NSF-Funded "Dark Web" to Battle Terrorists

BuzzSkyline writes "The National Science Foundation has announced a new University of Arizona project, which they call the Dark Web, intended to monitor all terrorist activity on the Internet. The project relies on 'advanced techniques such as Web spidering, link analysis, content analysis, authorship analysis, sentiment analysis and multimedia analysis [to] find, catalog and analyze extremist activities online.' The coolest part of the project is a tool called Writeprint, which 'automatically extracts thousands of multilingual, structural, and semantic features to determine who is creating "anonymous" content' with an accuracy of 95%, according to the release."

5%

[king-manic]

The coolest part of the project is a tool called Writeprint, which 'automatically extracts thousands of multilingual, structural, and semantic features to determine who is creating "anonymous" content' with an accuracy of 95%, according to the release."

So when they get it wrong, and the police storm my front door instead of my neighbors, will it still be "cool"?

Re:

[ShieldW0lf]

Man, I bet the British would have loved to have such a tool when they were occupying Ireland and Scotland. All those filthy Scottish and Irish terrorists would have been no trouble at all.

Re:

[Oligonicella]

Iraqi's don't care for them. I think that's the best vote.

About 20% of "colonists" opposed our Independence.

[khasim]

Back around 1776 there were a large number (about 20% of our population) of "Loyalists" who opposed our Independence.

If you had polled England at the time, and those Loyalists, you'd understand that the "terrorists" had control of the "colonies".

If England had won, every one of those "terrorists" who had signed their little "Declaration" would have been hanged. And their would have been rejoicing in the streets of the colonies.

Re:About 20% of "colonists" opposed our Independen

[veganboyjosh]

...And the locals would have welcomed the British with open arms...

Re:

[bmajik]

are you aware of any incidents involving colonists in 1776 blowing up markets full of children?

I'm pretty sure nothing of the sort happened, but i'm willing to hear evidence to the contrary.

Those were greatly less "evolved" times, and yet, my impression is that those at the forefront of political dissent were vastly more humane in spreading their message.

Re:About 20% of "colonists" opposed our Independen

[sumdumass]

Back in 1776, the terrorist were organized military fighting organized military. Can you seriously see no difference in fighting a war and blowing up random stranger walking down the streets hoping it is a soldier? Do you seriously think that stocking arms, ammunition, and other supplies or hiding in a church because you know the other side won't go there is comparable to fighting in the open?

I would agree if the insurgents would act in a military manor. but as of yet, they are completely happy with killing

"terrorist" vs. "freedom fighter"

[mi]

One man's "terrorist" is another man's "Freedom fighter".

The oft-repeated phrase is rather stupid. "Terrorist" is about methods (means) — targeting civilians. "Freedom fighter" is about goals (aims) — achieving a "freedom" of some sort or another.

There is not contradiction — one can be both in the same person's opinion. See also False Dichotomy [wikipedia.org]...

Re:

[Televiper2000]

If only those terms could be so eloquently applied in real life. But, in the real world they are often politicized. In fact the text you quoted uses quote marks around the words terrorist and freedom fighter to imply that it's merely what's said. Dictators will always run propaganda campaigns against rebels labeling them terrorists. Freedom fighters will always be left to achieving their goals through unconventional tactics that will more often than not be labeled "terrorist" tactics. Insurgent armies

Re:

[ShieldW0lf]

Terrorism is about creating fear in a population by attacking targets that have no military significance. When the IRA blows up a grocery store, that's terrorism.

The US wasn't attacked by terrorists. They were attacked by a tight knit military group that went after their critical infrastructure. The world trade center, the center of their economy. The pentagon, center of their military. And the commander in chief.

There have been no grocery stores blown up, no shopping malls, no attacks with Nuclear, C

Re:5%

[Jarjarthejedi]

I'm more curious how they're going to get 95% accuracy on who the person is without a large number of samples of non-anonymous writings from them. It seems obvious that they're really claiming that, with a large number of writing samples from the writer, they can get 95% accuracy. If they're actually claiming to be able to determine who anonymous people are without any non-anonymous writing by them then that's a system I have to see...

