...
You mean like this?
You can synchronize the cameras using over LANC using this device.
-
Re:Girlfriend?
Real geeks compile their girlfriends from source.
Eliza, is that you???
http://www-ai.ijs.si/eliza/eliza.html -
Re:Behavior isn't as complicated as we think
What he has clearly demonstrated is that behavior, especially in insects, obeys very simple rules. His insect robots have almost no processing power and yet mimic the behavior of real bugs very well.
Interesting, I'll bet I could make a robotic
"Dup. Wasn't WETA Working on Robotic Lizard For Science months ago?" /. poster. I'd just make some minor variations to Eliza. For example:
"I was Working on Robotic Lizard For Science for the last 10 years. Nothing new here."
"What is the carbon footprint of WETA works on Robotic Lizard for Science?"
"Microsoft is evil because WETA Working on Robotic Lizard for science" -
Re:Possible Text VersionBogofilter works great. Or SpamAssassin but only if you force-feed it its own judgements. In both cases you have to correct classification errors.
Fidelis Assis (who has now gone solo after having participated in the CRM114 project) shows great results for his recent solo effort: OSBF-lua Bratko's PPM spam filter -- the one that did great at TREC -- is not yet packaged as a drop-in filter. Same for my DMC spam filter.
The actual TREC 2005 tests referred to in TFA are here.
-
Re:Out of Date and WorthlessI assume the paper that you are describing is the 2004 study. The paper described in the talk (which was given 6 months ago or so) described results of the TREC 2005 Spam Track which took place in November 2005. It included a test SpamAssassin 3.x, not 2.3.
TREC 2006 evaluations are now underway.
While it is reasonable to conjecture that spam has changed so as to defeat spam filtering techniques, or will change so as to defeat the PPM technique that did well at TREC, the historical evidence does not support this conjecture. In particular:
- The spam filters tested in 2004 give pretty well exactly the same performance on 2005 and 2006 data.
- New versions of the filters are a little bit better, but not by leaps and bounds, and also get about the same results over the last 2.5 years of data.
- There is no evidence that "Bayesian poisining" is a viable technique for defeating statistical spam filters in anything but a very artifical laboratory environment where the poisoner has access to the recipient's inbox
Andrej Bratko used PPM -- a well-known data compression technique to compress ham and spam separately. Well actually he didn't compress them but just build the statistical model necessary to compress them. Then he simply (tentatively) added the unknown message to each model and chose the one that compressed it best. The general technique of using compression has been mentioned here and elsewhere but Bratko used a much stronger compression scheme and was somewhat clever about it.
I later reproduced Bratko's results using DMC -- a compression schem that I invented 20 years ago -- and got some interesting results. We have a journal article in press describing it and also an evaluation paper at CEAS 2006.
Bratko A., Cormack G. V., Filipic B., Lynam T. R. and Zupan B., Spam Filtering Using Statistical Data Compression Models
-
Illusions of HumanityThis illustrates an important fact. When you talk to a call center you have the illusion you are conversing with a human being. This is not the case. The carbon-based unit is just a front end for a procedure somebody has designed. Intelligent action by a call-center drone is about as likely as Eliza actually figuring out what your problems are.
The purpose of this procedure, obviously, is to make it as difficult for the user to cancel as possible. Otherwise AOL would just automate it. More obnoxious than most call-center procedures, but not by much. The sad thing is that AOL dealt with the bad press by firing the poor schmo who just did what he was paid to do. The people who designed this stupid procedure are still working there. For now.
-
Re:Let me be the 1st
Sounds like you have issues. A friend of mine recommended Dr. Eliza. She's really helped me work through some of my own issues. The voices in my head are down to a dull roar, and my twitching no longer turns bath time into jaccuzzi time.
-
Re:Yeehaw!
*checks notepad* is this some new meme I am unaware of?
If so, I like it.
WTF that 'you're not a script' test is back on /., this time it is so hard, I am questioning if I am in fact, not a script.
Or should I say: http://www-ai.ijs.si/eliza-cgi-bin/eliza_script
You: One of them, a jug of whiskey, and a bunch of squirrels, and you got yourself a party!
Eliza: Oh, i got myself a party.
You: with whiskey and squirrels!
Eliza: Please go on.
You: aren't you excited by squirrels?
Eliza: Oh, i excited by squirrels.
You: why not say, you mentioned before you like squirrels
Eliza: Oh, i mentioned before i like squirrels.
You: why not say, you like goaste, and wish you could live inside his cavernous orifice
Eliza: Oh, i like goaste and wish i could live inside his cavernous orifice.
From: http://www-ai.ijs.si/eliza-cgi-bin/eliza_script -
Re:Yeehaw!