Re:5%

[Anonymous Coward]

More likely it'll be along the lines of "These anon posts seem to be from the same person, and we should make more attempt to trace several of them to their source, rather than wasting our efforts on those over there..."

And simple to defeat?

[khasim]

Instead of posting anything anonymously yourself, just tell someone else to post it. There speling errors will not be the smae as your's and their sentence structure will be different.

Okay, they'll be able to group all of his posting as being posted by him ... but they won't be able to tie it to him unless he also posts a lot of stuff non-anonymously.

Re:

[badboy_tw2002]

Or they have YASWTP - Yet Anothe Secret Wiretap Program snatch one of the posts. And they're really only limited by what they can do in the States (or what they give lip service to as "not being able to do") - in other countries the gloves are pretty much off and only limited by how much the other country can figure out.

Don't think for a second that they aren't trying to actively hack some of the more popular places these things are being posted. If they can get one honey pot and the correlate that guys p

It's not even that difficult.

[khasim]

Every TCP/IP packet has a source address and a destination address.

So all that the government would need would be the addresses of the web sites (no matter where they are located) and taps on the pipelines. You can either try to catch the stuff going OUT of your country or going INTO their country (if you can't just tap the line of that website).

That will tell you who, in your country, is going there.

As long as it isn't using encryption, you'll even get what is being read/posted.

If it is using encryption, you still should have the location of the guy reading/posting. Or you can try cracking the encryption.

Once you have the location of the guy, you get a warrant and put a keylogger on his box or whatever.

There's no need for all of this crap about "darkweb". Google can already tell you what is posted on what websites. If these guys are smart enough to beat the basics, they're smart enough to know NOT to use the Internet for point-to-point communications.

Re:

[merreborn]

That really only works if you can count on your targets connecting directly to servers, and not using proxies/TOR.

Also, if the source IP in question happens to be, say, a NAT address that serves 100 terminals in a public library, then you don't have much to go on.

Re:

[Atlantis-Rising]

But you have far more to go on than you did before (~100 PCs compared to ~100,000,000)- and that is where conventional policing steps in, because that's what it's designed for.

Re:

[morgan_greywolf]

Every TCP/IP packet has a source address and a destination address.

But that doesn't necessarily mean that every TCP/IP packet has an accurate TCP/IP packet. See spoofing attack [wikipedia.org] for more info.

Re:

[g-san]

Or somebody copuld make intntional errors in somebeody's usual spelling habits and grammer.

Re:

[triffid_98]

Wow, in oda' news, terro'ists are now translatin' all uh deir anonymous postin's into JIBE, snatch dat mr. Ah be baaad... man. 'S coo', bro.

Re:

[Foobar of Borg]

1n 0+h3r n00s, 73rr0r1$7$ @r3 c0mmu.n1c@71ng s0131y 1n 1337

Re:

[1u3hr]

Okay, they'll be able to group all of his posting as being posted by him ... but they won't be able to tie it to him unless he also posts a lot of stuff non-anonymously.

I for instance, post here under a randomly chosen pseudonym. But on another forum, based in my neighbourhood and discussing local issues, I post under my own name and link to a personal web page. You can find my name, address and phone number there. If this system works as advertised (big if) it could correlate them. Though my style is dif

Re:5%

[alexhs]

Of course, when you register to DarkWeb, you give your identity. Obviously, 5% of registered people didn't enter their real identity.
Now, the biggest problem is to get terrorists to register to and use that DarkWeb thingy. But with such a kewl name and a good advertising campaign, it shouldn't be too hard.

Re:

[Smidge204]

Perhaps it's not so much identifying the real identity of the individual, but rather the ability to identify a particular anonymous writer apart from a whole group of anonymous writers.

In other words: they may not know the real names, but they can identify all the anonymous posts made by the same person with 95% accuracy. That seems much more doable compared to divining a person's real identity from nothing more than a pile of anonymous data.
=Smidge=

Re:5%

[colmore]

The worst thing is that for a search like this, 95% accuracy is TERRIBLE.

Let's say in 1,000,000 posters there are 20 secret terrorists. This system (assuming the 95% figure isn't just made up, and since it's a reliability figure coming from a government contractor - it is) will label 19 of the real terrorists as terrorists and *50,000* innocent internet users as terrorists. Since we already live in a world where being under government suspicion (but no charges) gets your assets frozen, phones tapped, and puts you on the no-fly list this is a BIG problem.

I go to a fairly international university. I've seen this 1984 B.S. shit on innocent people's jobs and educations first hand. As long as our elected representatives keep granting themselves and their officers these kinds of powers, we do not have the right to call ourselves the "land of the free."

Right now the US has in place a set of laws that would allow for an authoritarian (not-quite totalitarian, though if the press keeps dismantling itself, who knows) government. All it would take is the decision to enforce them to the letter; no consent from the voters would be needed.

Re:5% +++++

[davidsyes]

SOME of the other 5% will come from (or, alternatively, maybe the FIRST 95% comes from) use of Visual Analytics:

http://www.visualanalytics.com/ [visualanalytics.com]

Hell, just see:

http://www.google.com/search?q=visual+analytics&ie=utf-8&oe=utf-8&aq=t&rls=org.mozilla:en-US:official&client=firefox-a [google.com]

The thing is, I wonder if that NY Times (I think it was NYT) reporter/columnist under bushwhack/assault for "divulging" sensitive collection techniques to the "ter'rists" knew of Visual Analytics and could have shiel

Re:

[colmore]

You missed my point.

If you're searching for positives in a huge pool of negatives, even an extremely reliable test will return a majority of false positives.

In the scenario I described, out of a million internet users, the system turns up 50,000 only 19 of whom are terrorists, but all of whom are now under suspicion.

And don't paint suspicion about intelligence techniques in such black and white terms. I don't believe that the government whisks people away to gitmo because of a single computer test, however

Here's How

[asphaltjesus]

The likely scenario is they have a "person of interest." One way or another, they have some correspondence from this person. (think search warrant, warrant for arrest, etc) Then they go backwards through the application and associate anonymous content to the person of interest. I don't see a logical way to go at it from the another direction.

What's so awful about the whole idea is the unlikely event the evidence, or the method used to collect it is ever scrutinized.

For those of you still convinced hunti

Re:

[Walt Dismal]

CIA agent: "Hey Ted, come look at this. Ernest Hemingway isn't dead after all! He's writing for Al Qaeda."

Agent #2: "Wow! That bastard! Let's burn all his books!"

CIA agent: "And this other guy, he seems to be totally illiterate."

Agent #2: "Bob, you maroon, that's a George Bush speech they're quoting!"

CIA agent: "Ooop."

Re:

[StikyPad]

Clearly you're forgetting that computers never make mistakes!!

Re:

[zCyl]

My impression was that the 95% probability was that two postings were by the same author, not that they knew the identity of the author.

Which is impressive from an academic standpoint, but perhaps not very useful for mass surveillance. If you have 10,000 identified posts, and 10,000 unidentified posts, and you want to match them up with a one-to-one mapping, suddenly that 5% sounds much more troublesome. If each unidentified post scores a positive on 500 identified posts, and each identified post scores a

Re:

[hedwards]

Depending upon how reliable that estimate is, that would make this roughly as effective as DNA evidence in some cases.

DNA matches aren't typically 100%, the condition of the specimen isn't guaranteed to be of quality, and at this point, even with a 100% match it won't do you any good if you don't have a suspect. The accuracy isn't 100% because typically they won't check the entirety of the strands, they'll look at a relatively small section. It is pretty reliable, but as I already said, it isn't much good i

Re:

[ArcherB]

So when they get it wrong, and the police storm my front door instead of my neighbors, will it still be "cool"?

I would hope that if your neighbors are terrorists, you would have already called them in. I wouldn't want a bomb maker living next door to me!

Re:5%

[Anonymous Coward]

I call the FBI about all of my neighbors, just in case. I recommend you do the same.

It's better to be safe than sorry; why, just the other day, I saw some guy walking suspiciously down the street. I'm not one to overreact, but this guy was just suspicious if you know what I mean. He looked like he came from the Middle East, had shifty eyes, the full shebang.