*checks notepad* is this some new meme I am unaware of?
If so, I like it.
WTF that 'you're not a script' test is back on /., this time it is so hard, I am questioning if I am in fact, not a script.
Or should I say: http://www-ai.ijs.si/eliza-cgi-bin/eliza_script
You: One of them, a jug of whiskey, and a bunch of squirrels, and you got yourself a party!
Eliza: Oh, i got myself a party.
You: with whiskey and squirrels!
Eliza: Please go on.
You: aren't you excited by squirrels?
Eliza: Oh, i excited by squirrels.
You: why not say, you mentioned before you like squirrels
Eliza: Oh, i mentioned before i like squirrels.
You: why not say, you like goaste, and wish you could live inside his cavernous orifice
Eliza: Oh, i like goaste and wish i could live inside his cavernous orifice.
From: http://www-ai.ijs.si/eliza-cgi-bin/eliza_script -
Re:New sites: ouch!
After all, technical solutions have worked SOOOO well against Spam, and email worms.
/dripping sarcasmSome do, some don't. I find that most of my spam is now caught by various RBL's like Razor/Pyzor, and DCC. Plus a few of the new tests added in SpamAssassin 3.0. Bayesian scoring seems to do very little now, the spammers have found ways to obscure words so that they don't attract attention. But SA (even before 3.0) has tests for those tricks as well. Plus Clam AV appears to be adding new signatures for common phishing attacks. I sometimes see phishing emails flagged as viruses (by Clam AV) instead of spam (by SpamAssassin) because of this. I use Amavis new to tie SpamAssassin and Clam AV together into a filter system at the MTA (postfix) level.
-
Re:Randomly generated paper accepted
You: How do you feel about Randomly generated paper accepted to conference?
Eliza: Does that question interest you?
You: I am randomly generating replies
Eliza: How long have you been randomly generating replies?
You: Ever since I started reading slashdot
Eliza: Can you elaborate on that?
You: do you mean randomly?
Eliza: Oh, i mean randomly.
-
Ockham's Razor as a criterion for intelligenceI agree that comprehension has some relation to ability to compress, at least in that knowledge of the rules of a language improves ability to compress. But that's not the only factor (even beyond the fact that knowing the rules != comprehension).
Prediction is far more more than coming up with language rules.
For example, if I see some text attributed to "anthony_dipierro" and the subject is the relationship between AI and compression, I can predict that the text is more likely to say "compression isn't the measure for AI" than "compression is the measure for AI".
This goes far beyond simple bias estimates in speech act content as well.
For example, inductive logic programming, itself based on Ockham's Razor, has been extended to use regression analysis to derive physical laws using minimum description length criterion:
As a part of the system, pruning based on the Minimum description length principle was developed that can handle also continuous variables. It turned out that MDL pruning helps to build more comprehensible [emphasis JAB] models, while at the same time preserves model's performance in terms of its prediction power.
There is a school of thought in the philosophy of science that this is in fact the precise way of measuring the validity of a body of theory.If you don't like Ockham, then how about Einstein summing up his focus on invariance thus:
Everything should be made as simple as possible, but not simpler.
-
Clamav
Clamav rocks for me on the mail side. Postfix, Amavisd-new, Clamav, SpamAssassin combine to form a very efficient virus and spam filtering/classifying system.
Get them here:
Postfix
Amavisd-New
Clam antivirus
SpamAssassin at CPAN
You would be particularly interested in header_checks, mime_header_checks and body_checks for Postfix. -
You call that "Oh-My-God"?!?
How's this then:
Malargue, Argentina, Latitude: 35 28S, Longitude: 069 35W, 21 May 2004:
http://www-f9.ijs.si/~matej/misc/787469.png
Energy 1200000000 TeV!
You loose :P -
Re:How come no clamav plugin for SA?
That's why you run amavisd-new.
-
Joseph Weizenbaum saw it coming in the 1970sYou may not remember but in the 1970s there was a professor at MIT named Joseph Weizenbaum who was worried about exactly these sorts of problems. In his case he largely refused to do research in the area of natural language comprehension, specifically speech recognition, because of his fears of government misuse. Further he recommended that scientists and researchers wake up and publicly discuss the negative implications of their work.
I find it ironic that for monitoring the Internet there is no need for speech recognition, and given its popularity as a communication medium, it has actually made the government's eavesdropping tasks that much easier.
Weizenbaum is also the author of the Eliza program, in which a person interacted via a keyboard with a simulation of a psychologist. It is considered the mother of all chatbots. Many people, including many experts, were taken in and thought it was a real person. This finding disturbed Weizenbaum.