So I'm walking along and I see this guy. I almost kept going, minding my own business, but I thought about the danger this proud nation is in and I thought to myself, "If I don't do something, who will?"

And thank god I did.

I called 911 (blessed may that number always be in our hearts) and reported the likely perpetrator. I tailed him from a distance for a while, and my if he wasn't surprised when that officer pulled over next to him! You should have seen the look in his eyes, caught in the act!

So, long story short, turns out the police couldn't arrest him for anything (or he got off on some technicality, probably). I know one thing: he'll be more careful next time he decides to pull something. You've got me to thank for that.

Re:5%

[autocracy]

Oh, awesome... thanks for making sure he'll be more careful at his nefarious deeds. You've done us all proud there, Scooter.

Re:5%

[Reziac]

Place and time: somewhere in the Soviet Union in the 1930s. The phone rings at KGB headquarters.

          "Hello?"

          "My neighbor Ivan Asimov is an enemy of the State. He is hiding undeclared diamonds in his woodshed."

          "This will be noted."

          The next day, the KGB goons go over to Asimov's house. They search the shed where the firewood is
          kept, break every piece of wood, find no diamonds, swear at Asimov, and leave.

          The phone rings at Asimov's house.

          "Hello, Ivan! Did the KGB come?"

          "Yes."

          "Did they chop your firewood?"

          "Yes, they did."

          "Okay, now it's your turn to call. I need my vegetable patch plowed."

Re:5%

[mcpkaaos]

I'm sure your neighbor will think so.

Re:

[ajs]

The coolest part of the project is a tool called Writeprint, which 'automatically extracts thousands of multilingual, structural, and semantic features to determine who is creating "anonymous" content' with an accuracy of 95%, according to the release."

So when they get it wrong, and the police storm my front door instead of my neighbors, will it still be "cool"?

5% error rate is too high to base any first-order data on. My assumption would be that they'll use this information to determine what online content to spend their time working on. For example, if the modern equivalent of Echelon tells us that a terrorist in Iraq makes frequent calls to someone who makes frequent, high-signal calls to someone in the U.S. and that person is identified as the potential author of several anonymous postings to various forums, then you spend a whole lot of time analyzing those

Re:

[OrangeTide]

Basically the news is that they can cast a wider net. As far as we know the government's capabilities for monitoring high profile targets have not changed, it just scales much better now.

Re:

[non]

this would appear to be based on latent semantic analysis. see the wikipedia article [wikipedia.org] for some of the math. the group behind much of the work in this field are at U of Colorado. they have a site here [colorado.edu].

Re:

[Applekid]

And the best part will be that, during your trial (if you even get one), when you try to defend against an algorithm basically being a witness testifying against you, you will not be allowed to know it since that would involve divulging "national security secrets."

I wonder how they'd implement the witness protection program for code?

Re:

[dattaway]

95% is only 1 out of 20 times. Your door will be safe for the first 19 pages...

Re:

[Prof.Phreak]

So when they get it wrong, and the police storm my front door instead of my neighbors, will it still be "cool"?

Your lucky neighbor might think so!

Re:

[Oligonicella]

(remember when the branches of government were truely independent and this included the judiciary?)

No and neither do you as that has never been the case. Checks and balances precludes true independence.

This could have been used...

[halivar]

...to out Dan Lyons as "Fake Steve."

Other than that, I'm afraid this is the sort of technology that's only "cool" when it isn't being used on you.

Not to be confused with Darknet

[JamJam]

Not to be confused with Darknet http://en.wikipedia.org/wiki/Darknet [wikipedia.org] which is what I immediately thought from this article title.

Security is more important now than ever!

[ArcherB]

And yet another reason why you should lock down your wifi!

Attention NSF

[Anonymous Coward]

Attention, NSF: Here's a better, cheaper solution - point all those @#$@#$%ing existing VIAGRA and mortgage spambots out there at these forums you're monitoring.

Either the terra'rists give up after the spamming, or they kill the spammers. Either way, we win.

Re:

[Captain Splendid]

Somebody get this man some Venture capital STAT!

The quote you're looking for

[akad0nric0]

...is:

Quis custodiet, ipsos custodes
- Juvenal

Re:

[cromar]

Sorry, "this one" qualifies as a pronoun, which is a noun...

F or A?

[Slightly Askew]

Change NSF to NSA, and the summary would make just as much sense...except "terrorist" would be defined as whatever the current politicians in power decide it to mean.

Space race, nuclear power, this kind of technology. Just goes to show, if you have a good idea, find a way to use it to further the war machine and political agendas and prepare to get buried in money. Can someone please figure out a way to weaponize a cure for cancer?

Re:F or A?

[ScentCone]

Can someone please figure out a way to weaponize a cure for cancer?

You mean kind of like how there are now lots more skilled laser eye surgeons in the private sector competing to give you better prices for your business because once the military decided to back providing that service to its pilots, there was a giant leap in people being trained to do the work during their rotations?

As far as cancer: the military provides all kinds of basic medical research from which we all benefit. You'll see considerable military spending in epidemialogical studies, trauma treatment, etc. To the extent that, say, The Marine Corp is a weapon, the huge studies that can be conducted on the systematically collected health stats, DNA, etc., on a huge number of generally healthy people over several generations IS a part of all sorts of cancer (and other) studies.

Re:

[DocSavage64109]

You'll have to be a little more subtle if you really want anyone to fall for your strawman.

Re:

[ScentCone]

I'm guessing you're a fucking jarhead.

Actually, no. But then, you probably can't imagine that someone might actually respond to another post's comment by mentioning actual information that's relevent to the conversation, right? Or is someone less of a 'fag' if they cite the Coast Guard, instead? Which branch of the military does one get to mention that makes them less homosexual, from your perspective, exactly?

paint a bulls-eye on your forehead fag

And I don't have to guess. You're making it obvious

Re:

[ShieldW0lf]

Space race, nuclear power, this kind of technology. Just goes to show, if you have a good idea, find a way to use it to further the war machine and political agendas and prepare to get buried in money. Can someone please figure out a way to weaponize a cure for cancer?

1) Find a cure for cancer
2) Indiscriminantly irradiate the globe, giving everyone cancer
3) Distribute the cure only to card carrying citizens

There you go. Where do I get my money?

Another good tactic is to create diseases which, based

There are no false positives.

[Erris]

except "terrorist" would be defined as whatever the current politicians in power decide it to mean.

The current terminology is "potential terrorist", a category that includes everyone. The implementers mean it too.

Welcome to the new Constantinople, a power without peers, knowledge without truth, laws without justice and wealth without dignity. With tools like this the current power controls the future and the past. Dissidence will be impossible and change overs will only occur by coup. Without priva

Who's the extremist on-line?

[enrevanche]

This is something the National Science Foundation and University should be ashamed of. This will used to spy on Americans (and others) and will have little to do with terrorism. I'm sure it will be salable to many corporations as well.

These jerks are the "extremists on line".

Re:

[moore.dustin]

Ok I get the first part. Spying on our own citizen is bad, agreed, signed. Now can you explain why it is bad that it gets sold to corporations?

Also, they should not be ashamed of creating the technology, but ashamed of how it is used if it is wrong. That is like saying inventing the plane was bad because it would be used to fight wars. Bad example perhaps, but you get the idea.

"Spying"

[mosb1000]

Since all this information is readily available to anyone one with internet access, I don't think it's reasonable to call it spying. Seriously, if you post information on a message board where anyone in the whole entire fucking wold can read it, maybe you should expect that government officials and corporations can look at it a well!!!

Re:

[Vellmont]

I don't think it's reasonable to call it spying.

You're right, it's not spying, it's surveillance.

That doesn't really make it any better, however.

remember...

[weopenlatest]

...that the Bush administration's definition of 'terrorist' includes Democrats, pot smokers, vegetarians, and people with two arms and two legs.

Re:

[jollyreaper]

...that the Bush administration's definition of 'terrorist' includes Democrats, pot smokers, vegetarians, and people with two arms and two legs.