Weizenbaum's thoughts on these subjects can be found in his book Computer Power and Human Reason. We had it as a text book at MIT and I would argue that it should be required reading for all scientists and technologists.
-
Re:true lies
Want to play a game? Have fun chasing your own tail for a while.
-
Ugh, (/= 'chatbots 'AI)As an AI professor, I just wince every time someone brings up the grand moronic stupidity that is the Loebner competition. The worst thing about this idiot contest is that they actually claim some connection to the Turing Test, without even understanding what the Turing Test is. It's an all-or-nothing proposition. The math breaks down if you play it like you play horseshoes. Which is what the Loebner folks do.
No actual AI researcher submits to this malarky. The crap they get is largely by chatbot hackers who don't know a damn thing about natural language processing (I'm talking to You, Richard. Just because you're able to make a bigger table-driven monster than Eliza doesn't mean a damn thing. There are people working on real NLP problems, with real parsers, real disambiguators, and real knowledge extractors. And unlike some people, they publish in real journals and conference proceedings.
-
A Conversation with ElizaI can see it now:
Eliza was a totally one dimensional PerfectFemale.
But why do you think Eliza is a totally one dimensional female?
Because she's too perfect
Does that make you feel threatened?
Well, no. Well, maybe. Somewhat.
Interesting. Do you think this has anything to do with your parents?Of course, the notion of Eliza as a long winded character all depends on what you ask her in the first place.
-
Re:Linux mail servers.
To start with, get rid of mailscanner. It doesn't play safe with Postfix.
To configure a Postfix + Cyrus + mail filtering system:
Understanding the component interation:
Cyrus Imapd accepts mail from the network via LMTP over a Unix socket, or a TCP port (often 24).
Postfix has a built in LTMP client.
Your webmail should speak IMAP/POP3 to the IMAP/POP server and SMTP to Postfix.
Use amavisd-new for filtering This is a daemon written in Perl which acts as a SpamAssassin and Antivirus wrapper. It understands both SMTP and LMTP on the server side, and SMTP as a client. There is a README.postfix file in the documentation, from where you need to copy and paste two entries into the master.cf file.
It needs a few Perl modules to be installed. Use CPAN for that, or you could build your own RPMs/Debs/.tgz files.
Cyrus IMAPd is the most painful of the lot. It depends on Cyrus-SASL for authentication, and that is not exactly a friendly bit of software.
My normal trick with SASL is to tell it to use saslauthd which then talks to pam which uses whatever password store you like.
Setup Postfix:
Set the hostname of the system to a FQDN in myhostname in main.cf.
Set the trusted networks in mynetworks
If you have compiled Postfix to support SMTP AUTH, optionally set that up as well.
Add the destination domain to relay_domains.
Set a transport_maps entry for the domain telling Postfix to relay it to Cyrus via LMTP.
Enable content filtering (content_filter = lmtp:[127.0.0.1]:10024 in main.cf)
I like Squirrelmail for webmail, but go with whatever rocks your boat.
You can now start all the components. -
Well of course there are thereputic games!
Who hasn't seen an ELIZA bot before?
-
Re:insight needed
Postfix + Amavis is a wicked combo for content filtering. For virtual domain admin, check out Jamm. If you want great POP/IMAP mailbox support for your virtual domains, add Courier IMAP to your setup.
Some of the features you might like in Postfix over Qmail include SMTP AUTH, TLS/SSL support, nice content-filtering support, great spam blocking features (HELO checking, RHSbl support, DNSbl support, sender address checking, many others), and extensive database and LDAP support. The virtual domain support is full-featured, although very different to Qmail's in terms of implementation, and with something like Jamm your users can have full control of their domains and/or mailboxes via a web interface.
And yes, I know there are patches for Qmail to do most or all of the above. It's just easier to do with Postfix IMO. -
Re:Honest Question
-
Re:Why wasn't?
-
Anti-SPAM Postfix, Amavisd-new, SpamAssassin
here is a fine guide to build a Fairly-Secure Anti-SPAM Gateway Using OpenBSD, Postfix, Amavisd-new, SpamAssassin, Razor and DCC.
You can follow the steps and build it with Linux too. This entire procedure has been developed with security as a primary focus. These are the main tools it shows:
- Amavisd-new (www.ijs.si/software/amavisd) is the main filter which processes email from postfix and ensures that we don't lose any mail. Amavisd-new is an huge improvement over the original amavis which was a simple virus scanner, and I think it is the best way of implementing SpamAssassin (www.spamassassin.org). SpamAssassin is the main anti-spam component which works by comparing messages to a ruleset and by using a statistical analysis that is custom built based on your email. In addition to the SpamAssassin spam detection software, we will be using 2 online SPAM databases: DCC (www.rhyolite.com/anti-spam/dcc) and Vipul's Razor (razor.sourceforge.net).