Then why was Vietnam veteran and triple amputee Max Cleland branded a traitor?

Re:

[TubeSteak]

Then why was Vietnam veteran and triple amputee Max Cleland branded a traitor?

Well, if he was a Vietnam vet, then he probably smoked pot...
And everybody knows that smoking pot = supporting terrorism.
http://www.google.com/search?q=marijuana+terrorists [google.com]

Re:

[Don853]

...that the Bush administration's definition of 'terrorist' includes Democrats, pot smokers, vegetarians, and people with two arms and two legs.

Then why was Vietnam veteran and triple amputee Max Cleland branded a traitor?

"NSF-Funded"

[Basilius]

For those of us (like myself) that work closely with the banking industry, the phrase "NSF-Funded" produces quite a bit of cognitive dissonance.

OMG! I am dead

[140Mandak262Jamuna]

I should have known better than to cut and paste whole postings from the jihadi discussion fora to rebut them point by point. Now if that software can't tell from semantic structure, what I said and what I quoted, I can expect some visitors, look like. May be I will post in Slashdot and display some esoteric knowledge like the plural for forum is fora and may be that will throw a monkey's wrench into their Beysian filters. Ha ha ha.

Re:

[Randym]

...their Beysian filters. Ha ha ha.

I didn't know Hakim Bey was a mathematician too. Geez, that guy can do *everything*!

Anonymous Cowards?

[Fex303]

The coolest part of the project is a tool called Writeprint, which 'automatically extracts thousands of multilingual, structural, and semantic features to determine who is creating "anonymous" content' with an accuracy of 95%, according to the release.

A way to track down the ACs who keep posting homoerotic rants and random trolls.

This tech could destroy Slashdot as we know it!

RTFA People

[db32]

By analyzing these certain features, it can determine with more than 95 percent accuracy if the author has produced other content in the past. How fucking hard is that to read? Seriously? Every comment right now is on some bullshit tangent about hunting people down or other such nonsense, or how its impossible to figure out who it is without blah blah fucking blah. What it DOES say is that they can take a large ammount of anonymous information and tie it together to a single player. Not that it gives t

Re:

[bitRAKE]

Yeah, and I'm a hot 17 year old busty woman looking for a good time.

Re:

[mccalli]

So they are still an anonymous player, they just have their anonymous works attributed to them as an anonymous individual

Not necessarily, no. Take my history here for example - there are comments I've made as an A.C.. If this can work out my structure, then it should be able to tie back with 95% accuracy that those A.C. comments are from the individual behind the mccalli account.

That's just here. I've always posted under my real name in most places (been using the net a lot longer than the spam probl

Re:

[db32]

That is not what the article says. The article says "By analyzing these certain features, it can determine with more than 95 percent accuracy if the author has produced other content in the past." At no point does it actually say that it can identify a real individual, or that it can even identify each contributions source. Literally what it says is given enough information it can make a 95% accurate guess that an individual has contributed more than once. This does not mean it can identify all of that

My Name

[AbbyNormal]

is Abdul. I come from kingdom far away and have inheriteed moneys. I give you $1,000,000 if you please give me bank account information...etc.

SPAM should be fun to sort out and/or any spammers on blog comment sections.

Aliens moment

[Lurker2288]

"The project relies on 'advanced techniques such as Web spidering, link analysis, content analysis, authorship analysis, sentiment analysis and multimedia analysis [to] find, catalog and analyze extremist activities online."

Reminds me of something..."I'm ready, man, check it out. I am the ultimate badass! State of the badass art! You do NOT wanna fuck with me. Check it out!...Independently targeting particle beam phalanx, VWAP! Fry half a city with this puppy! We got tactical smart missiles, phase-plasma pulse rifles, RPGs! We got sonic, electronic ball-breakers! We got nukes, we got knives, sharp sticks..."

Computationally expensive beyond practicality

[sdaemon]

Sure you can crawl any information source and extrapolate anything you want out of it. I'd even be willing to believe the 95% accurate analysis, whatever. That's besides the point.

You can only extrapolate data you've read properly. The simplest of encryption and/or obfuscation schemes applied to this content would effectively protect against extrapolation. Sure, Big Brother can have software scrub the Net looking for suspicious content. But can they have software scrub the Net while applying decryption measures to everything found? While analyzing every image file for obfuscated content (or even something as simple as writing your terrorist plans on a piece of paper and scanning it in as an image)? While applying rot13 to every block of text found?

I would say no. The problem becomes computationally impossible at that point. There are theoretically infinite ways to hide, encrypt, or obfuscate data. To have a system check first for unhidden, unencrypted, un-obfuscated data, then also for each of those, is simply not doable unless one makes radical limitations to the format of the data itself.

I would say instead that this "Dark Web" will be invaluable in identifying characteristics of perfectly law-abiding forum posters, slashdotters, and so forth, and that the data gleaned will fetch a good price from directed marketeers, pharmaceutical companies, spammers, government bureaucracies, and other servants of the Dark Lord.

Re:

[phliar]

I would say instead that this "Dark Web" will be invaluable in identifying characteristics of perfectly law-abiding forum posters, slashdotters, and so forth, and that the data gleaned will fetch a good price from directed marketeers, pharmaceutical companies, spammers, government bureaucracies, and other servants of the Dark Lord.

Makes perfect sense -- after all, this project is in the "Eller College of Business" at the University of Arizona.

Re:

[surprise_audit]

So, how much of the web do the search engines crawl these days?? I think it was Altavista, back in the day, that said they thought they were only getting about 10% of web sites. Wouldn't this DarkWeb thing need to be bigger than Google to get/process enough pages to be useful??

Like we didn't know

[turing_m]

"The coolest part of the project is a tool called Writeprint, which 'automatically extracts thousands of multilingual, structural, and semantic features to determine who is creating "anonymous" content' with an accuracy of 95%, according to the release."

On the face of it, this is not that different from Amazon's Statistically Improbably Phrases, which have been around for at least several years now. Every brain is unique, and it's not surprising that each brain creates writings with some sort of statistical

Web Pages?

[stoicio]

I doubt that any self respecting terrorist is going to
expend resources making a web page that spiders can crawl.

Here's a hint:

Terrorist #1 sets up a WIFI home network with
limited external access and **no connection** to the
internet.

Non of the terrorists really want to know each other
since that would make them easier to find if one got caught.

All the other terrorists require is a GPS location relatively
close to the hot-spot. Not even the street address.

They park, or slow down,the car at the GPS coordinates,

The super awesome do-it-all tool been waiting for

[icepick72]

This Dark Web description sounds good, it even uses "semantic" technology but stop and think how little progress Google has made into the semantic web compared to what they want to do, contrasted with the talent they have hired. Considder the description of this NSF tool again. I predict there will be another /. posting in just over a year talking about how the project didn't quite work out as expected.

Heck, their job should be simple!

[Fantastic Lad]

They don't need expensive Dark Web nonsense.

They just need to pull up their own employee roster to see who's largely responsible for world terrorism.

Of course, the young recruits are probably still too busy puffing their chests smartly while humming the "Alias" theme music while quietly wishing that the NSA was the one which received the big Hollywood PR/propaganda effort to notice such sticky details as who was responsible for what. But what are a few sticky details? M's and W's all look the same.

-FL

The big question ..

[PFAK]

Does it adhear to web standards and respect /robots.txt ?

Uh-Huh

[irving47]

Right. Meantime, if I try to translate a letter written to me by a Russian business contact, whether I'm using babelfish or whatever hip new translator app, it makes us sound like retarded 8 year-olds to each other.

You don't need to

[exp(pi*sqrt(163))]

> Writeprint, which 'automatically...'...determine[s] who is creating "anonymous" content' with an accuracy of 95%

Why bother? Why not just make a press release claiming that you can do this when you can't. Then you can (1) get government funding and (2) you might even serve a useful purpose by disrupting communications between terrorists dumb enough to believe the (false) claim.

Idiotic replies about "identifying anonymous"

[OeLeWaPpErKe]

Obviously what this tool will do is link different anonymous content by the same creator.