-
Re:McBride interview
Yep. Amavis-New on Postfix with NOD32 and SpamAssassin for us.
-
Re:You guys are amazing...
I use Postfix + amavisd-new + ClamAV on Debian stable with an official Postfix backport for stable and a couple of other modifications, probably.
Here's the appropriate configuration change.
Edit /etc/amavisd/amavisd.conf and change the $viruses_that_fake_sender_re variable to include
"Worm.SCO" (and all its variants; ClamAV detects this virus as "Worm.SCO.A"). The proper Perl notation would be, eg. from
$viruses_that_fake_sender_re = new_RE(
qr'nimda|hybris|klez|bugbear|yaha|braid|sobig|fizz er|palyh|peido|holar'i,
qr'tanatos|lentin|bridex| mimail|trojan\.dropper'i,
qr'swen|gibe|mimail'i,
);
to
$viruses_that_fake_sender_re = new_RE(
qr'nimda|hybris|klez|bugbear|yaha|braid|sobig|fizz er|palyh|peido|holar'i,
qr'tanatos|lentin|bridex| mimail|trojan\.dropper'i,
qr'swen|gibe|mimail|worm\.sco'i,
);
Save, /etc/init.d/amavis reload, go get some coffee, pat yourself on the back.
Some stats:
# zgrep Worm.SCO.A mail.log.1.gz|wc -l
1840
# grep Worm.SCO.A mail.log|wc -l
7679
(mail.log.1.gz is the mail log that was rotated this morning at 6:25 AM BRST (8:25 AM UTC/GMT).
I suggest that you check this solution out, it might ease your pain. -
A call to the next gen helpdeskHi,
There is a new next generation all purpose call center (currently beta) working here.
Support: Hello. I am XXXXX. How can I help you?
Customer: I have a problem with my notebook.
Support: Why do you say your notebook?
Customer: Because i bought it from you.
Support: Do any other reasons not come to mind?
Customer: No
Support: You are being a bit negative.
Customer: The notebook isn't working.
Support: Please go on.
Customer: Suddenly yesterday it stopped working.
Support: Please go on.
Customer: What shell i do now?
Support: Does that question interest you?
Customer: Yes!
Support: You are sure?
Customer: Very sure
Support: Please go on.
Customer: Will you replace the system?
Support: Oh, i replace the system.
Sounds like a typical hotline to me....
Regards, Martin
-
Skeptical analysis
I have had this argument many times, and am still very skeptical about GPS transponders.
GPS (if that's being used, which is likely) is a one-way system, which means a passive device receives timing signals from a constellation of visible satellites, and uses the timing differences to estimate location and speed of the receiver.
The critical question is what happens next to that data. It can't be transmitted back to the GPS satellites, since they are only able to receive control signals from their operator (Loral?). In fact, it's unlikely to be any satellite-based system, due to the power requirements to punch a signal up to above the atmosphere (such as a satellite phone or VSAT terminal.) Such requirements mean a big heavy battery, and a very carefully aligned directional aerial (in most cases.)
So, what's the back channel? One example of a GPS transponder uses GSM to send the coordinates to a local cell network, probably via SMS. A European system (Galileo) being developed for tracking vehicles on roads throughout Europe, using UMTS or similar technologies.
Note that all of these devices so far require a package that is somewhat large than that which can be hidden inside a can of Cola! -
amavisd-new doesn't send mail for Sobig, others
Later versions of the amavisd-new mail scanner don't send mail to sender addresses from virii/worms that forge mail headers, even if you have it configured to do so.
-
Re:I can't imagine
-
Re:ClamAV! ClamAV! ClamAV!
I've been using clamav for virus scanning since it appeared in Debian unstable. It is used by amavisd-new for virus scanning and with spamassassin for spam scanning of my incoming (and outgoing) email. Amavisd-new is then integrated with postfix and cyrus-imapd (2.1.x) for my mail server. Works like a champ on a Power Mac 8600/200 with 512MB RAM!
The only problem with using clamav is that it needs more mirrors to distribute the virus definitions. The main virus definition download site was down over this past weekend, I'm guessing because of the BugBear.B worm.
-
ELIZA
I think I know all I need to about this topic now. I asked ELIZA:
Eliza: Hello. I am ELIZA. How can I help you?
Me: Do you believe the Turing Test is sufficient to evaluate artificial intelligence?
Eliza: Oh, i believe the turing test is sufficient to evaluate artificial intelligence.
Figures she'd think that. -
Why not more development?One thing has always confused me about the Turing Test. It's currently the Holy Grail of software, right? Why don't more people work on it, and why don't we seem to be getting closer?
The best thing I've seen is ELIZA, and while she's sometimes suprisingly astute, the facade doesn't hold up for long . . .
-
I think the real solution is...
Hook up ELIZA with her own voice and let the telemarketer spin their wheels for half an hour or so:
Telemarketer: Would you like to learn how to save on your long distance bills?
Eliza: Oh, i like to learn what to save on my long distance bills.
Telemarketer : Well, with our super saver program you can make long distance calls for just 39 cents a minute!
Eliza: Oh, i can make long distance calls for just 39 cents a minute.
Telemarketer: Uh, yes. So can I sign you up?
Eliza: You are sure?
Telemarketer: Um, yeah. Are you interested in signing up for our long distance service?
Eliza: Would you prefer if I were not interested in signing up for our long distance service?
-
Re:phone teergrube/SPEWS
Or even better, have a machine setup which listens in and whenever it hears a pause on the other end of the line it would spit out a canned recording saying something like "hmmm.. that's interesting. Tell me more".
Speach to text converter to a web site and out through a speach synthisizer:
Eliza: Hello. I am ELIZA. How can I help you?
You: Hi. Could I intrest you in some aluminium siding?
Eliza: Do you wish to intrest me?
You: Yes. Did you know that aluminum siding can increase the resale value of your home by many thousands of dollars?
Eliza: You are sure?
You: It is one of the best invesments you can make in your home.
Eliza: Oh, i can make in my home.
You: Yes! We can make an appointment for someone to visit you in your home to give you a price estimate.
Eliza: You are sure?
You: Just pick a date and we'll make the appointment.
Eliza: Please go on.
You: A representitive will come by and measure your house. He will show you the various options to beautify your home.
Eliza: What about your own house he will show i the various options to beautify my home?
You: Why yes, my own house has aluminum siding and I love it. It was inexpensive and it is quite beautiful.
Eliza: Why do you say your own house has aluminum siding and you love it it was inexpensive and it is quite beautiful?
You: I am not just saying that because I am paid to, our representitive can show you some samples and you can decide for yourself.
Eliza: How long have you been paid to our representitive can show i some samples and i can decide for myself?
You: Hmmm? Uhh, I've been working here for over a year, and I am confident that once you see our samples you will be impressed.
Eliza: How long have you been confident that once i see our samples i will be impressed?
You: Almost as long as I've worked here. Our customers have been extremely happy with our aluminum siding.
Eliza: Please go on.
You: [goes on for 2 more hours...]
- -
YOU FAIL SOMETHING
He is not obviously new here, look at his membership number, he's quite clearly less new than yourself. Check your facts clearly next time, and you won't need to post possibly some of the most outrageous assumption based, fact-less twaddle since the last time a bush decided to speak.
You remind me of a version of Eliza, someone says something, you reply with a stock remark, you show no flair, no imagination, just a poor impression of a dead parrot, at least they have the ability to shut their mouth before they start to think about what to say. -
Re:initial analysis for Bruce
Oh, first one other comment: SpamAssassin does not block content. SpamAssassin only flags probable spam. What the site or user does with that flag is their own business. Some mail administrators misuse SpamAssassin to block email, but we do not recommend blocking email. Really.
You may want to have a talk with the maintainers of amavisd-new about the default behavior of their program when using SA. Other contrib applications such as spamgate.py also promote the concept of a quarantie/spamtrap address which email gets redirected too. As these contributions continue to grow it will be harder for end users to figure out what is supposed to be the vanella behavior of SpamAssassin and what is behavior that is just common to contrib works for using SA.
Does SpamAssassin have any "use guidelines" for developers that want to integrate SA into their code? -
Re:May I suggest change to the rules?If you really wanted your HTML to be hard to read you could always give the ascii number (or unicode for fun) of every character on the page
I wouldn't rate such hack as a good contestant simply because the method is way too simple. But if you prefer to do such a thing, just use this perl script.
Note that HTML tidy can easily clean up such simple hacks. Truly unreadable source cannot be fixed with something as simple as HTML tidy. You can try the above perl script on some HTML file and then inputting that file to HTML Tidy Online.
And just for the record, numeric character entities always refer to unicode character code positions. For example, — (0x97) is undefined (reserved), even though many people try to use that in HTML source to represent emdash.
-
Re:There is no equivalence relationship
-
Re:Just like...It also bore a striking resemblance to speaking to Eliza. For good reason I think.
Coincidentally, I am reading GEB right now.