So if the same guy posts "kill all infidels" and includes instructions on jihadforpussies.com and writes about pink ladies underwear on getithere.com and reviews "movies" at grannysvideos.com they can go track his credit records for pink underwear instead of trying to figure out what internet cafe he posted his "big boy" stuff from.

This does not mean identifying every peace of content by name, just relating stuff by aut

Yay for our New "Dark Web" Overlords!

[dkf]

The coolest part of the project is a tool called Writeprint, which [...] 'determine[s] who is creating "anonymous" content'

They've managed to pin a lot of it on this guy called "Anonymous" Coward, who they think is related the playwright Noël Coward [wikipedia.org]...

Extremist =! Terrorists

[nurb432]

Another 'its for the children' type of maneuver.

This should scare anyone that likes their right to free speech. And yes, even terrorists should have the right to *speak*. If you restrict their right to speak, its not much of a stretch to restrict yours too.

Be afraid.

"Coolest"

[DragonWriter]

The coolest part of the project is a tool called Writeprint, which 'automatically extracts thousands of multilingual, structural, and semantic features to determine who is creating "anonymous" content' with an accuracy of 95%, according to the release.

Yeah, cool, 1 in 20 of the people identified as "terrorists" by this program will not even have created the content which led to their identification as terrorists.

Of course, the linking "creating particular content" to actually being a terrorist is a much har

A University?

[GodfatherofSoul]

I'm very disturbed that a university would think of this as something to brag about. So much for colleges as a haven for liberals...

Irreconcilable Differences

[xrayspx]

Our ideas of "cool" differ in extremely fundamental ways. Good to know. Usually someone might say something is "cool" and I might think "wow that's complete crap", but it's actually as if I have found my perfect opposite here. Congratulations.

Re:

[jbeaupre]

Not enough writing samples yet. It says:

25% Kerry
18% Gore
7% Osama
5% Hillary
45% 3rd grader Stevie Able of 1209 Mayburn St, Dallas, Tx

If you could post a few more messages please.

Re:

[Archangel Michael]

Am I the only person to google map this address? There is no 1209 Mayburn St in Dallas.

However, I did find the following address:

1209 N Mayburn St
Dearborn, MI 48128

Okay, its official. I'm a dork.

Re:

[Daimanta]

And I am smart. I was also interested but I am lazy. I knew that there would be a /. poster who would look it up. Thanks a bunch. Have a cookie.

Re:

[db32]

Not if you read the article. The summary lied to you. You are still anonymous, they just know what other anonymous things you did. By analyzing these certain features, it can determine with more than 95 percent accuracy if the author has produced other content in the past.

Um..so what is it they're REALLY trying to do?

[Quadraginta]

You really figure the NSF is going to blow a few million on trying to figure out who said "The President has his head up his ass!" on /. as an AC?

What for? Who cares? Sounds kind of weirdly paranoid. Or like someone has delusions of being way more important and influential than he is.

I mean, who seriously thinks that anything posted anonymously on random Internet message boards can possibly have any important political effect? As if J. Random Teenager posting from the school computer is going to suddenl

Re:

[PopeRatzo]

You really figure the NSF is going to blow a few million on trying to figure out who said "The President has his head up his ass!" on /. as an AC?

Considering the Bush Administration's security apparatus set up surveillance on a group of Quaker senior citizens just because they were against the War in Iraq, I wouldn't be a bit surprised.

Didn't you read the recent article about how the Bush Administration ran "roving bands" of anti-dissent thugs who would attend public appearances by the President, and if t

Re:

[grungeman]

Even if they only catch the ignorant, you wouldn't believe how many ignorant people are out there who support the terrorists directly or indirectly. Catching the ignorant means derstoying the base of the terrorist elite (if such thing exists), so this is neither a waste of time nor a wast of money.

Re:

[Potor]

there's no redundancy in that sentence ...

Re:

[DragonWriter]

The most common use, IME, when referring to media content, is the "unlike to offend, irritate, provoke strong emotion, etc." one that relates to appearance. The most common definition overall (and the one that is probably #1 in most dictionaries, consequently) is "harmless". Context matters to communication.